alphard/Bjornulf_custom_nodes
1
0
Fork 0
mirror of https://github.com/justUmen/Bjornulf_custom_nodes.git synced 2026-03-21 12:42:11 -03:00
Code Issues Packages Projects Releases Wiki Activity

0.77

Browse Source
This commit is contained in:
justumen
2025-03-19 17:36:25 +01:00
parent 44d69e8907
commit 39dfb0220a
76 changed files with 3207 additions and 955 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
.gitignore vendored
View File

@@ -10,4 +10,8 @@ ScriptsPerso/
civitai/NSFW_*
pickme.py
web/js/pickme.js
todo.py
todo.py
video_text_generator.py
run_workflow_from_api.py
remote_nodes.py
fix_face.py

581
API_civitai.py
View File

@@ -1,4 +1,6 @@
import os
from server import PromptServer
import aiohttp.web as web
import time
import requests
from PIL import Image, ImageSequence, ImageOps
@@ -168,11 +170,27 @@ def download_file(url, destination_path, model_name, api_token=None):
response.raise_for_status()
file_size = int(response.headers.get('content-length', 0))
# Initialize progress tracking if file size is known
if file_size > 0:
downloaded = 0
bar_width = 20 # Fixed width for the progress bar
with open(file_path, 'wb') as f:
for chunk in response.iter_content(chunk_size=8192):
if chunk:
f.write(chunk)
# Add progress reporting here if needed
# Progress bar logic
if file_size > 0:
downloaded += len(chunk)
progress = min(100, int((downloaded / file_size) * 100))
num_hashes = int(progress / (100 / bar_width))
bar = "[" + "#" * num_hashes + " " * (bar_width - num_hashes) + "]"
percentage = f"{progress:3d}%"
print(f"\r{bar} {percentage}", end="", flush=True)
# Add a newline after download completes to avoid overwriting
if file_size > 0:
print() # Moves to the next line after completion
return str(file_path)
except Exception as e:
@@ -222,95 +240,165 @@ import civitai
# ======================
# GENERATE WITH CIVITAI
# ======================
class APIGenerateCivitAI:
@classmethod
def INPUT_TYPES(cls):
"""Define the input types for the node."""
return {
"required": {
"api_token": ("STRING", {
"multiline": False,
"default": "",
"placeholder": "Enter your CivitAI API token here"
}),
"model_urn": ("STRING", {
"multiline": False,
"default": "urn:air:sdxl:checkpoint:civitai:133005@782002"
}),
"prompt": ("STRING", {
"multiline": True,
"default": "RAW photo, face portrait photo of 26 y.o woman"
}),
"api_token": ("STRING", {"default": "", "placeholder": "CivitAI API token"}),
"prompt": ("STRING", {"multiline": True, "default": "RAW photo, face portrait photo of 26 y.o woman"}),
"negative_prompt": ("STRING", {
"multiline": True,
"default": "(deformed iris, deformed pupils, semi-realistic, cgi, 3d, render, sketch, cartoon, drawing, anime)"
}),
"width": ("INT", {
"default": 1024,
"min": 128,
"max": 1024,
"step": 64
}),
"height": ("INT", {
"default": 768,
"min": 128,
"max": 1024,
"step": 64
}),
"steps": ("INT", {
"default": 20,
"min": 1,
"max": 50,
"step": 1
}),
"cfg_scale": ("FLOAT", {
"default": 7.0,
"min": 1.0,
"max": 30.0,
"step": 0.1
}),
"seed": ("INT", {
"default": -1,
"min": -1,
"max": 0x7FFFFFFFFFFFFFFF
}),
"number_of_images": ("INT", {
"default": 1,
"min": 1,
"max": 10,
"step": 1
}),
"timeout": ("INT", {
"default": 300,
"min": 60,
"max": 1800,
"step": 60,
"display": "Timeout (seconds)"
"default": "low quality, blurry, pixelated, distorted, artifacts"
}),
"width": ("INT", {"default": 1024, "min": 128, "max": 1024, "step": 64}),
"height": ("INT", {"default": 768, "min": 128, "max": 1024, "step": 64}),
"steps": ("INT", {"default": 20, "min": 1, "max": 50, "step": 1}),
"cfg_scale": ("FLOAT", {"default": 7.0, "min": 1.0, "max": 30.0, "step": 0.1}),
"seed": ("INT", {"default": -1, "min": -1, "max": 0x7FFFFFFFFFFFFFFF}),
"number_of_images": ("INT", {"default": 1, "min": 1, "max": 10, "step": 1}),
"timeout": ("INT", {"default": 300, "min": 60, "max": 1800, "step": 60}),
},
"optional":{
"add_LORA": ("add_LORA", {"forceInput": True}),
"optional": {
"model_urn": ("STRING", {"default": "urn:air:sdxl:checkpoint:civitai:101055@128078"}), #SDXL default
"add_LORA": ("STRING", {"multiline": True, "default": ""}),
"DO_NOT_WAIT": ("BOOLEAN", {"default": False, "label_on": "Save Links Only", "label_off": "Generate Now"}),
"links_file": ("STRING", {"default": "", "multiline": False}),
"LIST_from_style_selector": ("STRING", {
"default": "",
"multiline": True,
"placeholder": "e.g., Low Poly ;Samaritan 3D Cartoon;urn:air:sdxl:checkpoint:civitai:81270@144566;https://civitai.green/models/81270?modelVersionId=144566"
}),
}
}
RETURN_TYPES = ("IMAGE", "STRING",)
RETURN_NAMES = ("image", "generation_info",)
RETURN_TYPES = ("IMAGE", "STRING")
RETURN_NAMES = ("images", "generation_info")
FUNCTION = "generate"
CATEGORY = "Civitai"
def __init__(self):
self.output_dir = "output/API/CivitAI"
self.metadata_dir = "output/API/CivitAI/metadata"
os.makedirs(self.output_dir, exist_ok=True)
os.makedirs(self.metadata_dir, exist_ok=True)
self.links_dir = "Bjornulf/civitai_links"
os.makedirs(self.links_dir, exist_ok=True)
self._interrupt_event = threading.Event()
def get_next_number(self):
"""Get the next available number for file naming"""
files = [f for f in os.listdir(self.output_dir) if f.endswith('.png')]
if not files:
return 1
numbers = [int(f.split('.')[0]) for f in files]
return max(numbers) + 1
def generate(self, api_token, prompt, negative_prompt, width, height, steps, cfg_scale, seed, number_of_images, timeout, model_urn="", add_LORA="", DO_NOT_WAIT=False, links_file="", LIST_from_style_selector=""):
"""Generate images or save links based on DO_NOT_WAIT."""
if not api_token:
raise ValueError("API token is required")
os.environ["CIVITAI_API_TOKEN"] = api_token
civitai = get_civitai()
empty_image = torch.zeros((1, 512, 512, 3))
# Extract model_urn from LIST_from_style_selector if model_urn is empty and LIST_from_style_selector is provided
if not model_urn and LIST_from_style_selector:
parts = LIST_from_style_selector.split(';')
if len(parts) >= 3:
model_urn = parts[2].strip()
else:
raise ValueError("Invalid LIST_from_style_selector format: cannot extract model_urn")
if not model_urn:
raise ValueError("model_urn is required")
seed = random.randint(0, 0x7FFFFFFFFFFFFFFF) if seed == -1 else seed
jobs = []
# Prepare job requests
for i in range(number_of_images):
current_seed = seed + i
input_data = {
"model": model_urn,
"params": {
"prompt": prompt,
"negativePrompt": negative_prompt,
"scheduler": "EulerA",
"steps": steps,
"cfgScale": cfg_scale,
"width": width,
"height": height,
"clipSkip": 2,
"seed": current_seed
}
}
if add_LORA:
try:
lora_data = json.loads(add_LORA)
if "additionalNetworks" in lora_data:
input_data["additionalNetworks"] = lora_data["additionalNetworks"]
except Exception as e:
print(f"Error processing LORA data: {str(e)}")
response = civitai.image.create(input_data)
if 'token' not in response or 'jobs' not in response:
raise ValueError("Invalid API response")
jobs.append({
'token': response['token'],
'job_id': response['jobs'][0]['jobId'],
'input_data': input_data
})
# Save links if DO_NOT_WAIT is True
if DO_NOT_WAIT:
date_str = time.strftime("%d_%B_%Y").lower()
base_name = f"{date_str}_"
existing_files = [f for f in os.listdir(self.links_dir) if f.startswith(base_name) and f.endswith(".txt")]
next_number = max([int(f[len(base_name):-4]) for f in existing_files] or [0]) + 1
file_name = f"{date_str}_{next_number:03d}.txt"
file_path = os.path.join(self.links_dir, links_file if links_file else file_name)
mode = 'a' if links_file else 'w'
if not file_path.endswith(".txt"):
file_path += ".txt"
with open(file_path, mode) as f:
for job in jobs:
if LIST_from_style_selector:
f.write(f"{LIST_from_style_selector};Token: {job['token']};Job ID: {job['job_id']}\n")
else:
f.write(f"Token: {job['token']};Job ID: {job['job_id']}\n")
generation_info = {
"status": "links_saved",
"links_file": file_path,
"timestamp": time.strftime("%Y-%m-%d %H:%M:%S"),
"number_of_jobs": len(jobs)
}
return (empty_image, json.dumps(generation_info, indent=2))
# Generate images immediately (DO_NOT_WAIT=False)
images = []
infos = []
failed_jobs = []
for job in jobs:
try:
image_url = self.check_job_status(job['token'], job['job_id'], timeout)
image_response = requests.get(image_url)
if image_response.status_code != 200:
raise ConnectionError(f"Image download failed: {image_response.status_code}")
img = Image.open(BytesIO(image_response.content)).convert('RGB')
img_tensor = torch.from_numpy(np.array(img).astype(np.float32) / 255.0)
images.append(img_tensor.unsqueeze(0))
infos.append(self.format_generation_info(job['input_data'], job['token'], job['job_id'], image_url))
except Exception as e:
failed_jobs.append({'job': job, 'error': str(e)})
if not images:
generation_info = {"error": "All jobs failed", "failed_jobs": failed_jobs}
return (empty_image, json.dumps(generation_info, indent=2))
combined_tensor = torch.cat(images, dim=0)
combined_info = {
"successful_generations": len(images),
"total_requested": number_of_images,
"individual_results": infos,
"failed_jobs": failed_jobs if failed_jobs else None
}
return (combined_tensor, json.dumps(combined_info, indent=2))
def check_job_status(self, job_token, job_id, timeout=9999):
"""Check job status with timeout"""
@@ -341,196 +429,183 @@ class APIGenerateCivitAI:
raise InterruptedError("Generation interrupted by user")
raise TimeoutError(f"Job timed out after {timeout} seconds")
def save_image_and_metadata(self, img, generation_info, number):
"""Save both image and its metadata"""
# Save image
filename = f"{number:04d}.png"
filepath = os.path.join(self.output_dir, filename)
img.save(filepath)
def format_generation_info(self, input_data, token, job_id, image_url):
"""Format generation info (implementation assumed)."""
return {"token": token, "job_id": job_id, "image_url": image_url}
# Save metadata
metadata_filename = f"{number:04d}_metadata.json"
metadata_filepath = os.path.join(self.metadata_dir, metadata_filename)
with open(metadata_filepath, 'w') as f:
json.dump(generation_info, f, indent=4)
return filepath, metadata_filepath
def format_generation_info(self, input_data, job_token, job_id, image_url):
"""Format generation information for recovery"""
recovery_info = {
"timestamp": time.strftime("%Y-%m-%d %H:%M:%S"),
"generation_parameters": input_data,
"job_details": {
"token": job_token,
"job_id": job_id,
"image_url": image_url
class LoadCivitAILinks:
@classmethod
def INPUT_TYPES(cls):
"""Define the input types for the node."""
return {
"required": {
"api_token": ("STRING", {"default": "", "placeholder": "CivitAI API token"}),
"links_file_path": ("STRING", {
"default": "",
"placeholder": "Path to links file (priority if not empty)"
}),
"selected_file": (["Not selected"] + cls.get_links_files(), {
"default": "Not selected"
}),
"direct_links": ("STRING", {
"multiline": True,
"default": "",
"placeholder": "Enter links directly (e.g., Style;Model;URN;Link;Token: <token>;Job ID: <job_id>)"
}),
},
"recovery_command": f"curl -X GET '{image_url}' --output recovered_image.png",
"recovery_instructions": """
To recover this image:
1. Use the provided curl command to download the image
2. Or use the image_url directly in a browser
3. If the image is no longer available, you can retry generation with the same parameters
"""
"optional": {
"auto_save": ("BOOLEAN", {
"default": False,
"label_on": "Enable Auto-Save",
"label_off": "Disable Auto-Save"
}),
}
}
return recovery_info
def generate_single_image(self, input_data, job_token, job_id, timeout):
"""Generate a single image and return its tensor and info"""
try:
image_url = self.check_job_status(job_token, job_id, timeout)
if not image_url:
raise ValueError("No image URL received")
RETURN_TYPES = ("IMAGE", "STRING", "STRING")
RETURN_NAMES = ("images", "status_info", "LIST_style")
OUTPUT_IS_LIST = (False, False, True)
FUNCTION = "load_images"
CATEGORY = "Civitai"
image_response = requests.get(image_url)
if image_response.status_code != 200:
raise ConnectionError(f"Failed to download image: Status code {image_response.status_code}")
img = Image.open(BytesIO(image_response.content))
if img.mode != 'RGB':
img = img.convert('RGB')
number = self.get_next_number()
generation_info = self.format_generation_info(input_data, job_token, job_id, image_url)
image_path, metadata_path = self.save_image_and_metadata(img, generation_info, number)
img_tensor = torch.from_numpy(np.array(img).astype(np.float32) / 255.0)
img_tensor = img_tensor.unsqueeze(0)
return img_tensor, generation_info
except Exception as e:
raise Exception(f"Error generating single image: {str(e)}")
def generate(self, api_token, prompt, negative_prompt, width, height, model_urn, steps=20,
cfg_scale=7.0, seed=-1, number_of_images=1, timeout=300, add_LORA=""):
# Set the environment variable
if api_token:
os.environ["CIVITAI_API_TOKEN"] = api_token
# Get a fresh instance of civitai with the new token
civitai = get_civitai()
self._interrupt_event.clear()
empty_image = torch.zeros((1, height, width, 3))
try:
# Handle seed
if seed == -1:
seed = random.randint(0, 0x7FFFFFFFFFFFFFFF)
# Prepare jobs list
jobs = []
generation_tasks = []
for i in range(number_of_images):
current_seed = seed + i
input_data = {
"model": model_urn,
"params": {
"prompt": prompt,
"negativePrompt": negative_prompt,
"scheduler": "EulerA",
"steps": steps,
"cfgScale": cfg_scale,
"width": width,
"height": height,
"clipSkip": 2,
"seed": current_seed
}
}
# Handle add_LORA input if provided
if add_LORA:
try:
lora_data = json.loads(add_LORA)
if "additionalNetworks" in lora_data:
input_data["additionalNetworks"] = lora_data["additionalNetworks"]
except Exception as e:
print(f"Error processing LORA data: {str(e)}")
# Create generation job
response = civitai.image.create(input_data)
if not response or 'token' not in response or 'jobs' not in response:
raise ValueError("Invalid response from Civitai API")
jobs.append({
'token': response['token'],
'job_id': response['jobs'][0]['jobId'],
'input_data': input_data
})
# Process all jobs in parallel
images = []
infos = []
failed_jobs = []
for job in jobs:
try:
img_tensor, generation_info = self.generate_single_image(
job['input_data'],
job['token'],
job['job_id'],
timeout
)
images.append(img_tensor)
infos.append(generation_info)
except Exception as e:
failed_jobs.append({
'job': job,
'error': str(e)
})
if not images: # If all jobs failed
generation_info = {
"error": "All generation jobs failed",
"timestamp": time.strftime("%Y-%m-%d %H:%M:%S"),
"failed_jobs": failed_jobs
}
return (empty_image, json.dumps(generation_info, indent=2))
# Combine images into a batch
combined_tensor = torch.cat(images, dim=0)
# Combine generation info
combined_info = {
"successful_generations": len(images),
"total_requested": number_of_images,
"base_seed": seed,
"generation_parameters": jobs[0]['input_data'],
"individual_results": infos,
"failed_jobs": failed_jobs if failed_jobs else None
}
return (combined_tensor, json.dumps(combined_info, indent=2))
except InterruptedError:
generation_info = {
"error": "Generation interrupted by user",
"timestamp": time.strftime("%Y-%m-%d %H:%M:%S"),
"base_seed": seed
}
return (empty_image, json.dumps(generation_info, indent=2))
except Exception as e:
generation_info = {
"error": f"Civitai generation failed: {str(e)}",
"timestamp": time.strftime("%Y-%m-%d %H:%M:%S"),
"base_seed": seed if 'seed' in locals() else None
}
return (empty_image, json.dumps(generation_info, indent=2))
def __init__(self):
"""Initialize the node with the links directory."""
self.links_dir = "Bjornulf/civitai_links"
os.makedirs(self.links_dir, exist_ok=True)
@classmethod
def IS_CHANGED(cls, **kwargs):
def get_links_files(cls):
links_dir = "Bjornulf/civitai_links"
if not os.path.exists(links_dir):
return []
files = [f for f in os.listdir(links_dir) if f.endswith(".txt")]
return files
def load_images(self, api_token, links_file_path, selected_file, direct_links, auto_save=False):
"""Load images from links and optionally save them to style-based folders."""
if not api_token:
raise ValueError("API token is required")
os.environ["CIVITAI_API_TOKEN"] = api_token
civitai = get_civitai()
# Determine the source of links
lines = None
if links_file_path:
if not os.path.exists(links_file_path):
raise ValueError(f"File path '{links_file_path}' does not exist")
with open(links_file_path, 'r') as f:
lines = f.readlines()
elif selected_file != "Not selected":
file_path = os.path.join(self.links_dir, selected_file)
if not os.path.exists(file_path):
raise ValueError(f"Selected file '{file_path}' does not exist")
with open(file_path, 'r') as f:
lines = f.readlines()
elif direct_links:
lines = direct_links.splitlines()
else:
raise ValueError("No valid links source provided")
images = []
list_styles = [] # To store LIST_style strings
status_info = {
"loaded": 0,
"failed": 0,
"attempted": 0,
"timestamp": time.strftime("%Y-%m-%d %H:%M:%S")
}
for line in lines:
line = line.strip()
if not line:
continue
status_info["attempted"] += 1
try:
parts = line.split(";")
if len(parts) == 6:
style = parts[0].strip()
model_name = parts[1].strip()
model_urn = parts[2].strip()
model_link = parts[3].strip()
token = parts[4].split("Token: ")[1].strip()
job_id = parts[5].split("Job ID: ")[1].strip()
list_style = ';'.join(parts[:4])
elif len(parts) == 2 and "Token: " in parts[0] and "Job ID: " in parts[1]:
token = parts[0].split("Token: ")[1].strip()
job_id = parts[1].split("Job ID: ")[1].strip()
list_style = ""
else:
raise ValueError(f"Invalid link format: {line}")
# Fetch job status from CivitAI API
response = civitai.jobs.get(token=token)
job_status = next((job for job in response['jobs'] if job['jobId'] == job_id), None)
if not job_status or not job_status['result'].get('available'):
status_info["failed"] += 1
continue
# Download and process the image
image_url = job_status['result'].get('blobUrl')
image_response = requests.get(image_url)
if image_response.status_code != 200:
status_info["failed"] += 1
continue
img = Image.open(BytesIO(image_response.content))
if img.mode != 'RGB':
img = img.convert('RGB')
# Auto-save if enabled and style is available
if auto_save and len(parts) == 6:
style_folder = style.replace(" ", "_") # Replace spaces with underscores
save_dir = os.path.join(folder_paths.get_output_directory(), "civitai_autosave", style_folder)
os.makedirs(save_dir, exist_ok=True)
file_name = f"{job_id}.png"
file_path = os.path.join(save_dir, file_name)
img.save(file_path)
# Convert to tensor and collect
img_tensor = torch.from_numpy(np.array(img).astype(np.float32) / 255.0)
images.append(img_tensor.unsqueeze(0))
list_styles.append(list_style)
status_info["loaded"] += 1
except Exception as e:
status_info["failed"] += 1
print(f"Error processing link '{line}': {str(e)}")
if not images:
raise ValueError("No images loaded from the provided links")
combined_tensor = torch.cat(images, dim=0)
return (combined_tensor, json.dumps(status_info, indent=2), list_styles)
@classmethod
def IS_CHANGED(cls, api_token, links_file_path, selected_file, direct_links, auto_save):
"""Force node re-execution when inputs change."""
return float("NaN")
def interrupt(self):
"""Method to handle interruption"""
print("Interrupting CivitAI generation...")
self._interrupt_event.set()
@PromptServer.instance.routes.post("/get_civitai_links_files")
async def get_civitai_links_files(request):
try:
links_dir = "Bjornulf/civitai_links"
if not os.path.exists(links_dir):
return web.json_response({
"success": False,
"error": "Links directory does not exist"
}, status=404)
files = [f for f in os.listdir(links_dir) if f.endswith(".txt")]
return web.json_response({
"success": True,
"files": files
}, status=200)
except Exception as e:
error_msg = str(e)
return web.json_response({
"success": False,
"error": error_msg
}, status=500)
class APIGenerateCivitAIAddLORA:
@classmethod
def INPUT_TYPES(cls):

347
README.md
View File

@@ -1,14 +1,19 @@
# 🔗 Comfyui : Bjornulf_custom_nodes v0.77 🔗
A list of 142 custom nodes for Comfyui : Display, manipulate, create and edit text, images, videos, loras, generate characters and more.
A list of 162 custom nodes for Comfyui : Display, manipulate, create and edit text, images, videos, loras, generate characters and more.
You can manage looping operations, generate randomized content, trigger logical conditions, pause and manually control your workflows and even work with external AI tools, like Ollama or Text To Speech.
# Watch Video (Quick overview 28 minutes) :
⚠️ Warning : Very active development. Work in progress. 🏗
# Watch Video (Quick overview 28 minutes) + Updates in video playlist :
[![Watch the video](https://img.youtube.com/vi/jTg9QsgKYmA/hqdefault.jpg)](https://www.youtube.com/playlist?list=PL_hRi_PPLLWvUNfG_np_d3K0nBCabM3iQ)
# Coffee : ☕☕☕☕☕ 5/5
Support me and my work : ❤️❤️❤️ <https://ko-fi.com/bjornulf> ❤️❤️❤️
Very active development, new nodes, new features, new ideas, new possibilities...
Let me know if you have any issues, they must be solved for everyone !
Support me, this project and my other AI exciting projects : ❤️❤️❤️ <https://ko-fi.com/bjornulf> ❤️❤️❤️
# ☘ This project is part of my AI trio. ☘
@@ -29,7 +34,7 @@ Support me and my work : ❤️❤️❤️ <https://ko-fi.com/bjornulf> ❤️
`126.` [📒 Note](#126----note)
`127.` [🖼📒 Image Note (Load image)](#130----image-note-load-image)
`128.` [🖼👁 Preview (first) image](#128----preview-first-image)
`130.` [📥🖼📒 Image Note (Load image)](#130----image-note-load-image)
`130.` [📥🖼📒 Image Note (Load image)](#130----image-note-load-image)
`133.` [🖼👁 Preview 1-4 images (compare)](#128----preview-first-image)
## ✒ Text ✒
@@ -57,9 +62,10 @@ Support me and my work : ❤️❤️❤️ <https://ko-fi.com/bjornulf> ❤️
`138.` [📑👈 Select from List](#138----select-from-list)
`141.` [🌎✒👉 Global Write Pick Me](#141----global-write-pick-me)
`142.` [🌎📥 Load Global Pick Me](#142----load-global-pick-me)
`144` [📊🔍 Text Analyzer](#)
## 🔥 Text Generator 🔥
`81.` [🔥📝 Text Generator 📝🔥](#81----text-generator-)
`81.` [🔥📝🖼 Image Text Generator 🖼📝🔥](#81----text-generator-)
`82.` [👩‍🦰📝 Text Generator (Character Female)](#82----text-generator-character-female)
`83.` [👨‍🦰📝 Text Generator (Character Male)](#83----text-generator-character-male)
`84.` [👾📝 Text Generator (Character Creature)](#84----text-generator-character-creature)
@@ -147,6 +153,8 @@ Support me and my work : ❤️❤️❤️ <https://ko-fi.com/bjornulf> ❤️
`62.` [🦙👁 Ollama Vision](#62----ollama-vision)
`70.` [📏 Resize Image Percentage](#70----resize-image-percentage)
`80.` [🩷 Empty Latent Selector](#80----empty-latent-selector)
`146` [🖼📹🔪 Split Image/Video Grid](#)
`147` [🖼📹🔨 Reassemble Image/Video Grid](#)
## 🅰️ Variables 🅰️
`3.` [✒🗔🅰️ Advanced Write Text (+ 🎲 random option)](#3---🅰%EF%B8%8F-advanced-write-text---random-option)
@@ -174,6 +182,7 @@ Support me and my work : ❤️❤️❤️ <https://ko-fi.com/bjornulf> ❤️
`108.` [☁👑 Add Lora (API ONLY - CivitAI) 👑☁](#108----add-lora-api-only---civitai-)
`109.` [☁🎨 API Image Generator (Black Forest Labs - Flux) ☁](#109----api-image-generator-black-forest-labs---flux-)
`110.` [☁🎨 API Image Generator (Stability - Stable Diffusion) ☁](#110----api-image-generator-stability---stable-diffusion-)
`151` [📥🕑🤖 Load CivitAI Links](#)
## 📥 Take from CivitAI / Hugginface 📥
`98.` [📥 Load checkpoint SD1.5 (+Download from CivitAi)](#98----load-checkpoint-sd15-download-from-civitai)
@@ -193,7 +202,7 @@ Support me and my work : ❤️❤️❤️ <https://ko-fi.com/bjornulf> ❤️
`49.` [📹👁 Video Preview](#49----video-preview)
`50.` [🖼➜📹 Images to Video path (tmp video)](#50----images-to-video-path-tmp-video)
`51.` [📹➜🖼 Video Path to Images](#51----video-path-to-images)
`52.` [🔊📹 Audio Video Sync](#52----audio-video-sync)
`52.` [🔊📹 Audio Video Sync](#52----audio-video-sync)
`58.` [📹🔗 Concat Videos](#58----concat-videos)
`59.` [📹🔊 Combine Video + Audio](#59----combine-video--audio)
`60.` [🖼🖼 Merge Images/Videos 📹📹 (Horizontally)](#60----merge-imagesvideos--horizontally)
@@ -203,6 +212,9 @@ Support me and my work : ❤️❤️❤️ <https://ko-fi.com/bjornulf> ❤️
`78.` [📹➜📹 Convert Video](#78----convert-video)
`79.` [📹🔗 Concat Videos from list](#79----concat-videos-from-list)
`119.` [📥👑📹 Load Lora Hunyuan Video (+Download from CivitAi)](#119----load-lora-hunyuan-video-download-from-civitai)
`146` [🖼📹🔪 Split Image/Video Grid](#)
`147` [🖼📹🔨 Reassemble Image/Video Grid](#)
`149` [💾📹 Save Video (tmp_api.mp4/mkv/webm) ⚠️💣](#)
## 🤖 AI 🤖
`19.` [🦙💬 Ollama Talk](#19----ollama-talk)
@@ -213,6 +225,7 @@ Support me and my work : ❤️❤️❤️ <https://ko-fi.com/bjornulf> ❤️
`65.` [🦙 Ollama Persona Selector 🧑](#65----ollama-persona-selector-)
`66.` [🔊➜📝 STT - Speech to Text](#66----stt---speech-to-text)
`118.` [🔊 TTS Configuration ⚙](#118----tts-configuration-)
`160.` [🦙👁 Ollama Vision Prompt Selector](#)
## 🔊 Audio 🔊
`31.` [📝➜🔊 TTS - Text to Speech](#31----tts---text-to-speech-100-local-any-voice-you-want-any-language)
@@ -222,10 +235,13 @@ Support me and my work : ❤️❤️❤️ <https://ko-fi.com/bjornulf> ❤️
`118.` [🔊 TTS Configuration ⚙](#118----tts-configuration-)
`120.` [📝➜🔊 Kokoro - Text to Speech](#120----kokoro---text-to-speech)
`134.` [134 - 🔊▶ Play Audio](#134----play-audio)
`145` [🔊▶ Audio Preview (Audio player)](#)
`148` [💾🔊 Save Audio (tmp_api.wav/mp3) ⚠️💣](#)
## 💻 General / System 💻
`34.` [🧹 Free VRAM hack](#34----free-vram-hack)
`137.` [🌎🎲 Global Seed Manager](#137----global-seed-manager)
`143` [🧮 Basic Math](#)
## 🧍 Manual user Control 🧍
`35.` [⏸️ Paused. Resume or Stop, Pick 👇](#35---%EF%B8%8F-paused-resume-or-stop-)
@@ -237,6 +253,19 @@ Support me and my work : ❤️❤️❤️ <https://ko-fi.com/bjornulf> ❤️
## 🧠 Logic / Conditional Operations 🧠
`45.` [🔀 If-Else (input / compare_with)](#45----if-else-input--compare_with)
## [Coming soon] - not sorted in the categories above yet...
`150` [[Coming soon] 🎨📜 Style Selector (🎲 or ♻ or ♻📑) + Civitai urn](#)
`152` [[Coming soon] 💾 Save Tensors (tmp_api.pt) ⚠️💣](#)
`153` [[Coming soon] 📥 Load Tensor (tmp_api.pt)](#)
`154` [[Coming soon] 🔮 Remote VAE Decoder](#)
`155` [[Coming soon] 🔮 Remote VAE Decoder (Tiled)](#)
`156` [[Coming soon] 📥🔮 Load from Base64](#)
`157` [[Coming soon] 🔮⚡ Execute Workflow](#)
`158` [[Coming soon] 📥🔮📝 Text Manager Api (Execute Workflow)](#)
`159` [[Coming soon] 🔥📝📹 Video Text Generator 📹📝🔥](#)
`161` [[Coming soon] 🔧🧑 Fix Face](#)
`162` [[Coming soon] 🧑 Face Settings [Fix Face] ⚙](#)
# ☁ Usage in cloud :
Comfyui is great for local usage, but I sometimes need more power than what I have...
@@ -283,18 +312,18 @@ Then you can download it from the file manager JupyterLab.
If you have any issues with this template from Runpod, please let me know, I'm here to help. 😊
# 🏗 Dependencies (nothing to do for runpod ☁)
# 🏗 Dependencies (nothing to do for my runpod ☁)
Check `requirements.txt` for the latest dependencies...
Check `requirements.txt` for the latest dependencies.
## 🪟🐍 Windows : Install dependencies on windows with embedded python (portable version)
First you need to find this python_embedded `python.exe`, then you can right click or shift + right click inside the folder in your file manager to open a terminal there.
This is where I have it, with the command you need :
`H:\ComfyUI_windows_portable\python_embeded> .\python.exe -m pip install pydub ollama opencv-python`
`H:\ComfyUI_windows_portable\python_embeded> .\python.exe -m pip install ollama pydub opencv-python ffmpeg-python civitai-py fal_client sounddevice langdetect spacy textblob dill`
When you have to install something you can retake the same code and install the dependency you want :
When you have to install something else, you can retake the same code and install the dependency you want :
`.\python.exe -m pip install whateveryouwant`
You can then run comfyui.
@@ -412,7 +441,7 @@ Fix a lot of code everywhere, a little better logging system, etc...
WIP : Rewrite of all my ffmpeg nodes. (Still need improvements and fixes, will do that in 0.71?) Maybe don't use them yet...
- **0.71**: ❗Breaking changes for Global variable nodes. (add to global variable system a "filename", which is a a separate global variable file.) bug fix speech to text node, 5 new nodes 129-133. combine text limit raised to 100. improve Save image in folder node.
- **0.71-0.75**: Many bug fixing. Civitai nodes are working on windows. (encoding, links problem are solved ? - at least on my machines...)
- **0.76**: Removed kokoro_onnx from requirements.txt due to some conflict with other nodes (need to be installed manually if you want to use this node.)
- **0.76**: Removed kokoro_onnx from requirements.txt due to some conflict with other nodes, like Impact-Pack, with old version of numpy... (kokoro_onnx need to be installed manually if you want to use this node.)
New syntaxes for advanced text/line selector, ex: {left|right|middle|group=LMR}+{left|right|middle|group=LMR}+{left|right|middle|group=LMR} and {A(80%)|B(15%)|C(5%)}
2 new nodes switch : 🔛✨ Anything Switch On/Off (compatible with combine images) AND 🔛📝 Text Switch On/Off (Compatible with combine texts)
2 new pick Me global nodes, using an identifier instead of chain : 🌎✒👉 Global Write Pick Me AND 🌎📥 Load Global Pick Me
@@ -420,6 +449,15 @@ New syntaxes for advanced text/line selector, ex: {left|right|middle|group=LMR}+
1 new node to quickly select element from list : 📑👈 Select from List
1 new audio node : 🔊▶ Play Audio (Just play an audio file, will default to bell.m4a if none provided.) Can take AUDIO format or audio_path.
❗Breaking changes. Large rewrite for all FFMPEG related nodes. With options for video preview. (Still have few changes to make, next version.)
- **0.77**: Add Refresh button to Load image from folder.
Add new syntax for advanced write / line selector : 2 {apple|orange|banana|static_group=FRUIT}s, one {apple|orange|banana|static_group=FRUIT} on the left, one {apple|orange|banana|static_group=FRUIT} on the right.
TTS fix request voices.
❗Breaking changes. New vision model for Ollama.
20 new nodes (143 - 162), Text Analyzer, basic math, audio preview, etc..... (10 not realesed yet, but included as a public "roadmap".)
Remove faster-whisper from requirements.txt for people using python 3.13. (Need to install manually.)
Fix preview image node for the new comfyui layout.
Rewrite of FFMPEG nodes, COnfiguration and Video preview.
Many other bug fixes and improvements on many nodes, thx for the reports.
# 📝 Nodes descriptions
@@ -437,7 +475,9 @@ These are convenient because these are automatically recommended on drag and dro
## 2 - ✒ Write Text
**Description:**
Simple node to write text.
Simple node to write text / send text to another node.
Below is a simple example of how to use the write text node with a show node :
![write Text](screenshots/write.png)
@@ -464,6 +504,9 @@ Usage example :
Groups, with no duplicate, example : {left|right|middle|group=LMR}+{left|right|middle|group=LMR}+{left|right|middle|group=LMR}
Random based on percentage : {A(80%)|B(15%)|C(5%)}
❗ 0.77 - New syntax available :
2 {apple|orange|banana|static_group=FRUIT}s, one {apple|orange|banana|static_group=FRUIT} on the left, one {apple|orange|banana|static_group=FRUIT} on the right
## 4 - 🔗 Combine Texts
**Description:**
@@ -559,7 +602,9 @@ Example of usage to see the differences between different combinations :
## 13/14 - 📏 + 🖼 Resize and Save Exact name ⚠️💣
**Description:**
Resize an image to exact dimensions. The other node will save the image to the exact path.
Resize an image to exact dimensions. If height or width is set to 0, it will keep the aspect ratio.
The other node will save the image to the exact path.
⚠️💣 Warning : The image will be overwritten if it already exists.
![Resize and Save Exact](screenshots/resize_save_exact.png)
@@ -1213,7 +1258,7 @@ Here is one possible example for videos with node 60 and 61 :
### 62 - 🦙👁 Ollama Vision
**Description:**
Take an image as input and will describe the image. Uses `moondream` by default, but can select anything with node 63.
Take an image as input and will describe the image. Uses `moondream` by default, but can select anything with node 63, and a custom prompt with node 161.
![ollama vision](screenshots/ollama_vision.png)
@@ -1249,6 +1294,10 @@ Below, an example of a crazy scientist explaining gravity. (Notice that the LLM
**Description:**
Use `faster-whisper` to transform an AUDIO type or audio_path into text. (Autodetect language)
⚠️ Warning, this node is not compatible with python `3.13`, you need to use `3.12`. (As a gereneral rule, comfyui ecosystem still recommend you to use 3.12 : https://github.com/comfyanonymous/ComfyUI?tab=readme-ov-file#manual-install-windows-linux.)
(faster-whisper was removed from requirements.txt and must be installed manually.)
If you really want to use that node, you are responsible for installing that manually : `pip install faster-whisper`.
![stt](screenshots/stt_1.png)
![stt](screenshots/stt_2.png)
@@ -1369,7 +1418,7 @@ Just connect that to your KSampler.
![empty_latent](screenshots/empty_latent.png)
#### 81 - 🔥📝 Text Generator 📝🔥
#### 81 - 🔥📝🖼 Image Text Generator 🖼📝🔥
**Description:**
Main node to generate content, doesn't really do much by itself, just `camera angle` and `multicharacter action`. (For example : `... eat picnic, view from above.`)
@@ -1728,6 +1777,9 @@ So use that if you want to ignore a line.
Groups, with no duplicate, example : {left|right|middle|group=LMR}+{left|right|middle|group=LMR}+{left|right|middle|group=LMR}
Random based on percentage : {A(80%)|B(15%)|C(5%)}
❗ 0.77 - New syntax available :
2 {apple|orange|banana|static_group=FRUIT}s, one {apple|orange|banana|static_group=FRUIT} on the left, one {apple|orange|banana|static_group=FRUIT} on the right
#### 118 - 🔊 TTS Configuration ⚙
**Description:**
@@ -1821,6 +1873,8 @@ It's behavior is like a "Preview image" node. (See node 130 if you want a behavi
Sometimes I want to display an image to explain what something specific is doing visually. (For example a stack of loras will have a specific style.)
Here is a complex example on how i use that, for a list of loras stacks. (I then "select" a style by using node `125 - Model-Clip-Vae selector`)
**0.77** Multiline + image is now an issue on newer comfyui versions. Replaced with single line notes, couldn't find any clean js hack to solve the layout issue...
![Image note](screenshots/image_notes.png)
#### 128 - 🖼👁 Preview (first) image
@@ -1835,6 +1889,8 @@ Below is a visual example of what I just said :
![First image preview](screenshots/first_image_preview.png)
**0.77** Multiline + image is now an issue on newer comfyui versions. Replaced with single line notes, couldn't find any clean js hack to solve the layout issue...
#### 129 - 📌🅰️ Set Variable from Text
**Description:**
@@ -1919,7 +1975,7 @@ Well now you can quickly enable / disable it. (If disabled you will see it in re
![switch_text](screenshots/switch_text.png)
If connected with my combine text node, you can use a special option `ONLY_ME_combine_text` that will tell combine text to write ONLY the selected node. It will ignore all the otehrs. (Here will appear in blue.) :
If connected with my combine text node, you can use a special option `ONLY_ME_combine_text` that will tell combine text to write ONLY the selected node. It will ignore all the others. (Here will appear in blue.) :
![switch_text_onlyme](screenshots/switch_text_onlyme.png)
@@ -1987,4 +2043,261 @@ This node also automatically return a random value from the list with the global
Below is an example of write + load :
![global_write_pickme_load](screenshots/global_write_pickme_load.png)
![global_write_pickme_load](screenshots/global_write_pickme_load.png)
#### 143 - 🧮 Basic Math
**Description:**
Basic node for basic math.
Addition, multiplication...
Here is an example that I use to detect the number of "missing" frames in a video, compared to another video :
![basic math](screenshots/basic_math.png)
#### 144 - 📊🔍 Text Analyzer
**Description:**
Display details about an input text.
You can use it just to recover the number of characters, you can then for example use this number to avoid too long prompts and avoid OOM :
![text_ana_1](screenshots/text_ana_1.png)
Or extract more details about dialog and sentiment polarity of it. (For example, if you want a specific TTS voice for positive/neutral/negative.) :
![text_ana_2](screenshots/text_ana_2.png)
#### 145 - 🔊▶ Audio Preview (Audio player)
**Description:**
Simple preview audio node that can take as input an audio_path, mp3 or wav.
![audio_preview](screenshots/audio_preview.png)
#### 146 - 🖼📹🔪 Split Image/Video Grid
**Description:**
This node will split a grid of images or videos into separate images or videos.
Below is an example with a grid of 6 images, it will split them into 6 separate images (3 rows X 3 columns) :
![split_grid](screenshots/split_grid.png)
Why do I need that ? Sometimes I have several characters in a single image/video.
When I want to apply lip sync to a character, I need to split them into separate images/videos where4 the character is located, create the lip sync, and then reassemble them with node 147.
Or just to save VRAM, by processing only the part of the image that I need.
#### 147 - 🖼📹🔨 Reassemble Image/Video Grid
**Description:**
Reassemble an image/video that was grid split with node 146.
Will need to take the same grid size as input.
You can apply changes to a specific element of the grid, and then reassemble them.
You can select each part individually.
But there is also a "special" part, the MODIFIED one.
This part is selected with the MODIFIED_part and the MODIFIED_part_index (part number, if index set to 2, it will use `part_2`).
Below is an example where I modify part_2 and reassmbling it. (Without using part_2 output but selectionning it using 2 as MODIFIED_part_index) :
You can also work directly with part_2, but twith this setup you can quickly switch in between parts without changing the node layout.
![grid_assemble1](screenshots/grid_assemble_1.png)
Below is an example where I show you that you can also influence different parts "manually". (Here, I just send part_1 to replace part 7,8,9.)
![grid_assemble2](screenshots/grid_assemble_2.png)
For reassembling videos it might get tricky.
For example if you used the node `52 - 🔊📹 Audio Video Sync`, the modified part might be shorter/longer, change FPS etc...
You can try to re-synchronize them all if the duration was modified.
In this situation you need to use the `reference_video_part_index` to select the reference video part. (Probably the same as the MODIFIED_part_index.)
![](screenshots/.png)
#### 148 - 💾🔊 Save Audio (tmp_api.wav/mp3) ⚠️💣
**Description:**
Temporary save for an audio file, mp3 or wav.
Useful if you want to synchronize data in between API calls from several Comfyui instances. (Used in the background by my API nodes, like node 157 `🔮⚡ Execute Workflow`, to exchange audio files. - raw/processed -)
You can send audio_path (This will create file `tmp_api.mp3`) :
![savetmp_audio1](screenshots/savetmp_audio1.png)
Or an AUDIO type (This will create file `tmp_api.wav`) :
![savetmp_audio2](screenshots/savetmp_audio2.png)
Note that if you are sending an AUDIO type, this node will save it as a `wav` file. (Even if your selection is an mp3 file.)
#### 149 - 💾📹 Save Video (tmp_api.mp4/mkv/webm) ⚠️💣
**Description:**
Temporary save for a video file, mp4, mkv or webm.
Useful if you want to synchronize data in between API calls from several Comfyui instances. (Used in the background by my API nodes, like node 157 `🔮⚡ Execute Workflow`, to exchange video files. - raw/processed -)
![savetmp_video](screenshots/savetmp_video.png)
#### 150 - 🎨📜 Style Selector (🎲 or ♻ or ♻📑) + Civitai urn
**Description:**
A new Universal node for selecting styles.
Also contains some informations for Civitai urn. (Need test and better "database" of good/fast models, i recommended you to send the urns you like manually.)
Here is an example of usage with CivitAI API node :
![style_gen_1](screenshots/style_gen_1.png)
Here is a slightly more complex example using model urns :
![style_gen_2](screenshots/style_gen_2.png)
NOTE : CivitAI can be particularly slow !! So you can use the option DO_NOT_WAIT instead with an optional "links_file".
It will create a file with the links to recover the generated images later with the node 151.
So you can make as many requests as you want, and come back another time to recover the images with this txt file.
Note that you can also send the LIST directly with all the values.
![style_gen_3](screenshots/style_gen_3.png)
If you do that you can later recover the values like text and use the "autosave" option of node 151.
It will take the style and make a folder for it. Example from the image above : `./ComfyUI/output/civitai_autosave/Cinematic_photography/`
It is very useful if you use node 150 as a LOOP, you can recover the images later and organise them automatically in separate folders according to the selected style.
#### 151 - 📥🕑🤖 Load CivitAI Links
**Description:**
This node must be used after using the node 150 + DO_NOT_WAIT + links_file.
When saving DO_NOT_WAIT and the "links_file" from the node 150, it will create a txt file in "ComfyUI/Bjornulf/civitai_links/", you can load them with this node.
Below are the images recovered from the file "11_mars.txt":
![load_civit](screenshots/load_civit.png)
If you used from node 150 the `LIST_from_style_selector` you can later recover the values and use the "autosave" option of node 151.
It will take the style and make a folder for it. Example from the image above : `./ComfyUI/output/civitai_autosave/Cinematic_photography/`
It is very useful if you use node 150 as a LOOP, you can recover the images later and organise them automatically in separate folders according to the selected style.
#### 152 - [Coming soon] 💾 Save Tensors (tmp_api.pt) ⚠️💣
**Description:**
This is BETA, it will save a tensor in a file.
The goal is to optimize the VRAM usage by saving tensors in files.
#### 153 - [Coming soon] 📥 Load Tensor (tmp_api.pt)
**Description:**
This is BETA, it will load a tensor.
The goal is to optimize the VRAM usage by saving tensors in files.
#### 154 - 🔮 Remote VAE Decoder
**Description:**
Simple node to decode a VAE model from a remote machine.
The VAE decode node doesn't have to be on the same machine, so if you have a remote machine and want to save the VRAM, you can use that.
The remote machine need to have my custom nodes too.
![remote_vae](screenshots/remote_vae.png)
#### 155 - [Coming soon] 🔮 Remote VAE Decoder (Tiled)
**Description:**
Same as 154 but with tiled version.
![remote_vae_tiled](screenshots/remote_vae_tiled.png)
#### 156 - [Coming soon] 📥🔮 Load from Base64
**Description:**
This is BETA, it is used to recovered some values between Comfyui instances.
#### 157 - [Coming soon] 🔮⚡ Remote Execute Workflow
**Description:**
This is the start of a complex new environment and this is in BETA.
The goal of this node is to run a workflow on another Comfyui instance and recover what was generated.
For now it supports image/sound/video.
It can do anything that a normal workflow can do.
Create an image, create a video, upscale an image, etc...
I will share custom JSON and a lot more cool stuff for my Kofi supporters. (https://ko-fi.com/bjornulf)
Keep in touch for more informations... Wip...
Basically use run heavy duty workflows on another machine, and recover the results.
Below is an example of a lipsync workflow that I run on another Comfyui instance. (For now local network only, but will run on runpod too.)
I use that on my laptop (192.168.1.23), that recovers the results from my "ai dedicated" desktop machine (192.168.1..100) :
![execute_worflow](screenshots/execute_worflow.png)
#### 158 - [Coming soon] 📥🔮📝 Text Manager Api (Execute Workflow)
**Description:**
This is BETA.
This node need to be used to create functional workflows to run with node 157.
It is communicating with node 157 from another Comfyui instance.
#### 159 - [Coming soon] 🔥📝📹 Video Text Generator 📹📝🔥
**Description:**
This is BETA.
This is the start of a new node that will generate a video prompt.
It's goal is to be like the "Text Generator" node, but for videos.
The "old" Text Generator main node 81 is now renamed for image only : 🔥📝🖼 Image Text Generator 🖼📝🔥.
![video_text_generator](screenshots/video_text_generator.png)
#### 160 - 🦙👁 Ollama Vision Prompt Selector
**Description:**
New layout prompt selector for Ollama Vision.
Ollama vision is now divided in 2 separate nodes.
![ollama vision prompt](screenshots/ollama_vision.png)
#### 161 - [Coming soon] 🔧🧑 Fix Face
**Description:**
This is BETA.
First node of a new collection of "Fix Images" nodes.
Quickly fix, control and edit faces. (and more)
![fix_face](screenshots/fix_face_1.png)
Here is a zoom of that :
![fix_face](screenshots/fix_face_2_zoom.png)
You can also give specific text and custom denoise for each faces :
![fix_face](screenshots/fix_face_3.png)
You can also select a specific face, if set at 0 it will do all the faces, but below I set it to 1 :
![fix_face](screenshots/fix_face_4_ignore.png)
This node can also fix the whole person, but i will likely make another separate node for that later :
![fix_face](screenshots/fix_face_5_person.png)
You can preview the nodes effects very well with my Preview Image node, You can clearly see here the white box that is the detected "person" :
![fix_face](screenshots/fix_face_6_preview.png)
In the future I will make a system to "inject" a specific reusable character, probably connected to loras or suchs.
#### 162 - [Coming soon] 🧑 Face Settings [Fix Face] ⚙
**Description:**
Connect to Fix Face node to modify a specific face.

72
__init__.py
View File

@@ -40,6 +40,7 @@ from .character_description import CharacterDescriptionGenerator
from .text_to_speech import TextToSpeech, XTTSConfig
from .loop_combine_texts_by_lines import CombineTextsByLines
from .free_vram_hack import FreeVRAM
#, PurgeCLIPNode
from .pause_resume_stop import PauseResume
from .pick_input import PickInput
from .loop_images import LoopImages
@@ -71,7 +72,7 @@ from .ffmpeg_combine_video_audio import CombineVideoAudio
from .images_merger_horizontal import MergeImagesHorizontally
from .images_merger_vertical import MergeImagesVertically
from .ollama_talk import OllamaTalk
from .ollama_image_vision import OllamaImageVision
from .ollama_image_vision import OllamaImageVision, OllamaVisionPromptSelector
from .ollama_config_selector import OllamaConfig
from .ollama_system_persona import OllamaSystemPersonaSelector
from .ollama_system_job import OllamaSystemJobSelector
@@ -87,7 +88,7 @@ from .ffmpeg_convert import ConvertVideo
from .text_generator import TextGenerator, TextGeneratorScene, TextGeneratorStyle, TextGeneratorCharacterFemale, TextGeneratorCharacterMale, TextGeneratorOutfitMale, TextGeneratorOutfitFemale, ListLooper, ListLooperScene, ListLooperStyle, ListLooperCharacter, ListLooperOutfitFemale, ListLooperOutfitMale, TextGeneratorCharacterPose, TextGeneratorCharacterObject, TextGeneratorCharacterCreature
from .API_flux import APIGenerateFlux
from .API_StableDiffusion import APIGenerateStability
from .API_civitai import APIGenerateCivitAI, APIGenerateCivitAIAddLORA, CivitAIModelSelectorPony, CivitAIModelSelectorSD15, CivitAIModelSelectorSDXL, CivitAIModelSelectorFLUX_S, CivitAIModelSelectorFLUX_D, CivitAILoraSelectorSD15, CivitAILoraSelectorSDXL, CivitAILoraSelectorPONY, CivitAILoraSelectorHunyuan
from .API_civitai import APIGenerateCivitAI, APIGenerateCivitAIAddLORA, CivitAIModelSelectorPony, CivitAIModelSelectorSD15, CivitAIModelSelectorSDXL, CivitAIModelSelectorFLUX_S, CivitAIModelSelectorFLUX_D, CivitAILoraSelectorSD15, CivitAILoraSelectorSDXL, CivitAILoraSelectorPONY, CivitAILoraSelectorHunyuan, LoadCivitAILinks
from .API_falAI import APIGenerateFalAI
from .latent_resolution_selector import LatentResolutionSelector
from .loader_lora_with_path import LoaderLoraWithPath
@@ -116,8 +117,48 @@ from .play_sound import PlayAudio
from .switches import SwitchText, SwitchAnything
from .write_pickme_global import WriteTextPickMeGlobal, LoadTextPickMeGlobal
from .list_selector import ListSelector
from .text_analyzer import TextAnalyzer
from .math_node import MathNode
from .save_tmp_audio import SaveTmpAudio
from .save_tmp_video import SaveTmpVideo
from .audio_preview import AudioPreview
from .style_selector import StyleSelector
# from .switches import ConditionalSwitch
from .split_image import SplitImageGrid, ReassembleImageGrid
# from .video_text_generator import VideoTextGenerator
# from .run_workflow_from_api import ExecuteWorkflowNode, ApiDynamicTextInputs
# from .remote_nodes import RemoteVAEDecoderNodeTiled, RemoteVAEDecoderNode, LoadFromBase64, SaveTensors, LoadTensor
# from .fix_face import FixFace, FaceSettings
#RemoteTextEncodingWithCLIPs
NODE_CLASS_MAPPINGS = {
# "Bjornulf_PurgeCLIPNode": PurgeCLIPNode,
# "Bjornulf_RemoteTextEncodingWithCLIPs": RemoteTextEncodingWithCLIPs,
# "Bjornulf_FixFace": FixFace,
# "Bjornulf_FaceSettings": FaceSettings,
# "Bjornulf_SaveTensors": SaveTensors,
# "Bjornulf_LoadTensor": LoadTensor,
# "Bjornulf_LoadFromBase64": LoadFromBase64,
# "Bjornulf_RemoteVAEDecoderNode": RemoteVAEDecoderNode,
# "Bjornulf_RemoteVAEDecoderNodeTiled": RemoteVAEDecoderNodeTiled,
# "Bjornulf_VideoTextGenerator": VideoTextGenerator,
# "Bjornulf_ExecuteWorkflowNode": ExecuteWorkflowNode,
# "Bjornulf_ApiDynamicTextInputs": ApiDynamicTextInputs,
# "Bjornulf_ConditionalSwitch": ConditionalSwitch,
"Bjornulf_LoadCivitAILinks": LoadCivitAILinks,
"Bjornulf_SplitImageGrid": SplitImageGrid,
"Bjornulf_ReassembleImageGrid": ReassembleImageGrid,
"Bjornulf_StyleSelector": StyleSelector,
"Bjornulf_OllamaVisionPromptSelector": OllamaVisionPromptSelector,
"Bjornulf_AudioPreview": AudioPreview,
"Bjornulf_SaveTmpAudio": SaveTmpAudio,
"Bjornulf_SaveTmpVideo": SaveTmpVideo,
"Bjornulf_MathNode": MathNode,
"Bjornulf_TextAnalyzer": TextAnalyzer,
"Bjornulf_ListSelector": ListSelector,
"Bjornulf_WriteTextPickMeGlobal": WriteTextPickMeGlobal,
"Bjornulf_LoadTextPickMeGlobal": LoadTextPickMeGlobal,
@@ -273,6 +314,31 @@ NODE_CLASS_MAPPINGS = {
}
NODE_DISPLAY_NAME_MAPPINGS = {
# "Bjornulf_RemoteTextEncodingWithCLIPs": "[BETA] 🔮 Remote Text Encoding with CLIPs",
# "Bjornulf_ConditionalSwitch": "ConditionalSwitch",
# "Bjornulf_PurgeCLIPNode": "🧹📎 Purge CLIP",
# "Bjornulf_FixFace": "[BETA] 🔧🧑 Fix Face",
# "Bjornulf_FaceSettings": "[BETA] 🧑 Face Settings [Fix Face] ⚙",
# "Bjornulf_SaveTensors": "[BETA] 💾 Save Tensors (tmp_api.pt) ⚠️💣",
# "Bjornulf_LoadTensor": "[BETA] 📥 Load Tensor (tmp_api.pt)",
# "Bjornulf_RemoteVAEDecoderNode": "[BETA] 🔮 Remote VAE Decoder",
# "Bjornulf_RemoteVAEDecoderNodeTiled": "[BETA] 🔮 Remote VAE Decoder (Tiled)",
# "Bjornulf_LoadFromBase64": "[BETA] 📥🔮 Load from Base64",
# "Bjornulf_ApiDynamicTextInputs": "[BETA] 📥🔮📝 Text Manager Api (Execute Workflow)",
# "Bjornulf_ExecuteWorkflowNode": "[BETA] 🔮⚡ Remote Execute Workflow",
# "Bjornulf_VideoTextGenerator": "[BETA] 🔥📝📹 Video Text Generator 📹📝🔥",
"Bjornulf_LoadCivitAILinks": "📥🕑🤖 Load CivitAI Links",
"Bjornulf_StyleSelector": "🎨📜 Style Selector (🎲 or ♻ or ♻📑) + Civitai urn",
"Bjornulf_ReassembleImageGrid": "🖼📹🔨 Reassemble Image/Video Grid",
"Bjornulf_SplitImageGrid": "🖼📹🔪 Split Image/Video Grid",
"Bjornulf_SaveTmpAudio": "💾🔊 Save Audio (tmp_api.wav/mp3) ⚠️💣",
"Bjornulf_SaveTmpVideo": "💾📹 Save Video (tmp_api.mp4/mkv/webm) ⚠️💣",
"Bjornulf_AudioPreview": "🔊▶ Audio Preview (Audio player)",
"Bjornulf_MathNode": "🧮 Basic Math",
"Bjornulf_TextAnalyzer": "📊🔍 Text Analyzer",
"Bjornulf_OllamaVisionPromptSelector": "🦙👁 Ollama Vision Prompt Selector",
"Bjornulf_ListSelector": "📑👈 Select from List",
"Bjornulf_PlayAudio": "🔊▶ Play Audio",
"Bjornulf_SwitchText": "🔛📝 Text Switch On/Off",
@@ -327,7 +393,7 @@ NODE_DISPLAY_NAME_MAPPINGS = {
"Bjornulf_APIGenerateCivitAIAddLORA": "☁👑 Add Lora (API ONLY - CivitAI) 👑☁",
"Bjornulf_APIGenerateFlux": "☁🎨 API Image Generator (Black Forest Labs - Flux) 🎨☁",
"Bjornulf_APIGenerateStability": "☁🎨 API Image Generator (Stability - Stable Diffusion) 🎨☁",
"Bjornulf_TextGenerator": "🔥📝 Text Generator 📝🔥",
"Bjornulf_TextGenerator": "🔥📝 Image Text Generator 📝🔥",
"Bjornulf_TextGeneratorCharacterFemale": "👩‍🦰📝 Text Generator (Character Female)",
"Bjornulf_TextGeneratorCharacterMale": "👨‍🦰📝 Text Generator (Character Male)",
"Bjornulf_TextGeneratorCharacterPose": "💃🕺📝 Text Generator (Character Pose)",

80
audio_preview.py Normal file
View File

@@ -0,0 +1,80 @@
import os
import shutil
import time
import uuid
import urllib.request
import urllib.parse
class AudioPreview:
@classmethod
def INPUT_TYPES(cls):
return {
"required": {
"audio_path": ("STRING", {"default": ""}),
"autoplay": ("BOOLEAN", {"default": False}),
"loop": ("BOOLEAN", {"default": False}),
}
}
RETURN_TYPES = ()
FUNCTION = "preview_audio"
CATEGORY = "audio"
OUTPUT_NODE = True
def preview_audio(self, audio_path, autoplay, loop):
try:
# Validate input
if not audio_path or not isinstance(audio_path, str) or not audio_path.strip():
raise ValueError("No valid audio path provided.")
# Set up destination directory
dest_dir = os.path.join("temp", "Bjornulf")
os.makedirs(dest_dir, exist_ok=True)
# Generate unique filename components
timestamp = int(time.time())
uuid_str = str(uuid.uuid4()).replace('-', '')[:8] # Short unique string
base_name = "Bjornulf"
if audio_path.startswith("http://") or audio_path.startswith("https://"):
# Handle URL input
parsed_url = urllib.parse.urlparse(audio_path)
path = parsed_url.path
ext = os.path.splitext(path)[1]
if not ext:
raise ValueError("URL does not have a file extension.")
dest_name = f"{base_name}_{timestamp}_{uuid_str}{ext}"
dest_path = os.path.join(dest_dir, dest_name)
try:
urllib.request.urlretrieve(audio_path, dest_path)
except Exception as e:
raise ValueError(f"Failed to download audio from URL: {audio_path}. Error: {e}")
else:
# Handle local file input
audio_path = os.path.abspath(audio_path)
if not os.path.exists(audio_path):
raise FileNotFoundError(f"Audio file not found: {audio_path}")
ext = os.path.splitext(audio_path)[1]
dest_name = f"{base_name}_{timestamp}_{uuid_str}{ext}"
dest_path = os.path.join(dest_dir, dest_name)
shutil.copy2(audio_path, dest_path)
# Return UI data for frontend
return {
"ui": {
"audio": [dest_name, "Bjornulf"],
"metadata": {
"autoplay": autoplay,
"loop": loop
}
}
}
except Exception as e:
# Handle errors gracefully
return {
"ui": {
"audio": [],
"error": str(e)
}
}

20
audio_video_sync.py
View File

@@ -306,39 +306,41 @@ class AudioVideoSync:
temp_dir = os.path.join(self.temp_dir, "temp_frames")
os.makedirs(temp_dir, exist_ok=True)
# Extract frames using ffmpeg
# Clear existing files to avoid mixing frames from previous runs
for file in os.listdir(temp_dir):
os.remove(os.path.join(temp_dir, file))
# Extract frames using FFmpeg
subprocess.run([
'ffmpeg', '-i', video_path,
os.path.join(temp_dir, 'frame_%05d.png')
], check=True)
'ffmpeg', '-y', '-i', video_path, os.path.join(temp_dir, 'frame_%05d.png')
], check=True) # Added '-y' to overwrite without prompting
# Load frames and convert to tensor
frames = []
frame_files = sorted(os.listdir(temp_dir))
transform = transforms.Compose([
transforms.ToTensor(),
transforms.Lambda(lambda x: x * 255) # Scale to 0-255 range
transforms.Lambda(lambda x: x * 255)
])
for frame_file in frame_files:
image = Image.open(os.path.join(temp_dir, frame_file))
image = Image.open(os.path.join(temp_dir, frame_file)).convert('RGB')
frame_tensor = transform(image)
frames.append(frame_tensor)
# Stack frames into a single tensor
frames_tensor = torch.stack(frames)
# Ensure the tensor is in the correct format (B, C, H, W)
# Ensure correct format (B, C, H, W)
if frames_tensor.dim() == 3:
frames_tensor = frames_tensor.unsqueeze(0)
# Convert to uint8
frames_tensor = frames_tensor.byte()
# Clean up temporary directory
# Clean up
for frame_file in frame_files:
os.remove(os.path.join(temp_dir, frame_file))
os.rmdir(temp_dir)
return frames_tensor

59
ffmpeg_images_to_video.py
View File

@@ -122,6 +122,48 @@ class imagesToVideo:
except Exception as e:
logging.error(f"Unexpected error in FFmpeg-python: {str(e)}")
return False, f"Error: {str(e)}"
def get_next_filename(self, output_base, format="mp4"):
"""
Determines the next filename in a sequence with 4-digit numbering (0001, 0002, etc.).
Args:
output_base (str): The base path and prefix (e.g., 'output/imgs2video/me')
format (str): The file extension (e.g., 'mp4')
Returns:
str: The next filename in the sequence (e.g., 'output/imgs2video/me_0001.mp4')
"""
# Ensure output_base is clean
output_base = output_base.rstrip(os.sep)
# Pattern to match files with 4-digit numbers: e.g., 'output/imgs2video/me_0001.mp4'
pattern = f"{output_base}_[0-9][0-9][0-9][0-9].{format}"
# Get all files matching the pattern
existing_files = glob.glob(pattern)
# Extract numbers from filenames
numbers = []
for filepath in existing_files:
# Get the filename from the full path
filename = os.path.basename(filepath)
# Extract the 4-digit number between '_' and '.'
number_part = filename.split('_')[-1].split('.')[0]
# Verify it's a 4-digit number
if number_part.isdigit() and len(number_part) == 4:
numbers.append(int(number_part))
# Determine the next number
if numbers:
next_num = max(numbers) + 1
else:
next_num = 1
# Construct the next filename with zero-padding to 4 digits
next_filename = f"{output_base}_{next_num:04d}.{format}"
return next_filename
def image_to_video(self, images, fps, name_prefix, use_python_ffmpeg=False, audio=None, audio_path=None, FFMPEG_CONFIG_JSON=None):
ffmpeg_config = self.parse_ffmpeg_config(FFMPEG_CONFIG_JSON)
@@ -130,27 +172,24 @@ class imagesToVideo:
if ffmpeg_config and ffmpeg_config["output"]["container_format"] != "None":
format = ffmpeg_config["output"]["container_format"]
# Remove any extension from name_prefix and create output_base
name_prefix = os.path.splitext(name_prefix)[0]
output_base = os.path.join("output", name_prefix)
existing_files = glob.glob(f"{output_base}_*.{format}")
if existing_files:
max_num = max([int(f.split('_')[-1].split('.')[0]) for f in existing_files])
next_num = max_num + 1
else:
next_num = 1
# Get the next filename using the corrected function
output_file = self.get_next_filename(output_base, format)
output_file = f"{output_base}_{next_num:04d}.{format}"
temp_dir = "Bjornulf/temp_images_imgs2video"
# Clean up and prepare temporary directory
temp_dir = os.path.join("Bjornulf", "temp_images_imgs2video") # Use os.path.join for cross-platform compatibility
if os.path.exists(temp_dir) and os.path.isdir(temp_dir):
for file in os.listdir(temp_dir):
os.remove(os.path.join(temp_dir, file))
os.rmdir(temp_dir)
# Create necessary directories
os.makedirs(temp_dir, exist_ok=True)
os.makedirs(os.path.dirname(output_file) if os.path.dirname(output_file) else ".", exist_ok=True)
for i, img_tensor in enumerate(images):
img = Image.fromarray((img_tensor.cpu().numpy() * 255).astype(np.uint8))
if format == "webm":

19
free_vram_hack.py
View File

@@ -56,4 +56,21 @@ class FreeVRAM:
response.raise_for_status()
print("HTTP request triggered successfully")
except requests.exceptions.RequestException as e:
print(f"Failed to trigger HTTP request: {e}")
print(f"Failed to trigger HTTP request: {e}")
class PurgeCLIPNode:
@classmethod
def INPUT_TYPES(cls):
return {"required": {}}
RETURN_TYPES = ()
FUNCTION = "purge_clip"
CATEGORY = "utils"
def purge_clip(self):
# Check if the CLIP model is accessible in the global scope
global clip_model
if 'clip_model' in globals():
del clip_model # Delete the CLIP model reference
torch.cuda.empty_cache() # Clear VRAM cache
return ()

8
global_seed_manager.py
View File

@@ -1,14 +1,12 @@
from server import PromptServer
import os
from aiohttp import web
import random
class GlobalSeedManager:
@classmethod
def INPUT_TYPES(cls):
return {"required": {"seed": ( "INT", {
"default": 0,
"default": 1,
"min": 0,
"max": 4294967294
})}}
@@ -19,8 +17,8 @@ class GlobalSeedManager:
CATEGORY = "Bjornulf"
def generate_seed(self, seed: int):
# Generate new random seed
new_seed = random.randint(0, 2**31 - 1)
# Use the provided seed instead of generating a new one
new_seed = seed
seed_str = str(new_seed)
# Define file path

4
if_else.py
View File

@@ -12,6 +12,7 @@ class IfElse:
"STRING: input EQUAL TO compare_with",
"STRING: input NOT EQUAL TO compare_with",
"BOOLEAN: input IS TRUE",
"BOOLEAN: input IS FALSE",
"NUMBER: input GREATER THAN compare_with",
"NUMBER: input GREATER OR EQUAL TO compare_with",
"NUMBER: input LESS THAN compare_with",
@@ -61,6 +62,9 @@ class IfElse:
elif input_type == "BOOLEAN: input IS TRUE":
result = str(input).lower() in ("true", "1", "yes", "y", "on")
details += f"\nChecked if '{input}' is considered True"
elif input_type == "BOOLEAN: input IS FALSE":
result = str(input).lower() in ("false", "0", "no", "n", "off")
details += f"\nChecked if '{input}' is considered True"
else: # Numeric comparisons
try:
input_num = float(input)

193
line_selector.py
View File

@@ -14,31 +14,28 @@ class LineSelector:
def INPUT_TYPES(s):
return {
"required": {
"text": ("STRING", {"multiline": True}), # Input for multiple lines
"line_number": ("INT", {"default": 0, "min": 0, "max": 99999}), # 0 for random, >0 for specific line
"RANDOM": ("BOOLEAN", {"default": False}), # Force random selection
"LOOP": ("BOOLEAN", {"default": False}), # Return all lines as list
"LOOP_SEQUENTIAL": ("BOOLEAN", {"default": False}), # Sequential looping
"jump": ("INT", {"default": 1, "min": 1, "max": 100, "step": 1}), # Jump size for sequential loop
"pick_random_variable": ("BOOLEAN", {"default": True}), # Enable random choice functionality
"text": ("STRING", {"multiline": True}),
"line_number": ("INT", {"default": 0, "min": 0, "max": 99999}),
"RANDOM": ("BOOLEAN", {"default": False}),
"LOOP": ("BOOLEAN", {"default": False}),
"LOOP_SEQUENTIAL": ("BOOLEAN", {"default": False}),
"jump": ("INT", {"default": 1, "min": 1, "max": 100, "step": 1}),
"pick_random_variable": ("BOOLEAN", {"default": True}),
},
"optional": {
"variables": ("STRING", {"multiline": True, "forceInput": True}),
"seed": ("INT", {
"default": -1,
"min": -1,
"max": 0x7FFFFFFFFFFFFFFF
}),
"seed": ("INT", {"default": -1, "min": -1, "max": 0x7FFFFFFFFFFFFFFF}),
},
}
RETURN_TYPES = ("STRING", "INT", "INT") # String output, remaining cycles, current line number
RETURN_TYPES = ("STRING", "INT", "INT")
RETURN_NAMES = ("text", "remaining_cycles", "current_line")
OUTPUT_IS_LIST = (True, False, False) # Only text output can be a list
OUTPUT_IS_LIST = (True, False, False)
FUNCTION = "select_line"
CATEGORY = "Bjornulf"
def find_variables(self, text):
"""Identify nested curly brace sections in the text."""
stack = []
variables = []
for i, char in enumerate(text):
@@ -55,6 +52,7 @@ class LineSelector:
return variables
def parse_option(self, part):
"""Parse options within curly braces, handling CSV and weighted choices."""
if part.startswith('%csv='):
try:
filename = part.split('=', 1)[1].strip()
@@ -67,61 +65,83 @@ class LineSelector:
return (option.strip(), float(weight.split('%)')[0]))
return part.strip()
def process_content(self, content, seed):
random.seed(seed)
parts = []
def process_content(self, content, base_seed, position):
"""Process content within curly braces, handling groups and random choices."""
# Use a unique seed for regular choices based on position
random.seed(base_seed + position)
parts = [p.strip() for p in content.split('|')]
options = []
weights = []
group_defined = False
group_name = None
static_group = None
cycling_group = None
for p in content.split('|'):
p = p.strip()
if p.startswith('group='):
group_name = p.split('=', 1)[1].strip()
group_defined = True
continue
parsed = self.parse_option(p)
if isinstance(parsed, list): # CSV data
parts.extend(parsed)
weights.extend([1]*len(parsed))
elif isinstance(parsed, tuple): # Weighted option
parts.append(parsed[0])
weights.append(parsed[1])
for p in parts:
if p.startswith('static_group='):
static_group = p.split('=', 1)[1].strip()
elif p.startswith('group='):
cycling_group = p.split('=', 1)[1].strip()
else:
parts.append(parsed)
weights.append(1)
if group_defined:
return {'type': 'group', 'name': group_name, 'options': parts}
if any(w != 1 for w in weights):
total = sum(weights)
if total == 0: weights = [1]*len(parts)
return random.choices(parts, weights=[w/total for w in weights])[0]
return random.choice(parts) if parts else ''
parsed = self.parse_option(p)
if isinstance(parsed, list): # CSV data
options.extend(parsed)
weights.extend([1] * len(parsed))
elif isinstance(parsed, tuple): # Weighted option
options.append(parsed[0])
weights.append(parsed[1])
else:
options.append(parsed)
weights.append(1)
if static_group and cycling_group:
raise ValueError("Cannot specify both static_group and group in the same section.")
if static_group:
return {'type': 'static_group', 'name': static_group, 'options': options, 'weights': weights}
elif cycling_group:
return {'type': 'cycling_group', 'name': cycling_group, 'options': options, 'weights': weights}
else:
if options:
if any(w != 1 for w in weights):
total = sum(weights)
if total == 0:
weights = [1] * len(options)
return random.choices(options, weights=[w / total for w in weights])[0]
else:
return random.choice(options)
return ''
def process_advanced_syntax(self, text, seed):
# Process nested variables
"""Process the entire text for advanced syntax, handling nested variables and groups."""
variables = self.find_variables(text)
static_groups = {}
cycling_groups = {}
substitutions = []
groups = {}
for var in variables:
start, end = var['start'], var['end']
content = text[start+1:end-1]
processed = self.process_content(content, seed)
content = text[start + 1:end - 1]
processed = self.process_content(content, seed, start)
if isinstance(processed, dict):
if processed['type'] == 'group':
group_name = processed['name']
if group_name not in groups:
groups[group_name] = []
groups[group_name].append({
if processed['type'] == 'static_group':
name = processed['name']
if name not in static_groups:
static_groups[name] = []
static_groups[name].append({
'start': start,
'end': end,
'options': processed['options']
'options': processed['options'],
'weights': processed['weights']
})
elif processed['type'] == 'cycling_group':
name = processed['name']
if name not in cycling_groups:
cycling_groups[name] = []
cycling_groups[name].append({
'start': start,
'end': end,
'options': processed['options'],
'weights': processed['weights']
})
else:
substitutions.append({
@@ -130,15 +150,35 @@ class LineSelector:
'sub': processed
})
# Handle groups
for group_name, matches in groups.items():
# Handle static groups: choose one value per group name
random.seed(seed) # Reset seed for consistent static group behavior
for name, matches in static_groups.items():
if not matches or not matches[0]['options']:
continue
options = matches[0]['options']
weights = matches[0]['weights']
if any(w != 1 for w in weights):
total = sum(weights)
if total == 0:
weights = [1] * len(options)
chosen = random.choices(options, weights=[w / total for w in weights])[0]
else:
chosen = random.choice(options) if options else ''
for m in matches:
substitutions.append({
'start': m['start'],
'end': m['end'],
'sub': chosen
})
# Handle cycling groups: cycle through shuffled options
random.seed(seed) # Reset seed for consistent cycling group behavior
for name, matches in cycling_groups.items():
if not matches or not matches[0]['options']:
continue
options = matches[0]['options']
permuted = random.sample(options, len(options))
perm_cycle = cycle(permuted)
for m in matches:
substitutions.append({
'start': m['start'],
@@ -146,7 +186,7 @@ class LineSelector:
'sub': next(perm_cycle)
})
# Apply regular substitutions
# Apply substitutions in reverse order
substitutions.sort(key=lambda x: -x['start'])
result_text = text
for sub in substitutions:
@@ -155,43 +195,31 @@ class LineSelector:
return result_text
def select_line(self, text, line_number, RANDOM, LOOP, LOOP_SEQUENTIAL, jump, pick_random_variable, variables="", seed=-1):
# Parse variables
"""Select lines from the text based on the specified mode after processing advanced syntax."""
var_dict = {}
for line in variables.split('\n'):
if '=' in line:
key, value = line.split('=', 1)
var_dict[key.strip()] = value.strip()
# Replace variables in the text
for key, value in var_dict.items():
text = text.replace(f"<{key}>", value)
# Split the input text into lines, remove empty lines and lines starting with #
if seed < 0:
seed = random.randint(0, 0x7FFFFFFFFFFFFFFF)
if pick_random_variable:
text = self.process_advanced_syntax(text, seed)
lines = [line.strip() for line in text.split('\n')
if line.strip() and not line.strip().startswith('#')]
if line.strip() and not line.strip().startswith('#')]
if not lines:
return (["No valid lines found."], 0, 0)
import random
import os
# Set seed if provided
if seed < 0:
seed = random.randint(0, 0x7FFFFFFFFFFFFFFF)
# Process WriteTextAdvanced syntax if enabled
if pick_random_variable:
processed_lines = []
for line in lines:
processed_lines.append(self.process_advanced_syntax(line, seed))
lines = processed_lines
# Handle sequential looping
if LOOP_SEQUENTIAL:
counter_file = os.path.join("Bjornulf", "line_selector_counter.txt")
os.makedirs(os.path.dirname(counter_file), exist_ok=True)
try:
with open(counter_file, 'r') as f:
current_index = int(f.read().strip())
@@ -199,11 +227,10 @@ class LineSelector:
current_index = -jump
next_index = current_index + jump
if next_index >= len(lines):
with open(counter_file, 'w') as f:
f.write(str(-jump))
raise ValueError(f"Counter has reached the last line (total lines: {len(lines)}). Counter has be reset.")
raise ValueError(f"Counter has reached the last line (total lines: {len(lines)}). Counter has been reset.")
with open(counter_file, 'w') as f:
f.write(str(next_index))
@@ -211,11 +238,9 @@ class LineSelector:
remaining_cycles = max(0, (len(lines) - next_index - 1) // jump + 1)
return ([lines[next_index]], remaining_cycles, next_index + 1)
# Handle normal LOOP mode
if LOOP:
return (lines, len(lines), 0)
# Handle RANDOM or line_number selection
if RANDOM or line_number == 0:
selected = random.choice(lines)
else:

11
list_selector.py
View File

@@ -42,16 +42,15 @@ class ListSelector:
# Convert to 0-based index
index = selection - 1
# Get the selected number
# Get the selected element
selected = numbers[index]
# Convert to integer and string
# Try to convert to integer, but keep string value regardless
try:
selected_int = int(selected)
selected_str = str(selected_int)
except ValueError:
# If conversion fails, return 0
selected_int = 0
selected_str = "0"
selected_int = 0 # Return 0 for non-numeric values
selected_str = selected # Return original string value
return selected_int, selected_str, list_length

56
load_text.py
View File

@@ -1,56 +1,72 @@
import os
from server import PromptServer
import aiohttp.web as web
class LoadTextFromFolder:
default_dir = "Bjornulf/Text"
@classmethod
def INPUT_TYPES(cls):
"""Define input parameters for the node"""
default_dir = "Bjornulf/Text"
available_files = []
if os.path.exists(default_dir):
available_files = [f for f in os.listdir(default_dir)
if f.lower().endswith('.txt')]
available_files = cls.get_available_files()
if not available_files:
available_files = ["no_files_found"]
return {
"required": {
"text_file": (available_files, {"default": available_files[0]}),
}
}
@classmethod
def get_available_files(cls):
"""Get list of .txt files recursively from the default directory"""
available_files = []
if os.path.exists(cls.default_dir):
for root, dirs, files in os.walk(cls.default_dir):
for file in files:
if file.lower().endswith('.txt'):
full_path = os.path.join(root, file)
rel_path = os.path.relpath(full_path, cls.default_dir)
available_files.append(rel_path)
available_files.sort()
return available_files
RETURN_TYPES = ("STRING", "STRING", "STRING")
RETURN_NAMES = ("text", "filename", "full_path")
FUNCTION = "load_text"
CATEGORY = "Bjornulf"
def load_text(self, text_file):
"""Load text from the selected file"""
try:
if text_file == "no_files_found":
raise ValueError("No text files found in Bjornulf/Text folder")
filepath = os.path.join("Bjornulf/Text", text_file)
# Check if file exists
filepath = os.path.join(self.default_dir, text_file)
if not os.path.exists(filepath):
raise ValueError(f"File not found: {filepath}")
# Get absolute path
full_path = os.path.abspath(filepath)
# Get just the filename
filename = os.path.basename(filepath)
# Read text from file
with open(filepath, 'r', encoding='utf-8') as file:
text = file.read()
return (text, filename, full_path)
except (OSError, IOError) as e:
raise ValueError(f"Error loading file: {str(e)}")
@PromptServer.instance.routes.post("/get_text_files")
async def get_text_files(request):
try:
available_files = LoadTextFromFolder.get_available_files()
return web.json_response({
"success": True,
"files": available_files
}, status=200)
except Exception as e:
error_msg = str(e)
return web.json_response({
"success": False,
"error": error_msg
}, status=500)
class LoadTextFromPath:
@classmethod
def INPUT_TYPES(cls):

49
loader_lora_with_path.py
View File

@@ -8,42 +8,43 @@ class LoaderLoraWithPath:
return {
"required": {
"model": ("MODEL",),
"clip": ("CLIP",),
"lora_path": ("STRING", {"default": ""}),
"strength_model": ("FLOAT", {"default": 1.0, "min": -100.0, "max": 100.0, "step": 0.01}),
"strength_clip": ("FLOAT", {"default": 1.0, "min": -100.0, "max": 100.0, "step": 0.01}),
},
"optional": {
"clip": ("CLIP",),
}
}
RETURN_TYPES = ("MODEL", "CLIP")
FUNCTION = "load_lora" # Added this line
RETURN_TYPES = ("MODEL", "CLIP", "STRING")
FUNCTION = "load_lora"
CATEGORY = "Bjornulf"
def load_lora(self, model, clip, lora_path, strength_model, strength_clip):
def load_lora(self, model, lora_path, strength_model, strength_clip, clip=None):
try:
# Check if path exists
if not os.path.isfile(lora_path):
print(f"Error: Lora file not found at path: {lora_path}")
return (model, clip)
return (model, clip if clip is not None else None, lora_path if lora_path is not None else None)
# Load the Lora file
try:
lora = comfy.utils.load_torch_file(lora_path)
except Exception as e:
print(f"Error loading Lora file: {str(e)}")
return (model, clip)
lora = comfy.utils.load_torch_file(lora_path)
if clip is not None:
model_lora, clip_lora = comfy.sd.load_lora_for_models(
model, clip, lora, strength_model, strength_clip
)
return (model_lora, clip_lora, lora_path if lora_path is not None else None)
else:
model_lora = model.clone()
# Assuming ModelPatcher with diffusion_model
state_dict = model_lora.model.diffusion_model.state_dict()
for key in lora:
if 'unet' in key: # Filter for UNet keys; adjust as needed
if key in state_dict:
state_dict[key] += strength_model * lora[key]
model_lora.model.diffusion_model.load_state_dict(state_dict)
return (model_lora, None, lora_path if lora_path is not None else None)
# Apply the Lora
model_lora, clip_lora = comfy.sd.load_lora_for_models(
model,
clip,
lora,
strength_model,
strength_clip
)
return (model_lora, clip_lora)
except Exception as e:
print(f"Error in load_lora: {str(e)}")
return (model, clip)
return (model, clip if clip is not None else None)

62
math_node.py Normal file
View File

@@ -0,0 +1,62 @@
class MathNode:
@classmethod
def INPUT_TYPES(cls):
return {
"required": {
"operation": (["+", "-", "*", "/", "%"], {"default": "+"}),
"num_inputs": ("INT", {"default": 2, "min": 2, "max": 100, "step": 1}),
},
"hidden": {
**{f"value_{i}": ("*", {"forceInput": True}) for i in range(1, 101)}
}
}
RETURN_TYPES = ("FLOAT", "INT", "STRING")
FUNCTION = "compute"
CATEGORY = "Utilities/Math"
def compute(self, operation, num_inputs, **kwargs):
# Collect and convert inputs to float, defaulting to 0.0 if conversion fails or input is missing
values = []
for i in range(1, num_inputs + 1):
key = f"value_{i}"
raw_value = kwargs.get(key, 0.0)
try:
# Attempt to convert any input to float
value = float(raw_value)
except (ValueError, TypeError):
value = 0.0 # Fallback to 0.0 if conversion fails
values.append(value)
# If no valid values, return 0.0, 0, "0.0"
if not values:
return (0.0, 0, "0.0")
# Initialize result with the first value
result = values[0]
# Apply the selected operation cumulatively from left to right
for val in values[1:]:
if operation == "+":
result += val
elif operation == "-":
result -= val
elif operation == "*":
result *= val
elif operation == "/":
if val == 0:
result = 0.0 # Handle division by zero
else:
result /= val
elif operation == "%":
if val == 0:
result = 0.0 # Handle modulo by zero
else:
result %= val
# Create integer and string versions of the result
int_result = int(result) # Truncate to integer
str_result = str(result) # Convert to string
# Return the three outputs
return (result, int_result, str_result)

224
note_image.py
View File

@@ -9,27 +9,22 @@ import folder_paths
import node_helpers
from aiohttp import web
class ImageNote(SaveImage):
class ImageNote:
def __init__(self):
self.output_dir = folder_paths.get_temp_directory()
self.type = "temp"
self.prefix_append = "_temp_" + ''.join(random.choice("abcdefghijklmnopqrstupvxyz") for _ in range(5))
self.compress_level = 1
# Directory to store notes, created if it doesn’t exist
self.note_dir = os.path.join("ComfyUI", "Bjornulf", "imageNote")
os.makedirs(self.note_dir, exist_ok=True)
# Store last image path and hash to prevent unnecessary reloading
self.last_image_path = None
self.last_image_hash = None
self.last_output_images = None
@classmethod
def INPUT_TYPES(cls):
"""Define the input types for the node."""
return {
"optional": {
"images": ("IMAGE", ),
"image_path": ("STRING", {"default": ""}),
"note_text": ("STRING", {"default": "", "multiline": True})
"note": ("STRING", {"default": ""}),
"note_2": ("STRING", {"default": ""}),
"note_3": ("STRING", {"default": ""})
},
"hidden": {
"prompt": "PROMPT",
@@ -37,136 +32,164 @@ class ImageNote(SaveImage):
},
}
RETURN_TYPES = ("STRING", "STRING")
RETURN_NAMES = ("image_path", "note")
FUNCTION = "process_image"
OUTPUT_NODE = True
CATEGORY = "Bjornulf"
def compute_md5(self, image):
image_bytes = image.tobytes() if isinstance(image, Image.Image) else image
"""Compute MD5 hash of an image for note association."""
if isinstance(image, Image.Image):
image_bytes = image.tobytes()
elif isinstance(image, torch.Tensor):
image_bytes = (image.numpy() * 255).astype(np.uint8).tobytes()
else:
image_bytes = image
return hashlib.md5(image_bytes).hexdigest()
def process_image(self, images=None, image_path="", note_text="", prompt=None, extra_pnginfo=None):
output_images = None
output_note_text = ""
def process_image(self, images=None, image_path="", note="", note_2="", note_3="", prompt=None, extra_pnginfo=None):
"""Process the image and associate all notes."""
output_note = ""
ui_images = []
# If images are given, process them
if images is not None and len(images) > 0:
output_images = images
input_dir = folder_paths.get_input_directory()
output_dir = folder_paths.get_output_directory()
temp_dir = folder_paths.get_temp_directory()
# Collect all non-empty notes
all_notes = [n for n in [note, note_2, note_3] if n]
# Case 1: Image provided via file path
if image_path and os.path.isfile(image_path):
image = Image.open(image_path).convert("RGB")
image_hash = self.compute_md5(image)
# Determine image reference for UI
if image_path.startswith(input_dir):
type_ = "input"
filename = os.path.relpath(image_path, input_dir)
elif image_path.startswith(output_dir):
type_ = "output"
filename = os.path.relpath(image_path, output_dir)
else:
temp_filename = f"{image_hash}.png"
temp_path = os.path.join(temp_dir, temp_filename)
if not os.path.exists(temp_path):
image.save(temp_path)
type_ = "temp"
filename = temp_filename
# Handle notes: append new notes and read all
note_path = os.path.join(self.note_dir, f"{image_hash}.txt")
if all_notes:
with open(note_path, "a", encoding="utf-8") as f:
for n in all_notes:
f.write(n + "\n")
if os.path.exists(note_path):
with open(note_path, "r", encoding="utf-8") as f:
output_note = f.read().rstrip() # Remove trailing newline
ui_images = [{"filename": filename, "subfolder": "", "type": type_}]
result = (image_path, output_note)
# Case 2: Image provided as tensor
elif images is not None and len(images) > 0:
image_np = (images[0].numpy() * 255).astype(np.uint8)
image = Image.fromarray(image_np)
image_hash = self.compute_md5(image)
temp_filename = f"{image_hash}.png"
temp_path = os.path.join(temp_dir, temp_filename)
if not os.path.exists(temp_path):
image.save(temp_path)
# Handle notes: append new notes and read all
note_path = os.path.join(self.note_dir, f"{image_hash}.txt")
if all_notes:
with open(note_path, "a", encoding="utf-8") as f:
for n in all_notes:
f.write(n + "\n")
if os.path.exists(note_path):
with open(note_path, "r", encoding="utf-8") as f:
output_note_text = f.read()
elif note_text:
with open(note_path, "w", encoding="utf-8") as f:
f.write(note_text)
output_note_text = note_text
output_note = f.read().rstrip()
# If image_path is empty, do nothing
elif not image_path:
# logger.debug("No image path provided, skipping processing.")
return None, ""
ui_images = [{"filename": temp_filename, "subfolder": "", "type": "temp"}]
result = (temp_path, output_note)
# Process image from path only if it has changed
elif os.path.isfile(image_path):
if image_path == self.last_image_path:
# logger.debug("Image path has not changed, skipping reload.")
return super().save_images(images=self.last_output_images, prompt=prompt, extra_pnginfo=extra_pnginfo)
# Case 3: No image provided
else:
result = ("", "")
image = Image.open(image_path).convert("RGB")
image_hash = self.compute_md5(image)
return {"ui": {"images": ui_images}, "result": result}
if image_hash == self.last_image_hash:
# logger.debug("Image content has not changed, skipping reload.")
return super().save_images(images=self.last_output_images, prompt=prompt, extra_pnginfo=extra_pnginfo)
note_path = os.path.join(self.note_dir, f"{image_hash}.txt")
if os.path.exists(note_path):
with open(note_path, "r", encoding="utf-8") as f:
output_note_text = f.read()
elif note_text:
with open(note_path, "w", encoding="utf-8") as f:
f.write(note_text)
output_note_text = note_text
image_np = np.array(image).astype(np.float32) / 255.0
output_images = torch.from_numpy(image_np).unsqueeze(0)
# Update stored values
self.last_image_path = image_path
self.last_image_hash = image_hash
self.last_output_images = output_images
return super().save_images(images=output_images, prompt=prompt, extra_pnginfo=extra_pnginfo)
class ImageNoteLoadImage:
def __init__(self):
self.note_dir = os.path.join("ComfyUI", "Bjornulf", "imageNote")
os.makedirs(self.note_dir, exist_ok=True)
@classmethod
def INPUT_TYPES(s):
base_input_dir = folder_paths.get_input_directory() # Get base input directory
input_dir = os.path.join(base_input_dir, "Bjornulf", "imagenote_images") # Specify subdirectory
# Create the directory if it doesn't exist
base_input_dir = folder_paths.get_input_directory()
input_dir = os.path.join(base_input_dir, "Bjornulf", "imagenote_images")
if not os.path.exists(input_dir):
os.makedirs(input_dir, exist_ok=True) # Create directory and parents if needed
# Filter for image files only
os.makedirs(input_dir, exist_ok=True)
valid_extensions = ('.png', '.jpg', '.jpeg', '.gif', '.bmp', '.webp')
files = [f for f in os.listdir(input_dir) if
os.path.isfile(os.path.join(input_dir, f)) and
f.lower().endswith(valid_extensions)]
os.path.isfile(os.path.join(input_dir, f)) and
f.lower().endswith(valid_extensions)]
if not files:
# Provide a default option if no files are found
files = ["none"]
return {
"required": {
"image": (sorted(files), {"image_upload": True}),
"note": ("STRING", {"default": ""}),
"note_2": ("STRING", {"default": ""}),
"note_3": ("STRING", {"default": ""})
}
}
return {"required":
{
"image": (sorted(files), {"image_upload": True}),
# "note": ("STRING", {"default": ""}), # Added multiline option FAILURE
"note": ("STRING", {"multiline": True, "lines": 10})
}
}
RETURN_TYPES = ("IMAGE", "MASK", "STRING", "STRING") # Added note to return types
RETURN_NAMES = ("image", "mask", "image_path", "note") # Added note to return names
RETURN_TYPES = ("IMAGE", "MASK", "STRING", "STRING")
RETURN_NAMES = ("image", "mask", "image_path", "note")
FUNCTION = "load_image_alpha"
CATEGORY = "Bjornulf"
def load_image_alpha(self, image, note): # Added note parameter
def compute_md5(self, image):
"""Compute MD5 hash of an image for note association."""
if isinstance(image, Image.Image):
image_bytes = image.tobytes()
elif isinstance(image, torch.Tensor):
image_bytes = (image.numpy() * 255).astype(np.uint8).tobytes()
else:
image_bytes = image
return hashlib.md5(image_bytes).hexdigest()
def load_image_alpha(self, image, note="", note_2="", note_3=""):
image_path = folder_paths.get_annotated_filepath(image)
img = node_helpers.pillow(Image.open, image_path)
output_images = []
output_masks = []
w, h = None, None
excluded_formats = ['MPO']
for i in ImageSequence.Iterator(img):
i = node_helpers.pillow(ImageOps.exif_transpose, i)
if i.mode == 'I':
i = i.point(lambda i: i * (1 / 255))
image_converted = i.convert("RGBA") # Renamed to avoid shadowing
image_converted = i.convert("RGBA")
if len(output_images) == 0:
w = image_converted.size[0]
h = image_converted.size[1]
if image_converted.size[0] != w or image_converted.size[1] != h:
continue
image_np = np.array(image_converted).astype(np.float32) / 255.0 # Renamed to avoid shadowing
image_tensor = torch.from_numpy(image_np)[None,] # Renamed to avoid shadowing
image_np = np.array(image_converted).astype(np.float32) / 255.0
image_tensor = torch.from_numpy(image_np)[None,]
if 'A' in i.getbands():
mask = np.array(i.getchannel('A')).astype(np.float32) / 255.0
mask = 1. - torch.from_numpy(mask)
else:
mask = torch.zeros((64, 64), dtype=torch.float32, device="cpu")
output_images.append(image_tensor) # Renamed to avoid shadowing
output_images.append(image_tensor)
output_masks.append(mask.unsqueeze(0))
if len(output_images) > 1 and img.format not in excluded_formats:
@@ -176,19 +199,34 @@ class ImageNoteLoadImage:
output_image = output_images[0]
output_mask = output_masks[0]
return (output_image, output_mask, image_path, note) # Added note to return tuple
# Compute hash from the first image
first_image = output_image[0] if output_image.dim() == 4 else output_image
image_hash = self.compute_md5(first_image)
# Handle notes: append new notes and read all
all_notes = [n for n in [note, note_2, note_3] if n]
note_path = os.path.join(self.note_dir, f"{image_hash}.txt")
if all_notes:
with open(note_path, "a", encoding="utf-8") as f:
for n in all_notes:
f.write(n + "\n")
output_note = ""
if os.path.exists(note_path):
with open(note_path, "r", encoding="utf-8") as f:
output_note = f.read().rstrip()
return (output_image, output_mask, image_path, output_note)
@classmethod
def IS_CHANGED(s, image, note): # Added note to IS_CHANGED
def IS_CHANGED(s, image, note, note_2, note_3):
image_path = folder_paths.get_annotated_filepath(image)
m = hashlib.sha256()
with open(image_path, 'rb') as f:
m.update(f.read())
return m.digest().hex() + str(note) # Include note in hash
return m.digest().hex() + str(note) + str(note_2) + str(note_3)
@classmethod
def VALIDATE_INPUTS(s, image):
if not folder_paths.exists_annotated_filepath(image):
return "Invalid image file: {}".format(image)
return True

5
note_text.py
View File

@@ -22,4 +22,7 @@ class DisplayNote:
def display_text_pass(self, any, display_text):
# Simply pass through the input
return (any,)
if any is None:
return (None,)
else:
return (any,)

172
ollama_image_vision.py
View File

@@ -1,16 +1,7 @@
import torch
import base64
from PIL import Image
import numpy as np
from PIL import Image
from io import BytesIO
import requests
import json
import ollama
from ollama import Client
# import logging
import hashlib
from typing import Dict, Any
from PIL.PngImagePlugin import PngInfo
class OllamaImageVision:
@classmethod
@@ -18,46 +9,34 @@ class OllamaImageVision:
return {
"required": {
"IMAGE": ("IMAGE",),
"output_selection": ("INT", {"default": 7, "min": 1, "max": 8,
"step": 1, "display": "slider", "label": "Number of outputs (1-8)"}),
"process_below_output_selection": ("BOOLEAN", {"default": False, "label": "Process all up to selection"})
"OLLAMA_VISION_PROMPT": ("STRING", {"forceInput": True}),
"vram_retention_minutes": ("INT", {
"default": 0,
"min": 0,
"max": 99
}),
"seed": ("INT", {
"default": -1,
"min": -1,
"max": 2147483647
}),
"answer_single_line": ("BOOLEAN", {"default": False}),
},
"optional": {
"OLLAMA_CONFIG": ("OLLAMA_CONFIG", {"forceInput": True}),
"context": ("STRING", {"multiline": True}),
}
}
RETURN_TYPES = ("STRING", "STRING", "STRING", "STRING", "STRING", "STRING", "STRING", "STRING",)
RETURN_NAMES = ("1 - Basic Description", "2 - Advanced Description", "3 - Characters Description", "4 - Object Recognition", "5 - Semantic Understanding", "6 - Contextual Analysis", "7 - SDXL Prompt (words)", "8 - FLUX Prompt (sentences)")
FUNCTION = "process_image_base64"
RETURN_TYPES = ("STRING",)
RETURN_NAMES = ("response",)
FUNCTION = "process_image"
CATEGORY = "Bjornulf"
def __init__(self):
self.client = None
def create_prompts(self):
# return {
# "basic": "Describe the image in one sentence.",
# "advanced": "Describe the image in details.",
# "characters": "Describe the physical appearance of the character in vivid detail.",
# "objects": "List the key objects and elements visible in the image.",
# "semantic": "Provide an in-depth interpretation including mood, environment, and purpose in the image.",
# "context": "Describe the relationships and context between objects and people in the image."
# }
return {
# Do not describe what isn't there.
"basic": "Summarize the main content of the image in one concise sentence.",
"advanced": "Describe the scene thoroughly, capturing intricate details, colors, textures, and any significant actions or events occurring in the image.",
"characters": "Describe each character's physical appearance in vivid, descriptive terms, including clothing, expressions, body language, and notable features.",
"objects": "Identify and describe the primary objects in the image, detailing their size, position, color, and any unique characteristics.",
"semantic": "Analyze the image's mood, environment, and implied meaning. Discuss any symbolic elements, artistic style, and possible intent or story conveyed.",
"context": "Describe the relationships and interactions between objects and characters, focusing on spatial arrangement, implied actions, and any contextual clues suggesting connections or purpose.",
"SDXL": "Describe the image. The goal is to generate a concise, detailed, and effective description. Guidelines for describing the image:- Focus on visual elements, be specific about objects, colors, textures, and compositions. Use adjectives to describe key features. Avoid complete sentences or narrative descriptions. Prioritize important elements over minor details. Your input will be a detailed description of an image. Process this description and refine it into a prompt suitable for stable diffusion models using the following steps: 1. Identify the most important visual elements and characteristics. 2. Condense the description into a series of comma-separated phrases or words. 3. Prioritize specific, descriptive terms over general ones. Here are two examples of good outputs: Example 1:vibrant sunset, tropical beach, silhouetted palm trees, calm ocean, orange and purple sky, wispy clouds, golden sand, gentle waves, beachgoers in distance, serene atmosphere, warm lighting, panoramic view. Example 2: steampunk cityscape, towering clockwork structures, brass and copper tones, billowing steam, airships in sky, cobblestone streets, Victorian-era citizens, gears and pipes visible, warm sepia lighting, hazy atmosphere, intricate mechanical details. Your final output should be a single line of text containing the refined prompt, without any additional explanation or commentary. IMPORTANT : DO NOT Include information about the overall style or artistic technique.",
"FLUX": "Describe the given image in a detailed and structured format that is specifically designed for image generation. Use descriptive language to capture the essence of the image, including the environment, objects, characters, lighting, textures, and any other notable elements. The description must use some of these 9 points : 1. Scene Type: [Outdoor/Indoor/Abstract/Fantasy/Realistic/etc.] 2. Primary Subject: [Main focus or characters in the scene.] 3. Environment Details: [Describe the setting in vivid detail, including any landscapes, architecture, or surroundings.] 4. Lighting: [Specify the type, color, and intensity of the lighting.] 5. Colors and Tones: [Dominant colors and overall mood.] 6. Perspective: [Camera angle or viewpoint—close-up, wide shot, aerial, etc.] 7. Texture and Details: [Surface materials, patterns, and fine details.] 8. Emotion or Atmosphere: [Mood conveyed by the scene—serene, ominous, lively, etc.] 9. Unique Elements: [Special features or focal points that make the image distinctive.] For example: 1. Scene Type: Outdoor, natural landscape. 2. Primary Subject: A majestic lion standing atop a rocky hill. 3. Environment Details: A vast savannah with tall golden grass, sparse acacia trees, and distant mountains under a clear blue sky. 4. Lighting: Bright, warm sunlight casting long shadows. 5. Colors and Tones: Predominantly gold and blue, with subtle earthy browns and greens. 6. Perspective: Mid-range shot, slightly low angle to emphasize the lion's dominance. 7. Texture and Details: The lion's fur appears detailed with visible strands, and the rocks have a rough, weathered texture. 8. Emotion or Atmosphere: Majestic, powerful, and serene. 9. Unique Elements: A subtle wind effect in the grass and mane, adding movement to the scene. IMPORTANT : DO NOT Include information about the overall style or artistic technique."
}
def process_image_base64(self, IMAGE, OLLAMA_CONFIG=None, output_selection=6, process_below_output_selection=True):
def process_image(self, IMAGE, OLLAMA_VISION_PROMPT, answer_single_line, vram_retention_minutes, seed, OLLAMA_CONFIG=None, context=None):
from ollama import Client
# Default OLLAMA_CONFIG if not provided
if OLLAMA_CONFIG is None:
OLLAMA_CONFIG = {
"model": "moondream",
@@ -65,65 +44,78 @@ class OllamaImageVision:
}
selected_model = OLLAMA_CONFIG["model"]
ollama_url = OLLAMA_CONFIG["url"]
# Convert images to base64
images_base64 = []
for img in IMAGE:
# Convert tensor to numpy array
numpy_img = (255. * img.cpu().numpy()).clip(0, 255).astype(np.uint8)
# Create PIL Image
pil_image = Image.fromarray(numpy_img)
# Create a BytesIO object
buffered = BytesIO()
# Save the image into the BytesIO object
pil_image.save(buffered, format="PNG")
# Get the byte value and encode to base64
img_str = base64.b64encode(buffered.getvalue()).decode('utf-8')
images_base64.append(img_str)
# Clean up
buffered.close()
# Initialize client
client = Client(host=ollama_url)
# Get prompts
prompts = list(self.create_prompts().items())
# Process outputs based on selection and process_below_output_selection flag
responses = []
for i in range(8): # Always prepare 5 slots for output
if process_below_output_selection:
# Process all outputs up to output_selection
if i < output_selection:
prompt_type, prompt = prompts[i]
response = client.generate(
model=selected_model,
prompt=prompt,
images=images_base64
)
responses.append(response['response'].strip())
else:
responses.append("")
else:
# Process only the selected output (output_selection - 1)
if i == (output_selection - 1):
prompt_type, prompt = prompts[i]
response = client.generate(
model=selected_model,
prompt=prompt,
images=images_base64
)
responses.append(response['response'].strip())
else:
responses.append("")
return tuple(responses)
# Construct the final prompt
if context:
final_prompt = context + "\n" + OLLAMA_VISION_PROMPT
else:
final_prompt = OLLAMA_VISION_PROMPT
# Generate response with the final prompt
response = client.generate(
model=selected_model,
prompt=final_prompt,
images=images_base64,
keep_alive=f"{vram_retention_minutes}m"
)
if answer_single_line:
response['response'] = ' '.join(response['response'].split())
return (response['response'].strip(),)
def handle_error(self, error_message: str) -> tuple:
"""Handle errors by returning appropriate error messages for all outputs"""
error_response = f"Error: {error_message}"
return tuple([error_response] * 4)
class OllamaVisionPromptSelector: #Prompts made for gemma3
@classmethod
def INPUT_TYPES(cls):
return {
"required": {
"prompt_type": (["NONE", "basic", "advanced", "characters", "objects", "semantic", "basic_action", "advanced_action", "context", "SDXL", "FLUX", "video"],),
},
"optional": {
"prefix_custom_prompt": ("STRING", {"multiline": True, "default": ""}),
"suffix_custom_prompt": ("STRING", {"multiline": True, "default": "Do not include any introductory text or explanations, make it a clean one line answer."}),
}
}
RETURN_TYPES = ("STRING",)
RETURN_NAMES = ("OLLAMA_VISION_PROMPT",)
FUNCTION = "select_prompt"
CATEGORY = "Bjornulf"
def get_prompts(self):
"""Returns a dictionary of predefined prompts for each prompt_type."""
return {
"basic": "Summarize the main content of the image in one concise sentence.",
"advanced": "Describe the scene thoroughly, capturing intricate details, colors, textures, and significant actions.",
"characters": "Describe each character's physical appearance, including clothing, expressions, and notable features.",
"objects": "Identify and describe the primary objects, detailing their size, position, color, and characteristics.",
"semantic": "Analyze the image's mood, environment, and implied meaning, focusing on symbolic elements and atmosphere.",
"context": "Describe relationships and interactions between objects and characters, focusing on spatial arrangement and actions.",
"basic_action": "Describe the action that is happening in the image.",
"advanced_action": "Describe the action that is happening in the image in details.",
"SDXL": "Generate a concise, comma-separated list of visual elements from the image, suitable for Stable Diffusion XL. Focus on objects, colors, textures, and composition. Use adjectives for key features. Output only the final prompt. Example: vibrant sunset, tropical beach, silhouetted palm trees, calm ocean, orange sky.",
"FLUX": "Generate a detailed, structured description of the image for FLUX, including scene type, primary subject, environment details, lighting, colors, perspective, textures, atmosphere, and unique elements. Use concise descriptive phrases. Output only the final prompt. Example: A serene, moonlit forest clearing with a glowing, ethereal portal, surrounded by ancient, towering trees, casting long shadows in the silver light.",
"video": "Analyze the image to identify the characters. Then, imagine and describe in a single sentence a random dynamic actions they might perform next, involving movement or interaction, based on their appearance and the scene's context. Example : The woman is smirking seductively while staring at the camera, then suddenly winks. Do not add new characters that are not in the image."
}
def select_prompt(self, prompt_type, prefix_custom_prompt="", suffix_custom_prompt=""):
if prompt_type == "NONE":
selected_prompt = prefix_custom_prompt + suffix_custom_prompt
else:
prompts = self.get_prompts()
selected_prompt = prefix_custom_prompt + prompts.get(prompt_type, "") + suffix_custom_prompt
return (selected_prompt,) # Return as a tuple for ComfyUI compatibility

4
ollama_talk.py
View File

@@ -1,7 +1,5 @@
import requests
import json
import ollama
from ollama import Client
import logging
import hashlib
from typing import Dict, Any
@@ -47,6 +45,8 @@ class OllamaTalk:
current_instance = None
def __init__(self):
import ollama
from ollama import Client
self.last_content_hash = None
self.waiting = False
self.OLLAMA_CONFIG = None

2
pause_resume_stop.py
View File

@@ -8,6 +8,7 @@ import os
import io
import sys
import random
# from comfy_execution.graph import ExecutionBlocker
class Everything(str):
def __ne__(self, __value: object) -> bool:
@@ -68,6 +69,7 @@ class PauseResume:
PauseResume.should_stop = False # Reset for next run
PauseResume.is_paused = True
raise Exception("Workflow stopped by user")
# return (ExecutionBlocker("Workflow stopped by user"),) # Return ExecutionBlocker to stop gracefully, but error on next node.
PauseResume.is_paused = True
PauseResume.should_stop = False

4
pyproject.toml
View File

@@ -1,7 +1,7 @@
[project]
name = "bjornulf_custom_nodes"
description = "133 ComfyUI nodes : Display, manipulate, and edit text, images, videos, loras, generate characters and more. Manage looping operations, generate randomized content, use logical conditions and work with external AI tools, like Ollama or Text To Speech Kokoro, etc..."
version = "0.76"
description = "162 ComfyUI nodes : Display, manipulate, and edit text, images, videos, loras, generate characters and more. Manage looping operations, generate randomized content, use logical conditions and work with external AI tools, like Ollama or Text To Speech, etc..."
version = "0.77"
license = {file = "LICENSE"}
[project.urls]

7
requirements.txt
View File

@@ -1,10 +1,15 @@
ollama
pydub
opencv-python
faster_whisper
# Remove faster-whisper for the people using python 3.13 (old pkgutil)
# faster_whisper
ffmpeg-python
civitai-py
fal_client
sounddevice
#24, remove kokoro install by default (need to do that manually to use the kokoro node)
# kokoro_onnx
langdetect
spacy
textblob
dill

15
save_img_to_folder.py
View File

@@ -24,9 +24,12 @@ class SaveImageToFolder(SaveImage):
# Call the parent's save_images with filename_prefix set to "folder_name/"
# This will make the parent class save to the custom folder
return super().save_images(
images=images,
filename_prefix=f"{folder_name}/_",
prompt=prompt,
extra_pnginfo=extra_pnginfo
)
if images is None:
return (None,)
else:
return super().save_images(
images=images,
filename_prefix=f"{folder_name}/_",
prompt=prompt,
extra_pnginfo=extra_pnginfo
)

80
save_tmp_audio.py Normal file
View File

@@ -0,0 +1,80 @@
import os
import shutil
import numpy as np
import soundfile as sf
import torch
class SaveTmpAudio:
@classmethod
def INPUT_TYPES(cls):
return {
"optional": {
"audio": ("AUDIO",),
"audio_path": ("STRING",),
},
}
FUNCTION = "save_audio"
RETURN_TYPES = ()
OUTPUT_NODE = True
CATEGORY = "Bjornulf"
def save_audio(self, audio=None, audio_path=None):
# Ensure the output directory exists
os.makedirs("./output", exist_ok=True)
# Check if neither input is provided
if audio is None and audio_path is None:
raise ValueError("Either 'audio' or 'audio_path' must be provided")
# Case 1: Handle AUDIO input if provided
if audio is not None:
# Validate that audio is a dictionary with required keys
if not isinstance(audio, dict):
raise ValueError("AUDIO input must be a dictionary with 'waveform' and 'sample_rate'")
if 'waveform' not in audio or 'sample_rate' not in audio:
raise ValueError("AUDIO dictionary must contain 'waveform' and 'sample_rate' keys")
# Extract waveform and sample rate
waveform = audio['waveform']
sample_rate = audio['sample_rate']
# Ensure waveform is a PyTorch tensor
if not isinstance(waveform, torch.Tensor):
raise TypeError(f"Waveform must be a PyTorch tensor, got {type(waveform)}")
# Squeeze waveform to remove extra dimensions (e.g., from (1, 1, N) to (N,))
waveform = waveform.squeeze()
# Convert to NumPy array
audio_np = waveform.cpu().numpy()
# Validate the shape after squeezing
if audio_np.ndim > 2:
raise ValueError(f"Audio data has too many dimensions after squeezing: {audio_np.shape}")
# Scale floating-point data to 16-bit integers (assuming range [-1, 1])
if audio_np.dtype in (np.float32, np.float64):
audio_np = (audio_np * 32767).astype(np.int16)
# Save as WAV file
filename = "./output/tmp_api.wav"
sf.write(filename, audio_np, sample_rate)
# Case 2: Handle audio_path input if audio is not provided
elif audio_path is not None:
# Verify the file exists
if not os.path.exists(audio_path):
raise FileNotFoundError(f"Audio file not found: {audio_path}")
# Check for supported file extensions
ext = os.path.splitext(audio_path)[1].lower()
if ext not in ('.wav', '.mp3'):
raise ValueError("audio_path must be a .wav or .mp3 file")
# Copy the file to the output directory
filename = f"./output/tmp_api{ext}"
shutil.copy(audio_path, filename)
# Return UI information for ComfyUI
return {"ui": {"audio": [{"filename": filename, "type": "output"}]}}

1
save_tmp_image.py
View File

@@ -3,6 +3,7 @@ import numpy as np
from PIL import Image
import json
from PIL.PngImagePlugin import PngInfo
import torch
class SaveTmpImage:
@classmethod

34
save_tmp_video.py Normal file
View File

@@ -0,0 +1,34 @@
import os
import shutil
class SaveTmpVideo:
@classmethod
def INPUT_TYPES(cls):
return {
"required": {
"video_path": ("STRING", {"forceInput": True}),
},
}
FUNCTION = "save_video"
RETURN_TYPES = ()
OUTPUT_NODE = True
CATEGORY = "Bjornulf"
def save_video(self, video_path):
# Ensure the output directory exists
os.makedirs("./output", exist_ok=True)
# Verify the video file exists
if not os.path.exists(video_path):
raise FileNotFoundError(f"Video file not found: {video_path}")
# Get the file extension
ext = os.path.splitext(video_path)[1].lower()
if ext not in ('.mp4', '.mkv', '.webm'):
raise ValueError("video_path must be a .mp4, .mkv, or .webm file")
# Copy the file to the output directory with the same extension
filename = f"./output/tmp_api{ext}"
shutil.copy(video_path, filename)
print(f"Temporary video saved as: {filename}")
return {"ui": {"videos": [{"filename": filename, "type": "output"}]}}

BIN
screenshots/audio_preview.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 52 KiB

BIN
screenshots/basic_math.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 82 KiB

BIN
screenshots/execute_worflow.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 290 KiB

BIN
screenshots/fix_face_1.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 331 KiB

BIN
screenshots/fix_face_2_zoom.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 317 KiB

BIN
screenshots/fix_face_3.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 399 KiB

BIN
screenshots/fix_face_4_ignore.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 389 KiB

BIN
screenshots/fix_face_5_person.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 444 KiB

BIN
screenshots/fix_face_6_preview.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 149 KiB

BIN
screenshots/grid_assemble_1.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 446 KiB

BIN
screenshots/grid_assemble_2.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 453 KiB

BIN
screenshots/load_civit.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 170 KiB

BIN
screenshots/ollama_vision.png
View File

Binary file not shown.
Side by Side Swipe Overlay

Before

Width:  |  Height:  |  Size: 134 KiB

After

Width:  |  Height:  |  Size: 174 KiB

BIN
screenshots/random_int.png
View File

Binary file not shown.
Side by Side Swipe Overlay

Before

Width:  |  Height:  |  Size: 50 KiB

After

Width:  |  Height:  |  Size: 56 KiB

BIN
screenshots/remote_vae.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 25 KiB

BIN
screenshots/remote_vae_tiled.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 25 KiB

BIN
screenshots/savetmp_audio1.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 17 KiB

BIN
screenshots/savetmp_audio2.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 24 KiB

BIN
screenshots/savetmp_video.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 17 KiB

BIN
screenshots/split_grid.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 318 KiB

BIN
screenshots/style_gen_1.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 208 KiB

BIN
screenshots/style_gen_2.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 263 KiB

BIN
screenshots/style_gen_3.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 245 KiB

BIN
screenshots/switch_anything.png
View File

Binary file not shown.
Side by Side Swipe Overlay

Before

Width:  |  Height:  |  Size: 272 KiB

After

Width:  |  Height:  |  Size: 290 KiB

BIN
screenshots/text_ana_1.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 64 KiB

BIN
screenshots/text_ana_2.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 91 KiB

BIN
screenshots/video_text_generator.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 32 KiB

4
show_text.py
View File

@@ -1,7 +1,6 @@
class Everything(str):
def __ne__(self, _):
return False
class ShowText:
@classmethod
def INPUT_TYPES(cls):
@@ -21,7 +20,6 @@ class ShowText:
if isinstance(value, int):
return 'integer'
elif isinstance(value, float):
# Check if it has a decimal part
if value % 1 == 0:
return 'float' if str(value).endswith('.0') else 'integer'
return 'float'
@@ -39,5 +37,5 @@ class ShowText:
return 'other type'
def show_text(self, text_int_float):
type_info = [f"{value}" for value in text_int_float]
type_info = [f"{value}" for value in text_int_float if value is not None]
return {"ui": {"text": type_info}}

16
speech_to_text.py
View File

@@ -5,13 +5,8 @@ import numpy as np
import tempfile
import wave
import subprocess # Added for ffmpeg
try:
import faster_whisper
WHISPER_AVAILABLE = True
except ImportError:
WHISPER_AVAILABLE = False
print("faster-whisper not found. To use local transcription, install with: pip install faster-whisper")
import sys
import logging
class SpeechToText:
def __init__(self):
@@ -56,8 +51,7 @@ class SpeechToText:
return temp_file.name
def load_local_model(self, model_size):
if not WHISPER_AVAILABLE:
return False, "faster-whisper not installed. Install with: pip install faster-whisper"
import faster_whisper
try:
if self.local_model is None:
@@ -84,6 +78,10 @@ class SpeechToText:
return False, f"Error during local transcription: {str(e)}", None
def transcribe_audio(self, model_size, AUDIO=None, audio_path=None, video_path=None):
# Check Python version and warn if 3.12 or higher
if sys.version_info > (3, 12):
logging.warning("⚠️⚠️⚠️ Warning: You are using Python {}.{} or higher. This may cause compatibility issues with some dependencies (e.g., faster_whisper). Consider using Python 3.11 or 3.12 instead. ⚠️⚠️⚠️".format(sys.version_info.major, sys.version_info.minor))
import faster_whisper
transcript = "No valid audio input provided"
detected_language = ""
temp_wav_path = None

206
split_image.py Normal file
View File

@@ -0,0 +1,206 @@
import torch
class SplitImageGrid:
@classmethod
def INPUT_TYPES(cls):
return {
"required": {
"image": ("IMAGE",),
"rows": ("INT", {"default": 1, "min": 1, "max": 9}),
"columns": ("INT", {"default": 1, "min": 1, "max": 9}),
"MODIFIED_part_index": ("INT", {"default": 1, "min": 1, "max": 9}),
}
}
RETURN_TYPES = ["IMAGE"] * 9 + ["INT", "INT", "IMAGE", "INT"]
RETURN_NAMES = [f"part_{i}" for i in range(1, 10)] + ["rows", "columns", "MODIFIED_part", "MODIFIED_part_index"]
FUNCTION = "split"
CATEGORY = "image"
def split(self, image, rows, columns, MODIFIED_part_index):
# Get image dimensions
B, H, W, C = image.shape
# Removed check: if H % rows != 0 or W % columns != 0:
# raise ValueError("Image dimensions must be divisible by rows and columns")
# Calculate base part dimensions
part_height = H // rows
part_width = W // columns
O = 2 # Overlap of 2 pixels
parts = []
# Split image with overlap
for r in range(rows):
for c in range(columns):
# Define slicing indices with overlap
h_start = max(0, r * part_height - O) # Extend O pixels up, but not beyond top
h_end = min(H, (r + 1) * part_height + O) # Extend O pixels down, not beyond bottom
w_start = max(0, c * part_width - O) # Extend O pixels left, not beyond left edge
w_end = min(W, (c + 1) * part_width + O) # Extend O pixels right, not beyond right edge
part = image[:, h_start:h_end, w_start:w_end, :]
parts.append(part)
# Pad unused parts with None
while len(parts) < 9:
parts.append(None)
# Adjust MODIFIED_part_index to 0-based and validate
MODIFIED_index = MODIFIED_part_index - 1
if MODIFIED_index < 0 or MODIFIED_index >= rows * columns:
raise ValueError(f"MODIFIED_part_index {MODIFIED_part_index} is out of range for {rows}x{columns} grid")
MODIFIED_part = parts[MODIFIED_index]
return tuple(parts + [rows, columns, MODIFIED_part, MODIFIED_part_index])
class ReassembleImageGrid:
@classmethod
def INPUT_TYPES(cls):
return {
"required": {
"original": ("IMAGE",),
"rows": ("INT", {"default": 1, "min": 1, "max": 10}),
"columns": ("INT", {"default": 1, "min": 1, "max": 10}),
},
"optional": {
"part_1": ("IMAGE",),
"part_2": ("IMAGE",),
"part_3": ("IMAGE",),
"part_4": ("IMAGE",),
"part_5": ("IMAGE",),
"part_6": ("IMAGE",),
"part_7": ("IMAGE",),
"part_8": ("IMAGE",),
"part_9": ("IMAGE",),
"MODIFIED_part": ("IMAGE",),
"MODIFIED_part_index": ("INT", {"default": 0, "min": 0, "max": 9}),
"reference_video_part_index": ("INT", {"default": 0, "min": 0, "max": 9}),
"auto_resize": ("BOOLEAN", {"default": True}), # Add option to enable/disable auto-resizing
}
}
RETURN_TYPES = ["IMAGE"]
RETURN_NAMES = ["image"]
FUNCTION = "reassemble"
CATEGORY = "image"
def repeat_frames(self, tensor, k):
"""Repeat the tensor k times along the batch dimension."""
return tensor.repeat(k, 1, 1, 1) if k > 1 else tensor
def adjust_frame_count_with_repeat(self, tensor, B_ref, B_original):
"""Adjust frame count, considering repetition if B_ref ≈ k * B_original."""
if B_original == 0:
raise ValueError("Original frame count is zero")
k = round(B_ref / B_original)
if k > 0 and abs(B_ref - k * B_original) <= 1:
repeated = self.repeat_frames(tensor, k)
if repeated.shape[0] > B_ref:
repeated = repeated[:B_ref]
elif repeated.shape[0] < B_ref:
pad_size = B_ref - repeated.shape[0]
last_frame = repeated[-1:].repeat(pad_size, 1, 1, 1)
repeated = torch.cat([repeated, last_frame], dim=0)
return repeated
else:
return self.adjust_frame_count(tensor, B_ref)
def adjust_frame_count(self, tensor, target_B):
"""Adjust the frame count of a tensor to match target_B by repeating or skipping frames."""
B = tensor.shape[0]
if B == target_B:
return tensor
indices = torch.linspace(0, B - 1, steps=target_B).round().long()
indices = indices.clamp(0, B - 1)
return tensor[indices]
def resize_tensor(self, tensor, target_height, target_width):
"""Resize tensor to target dimensions using interpolation."""
import torch.nn.functional as F
B, H, W, C = tensor.shape
# PyTorch's F.interpolate expects [B, C, H, W] format
# So we need to permute, resize, then permute back
tensor_BCHW = tensor.permute(0, 3, 1, 2) # [B, H, W, C] -> [B, C, H, W]
# Resize using bilinear interpolation
resized = F.interpolate(
tensor_BCHW,
size=(target_height, target_width),
mode='bilinear',
align_corners=False
)
# Convert back to [B, H, W, C] format
return resized.permute(0, 2, 3, 1) # [B, C, H, W] -> [B, H, W, C]
def reassemble(self, original, rows, columns, part_1=None, part_2=None, part_3=None,
part_4=None, part_5=None, part_6=None, part_7=None, part_8=None,
part_9=None, MODIFIED_part=None, MODIFIED_part_index=0,
reference_video_part_index=0, auto_resize=True):
# Get original dimensions
B, H, W, C = original.shape
# Calculate part dimensions
part_height = H // rows
part_width = W // columns
O = 2 # Overlap pixels, matching SplitImageGrid
parts = [part_1, part_2, part_3, part_4, part_5, part_6, part_7, part_8, part_9]
# Override with MODIFIED_part if provided
if MODIFIED_part is not None and MODIFIED_part_index > 0:
index = MODIFIED_part_index - 1
if index < 0 or index >= 9:
raise ValueError(f"Invalid MODIFIED_part_index: {MODIFIED_part_index}")
parts[index] = MODIFIED_part
# Handle reference part logic
if reference_video_part_index > 0:
ref_index = reference_video_part_index - 1
if parts[ref_index] is None:
raise ValueError(f"Reference part {reference_video_part_index} is not provided")
B_ref = parts[ref_index].shape[0]
original = self.adjust_frame_count_with_repeat(original, B_ref, B)
for i in range(9):
if parts[i] is not None and i != ref_index:
parts[i] = self.adjust_frame_count_with_repeat(parts[i], B_ref, B)
elif i == ref_index:
parts[i] = parts[i]
else:
B_ref = B
# Clone original to avoid modifying it
reassembled = original.clone()
# Reassemble the parts into the grid
for i, part in enumerate(parts, start=1):
if part is not None:
# Determine part position
row = (i - 1) // columns
col = (i - 1) % columns
# Calculate cropping offsets based on position
crop_top = O if row > 0 else 0
crop_left = O if col > 0 else 0
# Get the cropped part
cropped_part = part[:, crop_top:, crop_left:, :]
# Check if resize is needed and enabled
if auto_resize and (cropped_part.shape[1] != part_height or cropped_part.shape[2] != part_width):
print(f"Resizing part {i} from {cropped_part.shape[1:3]} to ({part_height}, {part_width})")
cropped_part = self.resize_tensor(cropped_part, part_height, part_width)
elif not auto_resize and (cropped_part.shape[1] != part_height or cropped_part.shape[2] != part_width):
# If auto-resize is disabled, still throw the error
raise ValueError(f"Cropped part {i} has incorrect shape. Expected ({part_height}, {part_width}, {C}), got {cropped_part.shape[1:]}")
# Validate frame count
if cropped_part.shape[0] != B_ref:
raise ValueError(f"Cropped part {i} has incorrect frame count. Expected {B_ref}, got {cropped_part.shape[0]}")
# Place cropped part into reassembled image
h_start = row * part_height
h_end = h_start + part_height
w_start = col * part_width
w_end = w_start + part_width
reassembled[:, h_start:h_end, w_start:w_end, :] = cropped_part
return (reassembled,)

325
style_selector.py Normal file
View File

@@ -0,0 +1,325 @@
import os
import random
from aiohttp import web
from server import PromptServer
# Shared data structures (unchanged)
CATEGORIES = ["Painting", "Photography", "Digital Art", "3D Rendering", "Illustration"]
BRANCHES = {
"Painting": [
"Renaissance", "Baroque", "Rococo", "Neoclassicism",
"Romanticism", "Realism", "Impressionism", "Post-Impressionism",
"Expressionism", "Fauvism", "Cubism", "Futurism", "Dadaism",
"Surrealism", "Abstract Expressionism", "Pop Art", "Op Art",
"Minimalism"
],
"Photography": [
"Black and White", "Color", "Vintage", "Sepia Tone", "HDR",
"Long Exposure", "Macro", "Portrait", "Landscape", "Street",
"Fashion", "Analog Film", "Cinematic"
],
"Digital Art": [
"Digital Painting", "Vector Art", "Pixel Art", "Fractal Art",
"Algorithmic Art", "Glitch Art"
],
"3D Rendering": [
"Low Poly", "Voxel", "Isometric", "Ray Tracing"
],
"Illustration": [
"Line Art", "Cartoon", "Comic Book", "Manga", "Anime",
"Technical Illustration", "Botanical Illustration",
"Architectural Rendering", "Concept Art", "Storyboard Art"
],
}
BRANCHES_MODELS = {
"Photography": [
("SDXL", "urn:air:sdxl:checkpoint:civitai:101055@128078", "https://civitai.green/models/101055?modelVersionId=128078"),
("Juggernaut XL", "urn:air:sdxl:checkpoint:civitai:133005@166909", "https://civitai.green/models/133005?modelVersionId=166909"),
("Realistic Stock Photo", "urn:air:sdxl:checkpoint:civitai:139565@154593", "https://civitai.green/models/139565?modelVersionId=154593"),
],
"Illustration": [
("Hassaku XL", "urn:air:sdxl:checkpoint:civitai:140272@176059", "https://civitai.green/models/140272?modelVersionId=176059"),
("[Lah] Mysterious", "urn:air:sdxl:checkpoint:civitai:118441@162380", "https://civitai.green/models/118441?modelVersionId=162380"),
("Copax TimeLessXL", "urn:air:sdxl:checkpoint:civitai:118111@1108377", "https://civitai.green/models/118111?modelVersionId=172160"),
],
"3D Rendering": [
("Samaritan 3D Cartoon", "urn:air:sdxl:checkpoint:civitai:81270@144566", "https://civitai.green/models/81270?modelVersionId=144566"),
("FormulaXL", "urn:air:sdxl:checkpoint:civitai:129922@160525", "https://civitai.green/models/129922?modelVersionId=160525"),
],
"Digital Art": [
("BriXL", "urn:air:sdxl:checkpoint:civitai:131703@166762", "https://civitai.green/models/131703?modelVersionId=166762"),
("SDXL Unstable Diffusers", "urn:air:sdxl:checkpoint:civitai:84040@395107", "https://civitai.green/models/84040?modelVersionId=395107"),
],
"Painting": [
("Copax TimeLessXL", "urn:air:sdxl:checkpoint:civitai:118111@1108377", "https://civitai.green/models/118111?modelVersionId=172160"),
("PixelPaint - Beautiful Painting Style", "urn:air:sdxl:checkpoint:civitai:284101@319693", "https://civitai.green/models/284101?modelVersionId=319693"),
],
}
MODELS = {model_name: (urn, link) for category in BRANCHES_MODELS for model_name, urn, link in BRANCHES_MODELS[category]}
# Counter files
STYLE_LIST_COUNTER_FILE = os.path.join("Bjornulf", "style_list_counter.txt")
MODEL_LIST_COUNTER_FILE = os.path.join("Bjornulf", "model_list_counter.txt")
class StyleSelector:
@classmethod
def INPUT_TYPES(cls):
ALL_STYLES = sorted(set(style for styles in BRANCHES.values() for style in styles))
ALL_MODELS = ["None"] + sorted(MODELS.keys())
return {
"required": {
"category": (CATEGORIES,),
"style": (ALL_STYLES,),
"model": (ALL_MODELS, {"default": "None"}),
"seed": ("INT", {"default": -1, "min": -1, "max": 0x7FFFFFFFFFFFFFFF}),
"LOOP_random_LIST": ("BOOLEAN", {"default": False}),
"LOOP_style_LIST": ("BOOLEAN", {"default": False}),
"LOOP_SEQUENTIAL": ("BOOLEAN", {"default": False}),
"jump": ("INT", {"default": 1, "min": 1, "max": 100, "step": 1}),
}
}
RETURN_TYPES = ("STRING", "STRING", "STRING", "STRING")
RETURN_NAMES = (
"selected_category",
"selected_style_LIST",
"recommended_with_selected_category",
"random_LIST_with_selected_category"
)
OUTPUT_IS_LIST = (False, True, False, True)
FUNCTION = "select_style"
CATEGORY = "Bjornulf"
def format_style(self, s, descriptor, model=None):
"""Helper function to format a style string."""
if model != "None" and model in MODELS:
urn, link = MODELS[model]
return f"{s} {descriptor};{model};{urn};{link}"
return f"{s} {descriptor}"
def get_next_index(self, counter_file, jump, max_items):
"""Get the next index from the counter file, stopping at max_items."""
os.makedirs(os.path.dirname(counter_file), exist_ok=True)
try:
with open(counter_file, 'r') as f:
current_index = int(f.read().strip())
except (FileNotFoundError, ValueError):
current_index = -jump # Start before the first index (e.g., -1 if jump=1)
next_index = current_index + jump
if next_index >= max_items:
raise ValueError(f"Counter has reached its limit of {max_items}. Reset counter to continue.")
with open(counter_file, 'w') as f:
f.write(str(next_index))
return next_index
def select_style(self, category, style, seed, LOOP_random_LIST, LOOP_style_LIST, LOOP_SEQUENTIAL, jump, model=None):
DESCRIPTORS = {
"Painting": "painting",
"Photography": "photography",
"Illustration": "illustration",
}
rng = random.Random() if seed == -1 else random.Random(seed)
descriptor = DESCRIPTORS.get(category, "")
# Get styles and models for the category
styles = BRANCHES.get(category, [])
models = BRANCHES_MODELS.get(category, [])
# selected_category: Single string
selected_category = category
# selected_style_LIST: List based on loop mode
if not styles:
selected_style_LIST = ["No styles found."]
elif LOOP_SEQUENTIAL and LOOP_style_LIST:
# Sequential mode for styles
max_styles = len(styles)
next_index = self.get_next_index(STYLE_LIST_COUNTER_FILE, jump, max_styles)
selected_style = styles[next_index]
selected_style_LIST = [self.format_style(selected_style, descriptor, model)]
elif LOOP_style_LIST:
# Return full list of styles
selected_style_LIST = [self.format_style(s, descriptor, model) for s in styles]
else:
# Default: single selected style
selected_style_LIST = [self.format_style(style, descriptor, model)]
# recommended_with_selected_category: Single string based on input style
if models:
recommended_model = models[0][0] # First model's name
recommended_with_selected_category = self.format_style(style, descriptor, recommended_model)
else:
recommended_with_selected_category = ""
# random_LIST_with_selected_category: List based on LOOP_random_LIST and LOOP_SEQUENTIAL
if LOOP_SEQUENTIAL and LOOP_random_LIST:
# Sequential mode for models
if not models:
random_LIST_with_selected_category = []
else:
max_models = len(models)
next_index = self.get_next_index(MODEL_LIST_COUNTER_FILE, jump, max_models)
selected_model = models[next_index][0]
random_LIST_with_selected_category = [self.format_style(style, descriptor, selected_model)]
elif LOOP_random_LIST:
# Return list of selected style with all models
random_LIST_with_selected_category = [self.format_style(style, descriptor, m[0]) for m in models] if models else []
else:
# Default: single random style and model
if models:
random_model = rng.choice(models)[0]
random_style = rng.choice(styles)
random_LIST_with_selected_category = [self.format_style(random_style, descriptor, random_model)]
else:
random_LIST_with_selected_category = []
return (
selected_category,
selected_style_LIST,
recommended_with_selected_category,
random_LIST_with_selected_category
)
# API endpoints for counter management
@PromptServer.instance.routes.post("/reset_style_list_counter")
async def reset_style_list_counter(request):
try:
os.remove(STYLE_LIST_COUNTER_FILE)
return web.json_response({"success": True}, status=200)
except FileNotFoundError:
return web.json_response({"success": True}, status=200)
except Exception as e:
return web.json_response({"success": False, "error": str(e)}, status=500)
@PromptServer.instance.routes.post("/reset_model_list_counter")
async def reset_model_list_counter(request):
try:
os.remove(MODEL_LIST_COUNTER_FILE)
return web.json_response({"success": True}, status=200)
except FileNotFoundError:
return web.json_response({"success": True}, status=200)
except Exception as e:
return web.json_response({"success": False, "error": str(e)}, status=500)
@PromptServer.instance.routes.post("/get_style_list_counter")
async def get_style_list_counter(request):
try:
with open(STYLE_LIST_COUNTER_FILE, 'r') as f:
current_index = int(f.read().strip())
return web.json_response({"success": True, "value": current_index + 1}, status=200)
except (FileNotFoundError, ValueError):
return web.json_response({"success": True, "value": 0}, status=200)
except Exception as e:
return web.json_response({"success": False, "error": str(e)}, status=500)
@PromptServer.instance.routes.post("/get_model_list_counter")
async def get_model_list_counter(request):
try:
with open(MODEL_LIST_COUNTER_FILE, 'r') as f:
current_index = int(f.read().strip())
return web.json_response({"success": True, "value": current_index + 1}, status=200)
except (FileNotFoundError, ValueError):
return web.json_response({"success": True, "value": 0}, status=200)
except Exception as e:
return web.json_response({"success": False, "error": str(e)}, status=500)
# class StyleSelector:
# @classmethod
# def INPUT_TYPES(cls):
# # Input configuration remains unchanged
# ALL_STYLES = sorted(set(style for styles in BRANCHES.values() for style in styles))
# ALL_MODELS = ["None"] + sorted(MODELS.keys())
# return {
# "required": {
# "category": (CATEGORIES,),
# "style": (ALL_STYLES,),
# "model": (ALL_MODELS, {"default": "None"}),
# "seed": ("INT", {"default": -1, "min": -1, "max": 0x7FFFFFFFFFFFFFFF}),
# "LOOP_random_LIST": ("BOOLEAN", {"default": False}),
# "LOOP_style_LIST": ("BOOLEAN", {"default": False}),
# }
# }
# RETURN_TYPES = ("STRING", "STRING", "STRING", "STRING")
# RETURN_NAMES = (
# "selected_category",
# "selected_style_LIST",
# "recommended_with_selected_category",
# "random_LIST_with_selected_category"
# )
# OUTPUT_IS_LIST = (False, True, False, True) # Updated: third element is now False
# FUNCTION = "select_style"
# CATEGORY = "Bjornulf"
# def select_style(self, category, style, seed, LOOP_random_LIST, LOOP_style_LIST, model=None):
# DESCRIPTORS = {
# "Painting": "painting",
# "Photography": "photography",
# "Illustration": "illustration",
# }
# rng = random.Random() if seed == -1 else random.Random(seed)
# descriptor = DESCRIPTORS.get(category, "")
# # selected_category: Single string (unchanged)
# selected_category = category
# # selected_style_LIST: List (unchanged)
# if LOOP_style_LIST:
# styles = BRANCHES.get(category, [])
# if model != "None" and model in MODELS:
# urn, link = MODELS[model]
# selected_style_LIST = [
# f"{s} {descriptor};{model};{urn};{link}" for s in styles
# ]
# else:
# selected_style_LIST = [f"{s} {descriptor}" for s in styles]
# else:
# if model != "None" and model in MODELS:
# urn, link = MODELS[model]
# selected_style_LIST = [f"{style} {descriptor};{model};{urn};{link}"]
# else:
# selected_style_LIST = [f"{style} {descriptor}"]
# # recommended_with_selected_category: Now a single string
# if category in BRANCHES_MODELS and BRANCHES_MODELS[category]:
# recommended_model = BRANCHES_MODELS[category][0] # First model
# recommended_with_selected_category = f"{style} {descriptor};{recommended_model[0]};{recommended_model[1]};{recommended_model[2]}"
# else:
# recommended_with_selected_category = ""
# # random_LIST_with_selected_category: List (unchanged)
# if LOOP_random_LIST:
# if category in BRANCHES_MODELS and BRANCHES_MODELS[category]:
# models = BRANCHES_MODELS[category]
# random_LIST_with_selected_category = [
# f"{style} {descriptor};{m[0]};{m[1]};{m[2]}" for m in models
# ]
# else:
# random_LIST_with_selected_category = []
# else:
# if category in BRANCHES_MODELS and BRANCHES_MODELS[category]:
# random_model = rng.choice(BRANCHES_MODELS[category])
# random_style_base = rng.choice(BRANCHES[category])
# random_style = f"{random_style_base} {descriptor}" if descriptor else random_style_base
# random_LIST_with_selected_category = [
# f"{random_style};{random_model[0]};{random_model[1]};{random_model[2]}"
# ]
# else:
# random_LIST_with_selected_category = []
# return (
# selected_category,
# selected_style_LIST,
# recommended_with_selected_category,
# random_LIST_with_selected_category
# )

27
switches.py
View File

@@ -45,4 +45,29 @@ class SwitchText:
if switch:
return (STRING,)
else:
return ("",)
return ("",)
class ConditionalSwitch:
@classmethod
def INPUT_TYPES(cls):
return {
"required": {
"condition": ("BOOLEAN", {"default": False}), # True or False to pick the path
},
"optional": {
"input_data": (Everything("*"), {"forceInput": True}), # Passthrough any data type
}
}
RETURN_TYPES = (Everything("*"), Everything("*")) # Two outputs, dynamically typed
RETURN_NAMES = ("TRUE", "FALSE") # Named outputs for clarity
FUNCTION = "switch"
CATEGORY = "Utilities"
def switch(self, condition, input_data=None):
if condition:
# If condition is True, send data to TRUE output, FALSE gets None
return (input_data, None)
else:
# If condition is False, send data to FALSE output, TRUE gets None
return (None, input_data)

112
text_analyzer.py Normal file
View File

@@ -0,0 +1,112 @@
import re
import subprocess
import sys
# Load the spaCy model once globally to keep it lightweight
def ensure_model(model_name="en_core_web_sm"):
try:
# Try to load the model
import spacy
spacy.load(model_name)
except OSError:
# If the model isn't found, download it
print(f"Model '{model_name}' not found. Downloading now...")
subprocess.check_call([sys.executable, "-m", "spacy", "download", model_name])
print(f"Model '{model_name}' downloaded successfully.")
class TextAnalyzer:
@classmethod
def INPUT_TYPES(cls):
return {
"required": {
"text": ("STRING", {"multiline": True})
}
}
# Define the output types and their names
RETURN_TYPES = ("INT", "INT", "INT", "STRING", "STRING", "FLOAT", "STRING", "STRING", "STRING", "STRING", "STRING", "STRING")
RETURN_NAMES = ("number_lines", "number_words", "number_characters", "language", "sentiment", "sentiment_polarity", "type", "character", "sentence", "subject", "action", "target")
FUNCTION = "analyze"
CATEGORY = "Bjornulf"
def analyze(self, text):
from langdetect import detect
from textblob import TextBlob
import spacy
ensure_model()
nlp = spacy.load("en_core_web_sm")
# **Statistics**
# Count lines by splitting on newline characters
lines = len(text.split('\n'))
# Count words by splitting on whitespace
words = len(text.split())
# Count total characters including spaces and punctuation
characters = len(text)
# **Dialog or Description Detection**
# Check if the text starts with a name followed by a colon (e.g., "Jessica:")
dialog_match = re.match(r'^([A-Za-z]+):', text)
if dialog_match:
type_ = 'dialog'
character = dialog_match.group(1) # Extract the character name
spoken_text = text[dialog_match.end():].strip() # Text after the colon
else:
type_ = 'description'
character = None
spoken_text = text
# **Language Detection**
try:
language = detect(spoken_text)
except:
language = 'unknown'
# **Sentiment Analysis**
blob = TextBlob(spoken_text)
polarity = blob.sentiment.polarity
if polarity > 0:
sentiment = 'positive'
elif polarity < 0:
sentiment = 'negative'
else:
sentiment = 'neutral'
# **Subject, Action, Target Extraction**
# Only perform NLP if the language is English (for spaCy compatibility)
if language == 'en':
doc = nlp(spoken_text)
action = None
subject = None
target = None
# Look for the main verb (action) and its subject and object
for token in doc:
if token.dep_ == 'ROOT' and token.pos_ == 'VERB':
action = token.text
subject_tokens = [w for w in token.children if w.dep_ == 'nsubj']
target_tokens = [w for w in token.children if w.dep_ == 'dobj']
if subject_tokens:
subject = subject_tokens[0].text
if target_tokens:
target = target_tokens[0].text
break
else:
subject, action, target = None, None, None
# **Return Results**
# Convert None to empty strings for ComfyUI compatibility
return (
lines,
words,
characters,
language,
sentiment,
polarity,
type_,
character or "",
spoken_text or "",
subject or "",
action or "",
target or ""
)

53
text_generator.py
View File

@@ -1638,7 +1638,8 @@ class TextGeneratorOutfitFemale:
}
}
RETURN_TYPES = ("GEN_OUTFIT",)
RETURN_TYPES = ("GEN_OUTFIT,STRING",)
RETURN_NAMES = ("GEN_OUTFIT",)
FUNCTION = "generate"
CATEGORY = "Bjornulf"
@@ -1796,7 +1797,8 @@ class TextGeneratorOutfitMale:
}
}
RETURN_TYPES = ("GEN_OUTFIT",)
RETURN_TYPES = ("GEN_OUTFIT,STRING",)
RETURN_NAMES = ("GEN_OUTFIT",)
FUNCTION = "generate"
CATEGORY = "Bjornulf"
@@ -1976,13 +1978,14 @@ class TextGeneratorCharacterFemale:
"CUSTOM_PROMPT": ("STRING", {"multiline": True, "default": ""}),
},
"optional": {
"add_GEN_CHARACTER": ("GEN_CHARACTER",),
"GEN_OUTFIT": ("GEN_OUTFIT",),
"GEN_POSE": ("GEN_POSE",),
"add_GEN_CHARACTER": ("GEN_CHARACTER,STRING",),
"GEN_OUTFIT": ("GEN_OUTFIT,STRING",),
"GEN_POSE": ("GEN_POSE,STRING",),
}
}
RETURN_TYPES = ("GEN_CHARACTER",)
RETURN_TYPES = ("GEN_CHARACTER,STRING",)
RETURN_NAMES = ("GEN_CHARACTER",)
FUNCTION = "generate"
CATEGORY = "Bjornulf"
@@ -2120,13 +2123,14 @@ class TextGeneratorCharacterMale:
"CUSTOM_PROMPT": ("STRING", {"multiline": True, "default": ""}),
},
"optional": {
"add_GEN_CHARACTER": ("GEN_CHARACTER",),
"GEN_OUTFIT": ("GEN_OUTFIT",),
"GEN_POSE": ("GEN_POSE",),
"add_GEN_CHARACTER": ("GEN_CHARACTER,STRING",),
"GEN_OUTFIT": ("GEN_OUTFIT,STRING",),
"GEN_POSE": ("GEN_POSE,STRING",),
}
}
RETURN_TYPES = ("GEN_CHARACTER",)
RETURN_TYPES = ("GEN_CHARACTER,STRING",)
RETURN_NAMES = ("GEN_CHARACTER",)
FUNCTION = "generate"
CATEGORY = "Bjornulf"
@@ -2253,7 +2257,8 @@ class TextGeneratorStyle:
}
}
RETURN_TYPES = ("GEN_STYLE",)
RETURN_TYPES = ("GEN_STYLE,STRING",)
RETURN_NAMES = ("GEN_STYLE",)
FUNCTION = "generate"
CATEGORY = "Bjornulf"
@@ -2316,7 +2321,8 @@ class TextGeneratorScene:
}
}
RETURN_TYPES = ("GEN_SCENE",)
RETURN_TYPES = ("GEN_SCENE,STRING",)
RETURN_NAMES = ("GEN_SCENE",)
FUNCTION = "generate"
CATEGORY = "Bjornulf"
@@ -2373,9 +2379,9 @@ class TextGenerator:
"CUSTOM_PROMPT": ("STRING", {"multiline": True, "default": ""}),
},
"optional": {
"GEN_STYLE": ("GEN_STYLE",),
"GEN_CHARACTER": ("GEN_CHARACTER",),
"GEN_SCENE": ("GEN_SCENE",),
"GEN_STYLE": ("GEN_STYLE,STRING",),
"GEN_CHARACTER": ("GEN_CHARACTER,STRING",),
"GEN_SCENE": ("GEN_SCENE,STRING",),
}
}
@@ -2889,12 +2895,13 @@ class TextGeneratorCharacterPose:
"CUSTOM_PROMPT": ("STRING", {"multiline": True, "default": ""}),
},
"optional": {
"GEN_OBJECT": ("GEN_OBJECT",),
"add_GEN_POSE": ("GEN_POSE",),
"GEN_OBJECT": ("GEN_OBJECT,STRING",),
"add_GEN_POSE": ("GEN_POSE,STRING",),
}
}
RETURN_TYPES = ("GEN_POSE",)
RETURN_TYPES = ("GEN_POSE,STRING",)
RETURN_NAMES = ("GEN_POSE",)
FUNCTION = "generate"
CATEGORY = "Bjornulf"
@@ -2968,11 +2975,12 @@ class TextGeneratorCharacterObject:
"CUSTOM_PROMPT_SUFFIX": ("STRING", {"default": "", "multiline": True}),
},
"optional": {
"add_GEN_OBJECT": ("GEN_OBJECT",),
"add_GEN_OBJECT": ("GEN_OBJECT,STRING",),
}
}
RETURN_TYPES = ("GEN_OBJECT",)
RETURN_TYPES = ("GEN_OBJECT,STRING",)
RETURN_NAMES = ("GEN_OBJECT",)
FUNCTION = "generate"
CATEGORY = "Bjornulf"
@@ -3017,11 +3025,12 @@ class TextGeneratorCharacterCreature:
"CUSTOM_PROMPT": ("STRING", {"multiline": True, "default": ""}),
},
"optional": {
"add_GEN_CHARACTER": ("GEN_CHARACTER",),
"add_GEN_CHARACTER": ("GEN_CHARACTER,STRING",),
}
}
RETURN_TYPES = ("GEN_CHARACTER",)
RETURN_TYPES = ("GEN_CHARACTER,STRING",)
RETURN_NAMES = ("GEN_CHARACTER",)
FUNCTION = "generate"
CATEGORY = "Bjornulf"

32
text_to_speech.py
View File

@@ -98,21 +98,25 @@ class TextToSpeech:
try:
response = requests.get(url, stream=True)
response.raise_for_status()
response.raise_for_status() # Raises an HTTPError for 4xx/5xx status codes
audio_data = io.BytesIO()
for chunk in response.iter_content(chunk_size=8192):
audio_data.write(chunk)
audio_data.seek(0)
# Check if the audio data is empty
if audio_data.getbuffer().nbytes == 0:
raise ValueError("Received empty audio data from server")
return audio_data
except requests.RequestException as e:
print(f"Error generating audio: {e}")
return io.BytesIO()
raise # Re-raise the exception to stop the workflow
except Exception as e:
print(f"Unexpected error: {e}")
return io.BytesIO()
raise # Re-raise any other unexpected exceptions
def play_audio(self, audio: AudioSegment) -> None:
if sys.platform.startswith('win'):
@@ -195,28 +199,6 @@ class TextToSpeech:
audio_output, _, duration = self.process_audio_data(autoplay, audio_data, full_path if save_audio else None)
return (audio_output, save_path, full_path, duration)
# GET VOICE FROM TTS SERVER (Not good for now)
# @PromptServer.instance.routes.post("/bjornulf_TTS_get_voices")
# async def get_voices(request):
# try:
# data = await request.json()
# TTS_url = data.get('url', 'http://localhost:8020')
# # Use requests instead of client_session
# response = requests.get(f"{TTS_url}/speakers")
# response.raise_for_status() # Raise an exception for bad status codes
# voices = response.json()
# # Transform the response to just get the voice_ids
# voice_ids = [voice["voice_id"] for voice in voices]
# return web.json_response({"voices": voice_ids})
# except requests.RequestException as e:
# print(f"Error fetching voices: {str(e)}")
# return web.json_response({"error": str(e)}, status=500)
# except Exception as e:
# print(f"Unexpected error: {str(e)}")
# return web.json_response({"error": str(e)}, status=500)
# Scan folder
@PromptServer.instance.routes.post("/bjornulf_TTS_get_voices")
async def get_voices(request):

2
text_to_variable.py
View File

@@ -9,6 +9,6 @@ class TextToVariable:
CATEGORY = "Custom"
def process(self, variable_name, text_value):
text_value = text_value.replace("\n", ";")
text_value = text_value.replace("\n", "")
output_string = f"{variable_name} = {text_value}"
return (output_string,)

73
video_pingpong.py
View File

@@ -3,13 +3,17 @@ import os
import shutil
from PIL import Image
import numpy as np
import glob
import subprocess
class VideoPingPong:
@classmethod
def INPUT_TYPES(cls):
return {
"required": {
"optional": {
"images": ("IMAGE",),
"video_path": ("STRING", {"default": ""}),
"use_python_ffmpeg": ("BOOLEAN", {"default": False}),
},
}
@@ -18,56 +22,75 @@ class VideoPingPong:
FUNCTION = "pingpong_images"
CATEGORY = "Bjornulf"
def pingpong_images(self, images):
# Create a clean folder to store the images
def extract_frames(self, video_path, temp_dir, use_python_ffmpeg):
"""Extract frames from a video file using FFmpeg, preserving original settings."""
if use_python_ffmpeg:
try:
import ffmpeg
(
ffmpeg
.input(video_path)
.output(os.path.join(temp_dir, 'frame_%04d.png'), start_number=0)
.run()
)
except ImportError:
raise RuntimeError("ffmpeg-python is not installed. Please install it or set use_python_ffmpeg to False.")
except ffmpeg.Error as e:
raise RuntimeError(f"Failed to extract frames using ffmpeg-python: {e}")
else:
ffmpeg_cmd = ['ffmpeg', '-i', video_path, '-start_number', '0', os.path.join(temp_dir, 'frame_%04d.png')]
try:
subprocess.run(ffmpeg_cmd, check=True)
except subprocess.CalledProcessError as e:
raise RuntimeError(f"Failed to extract frames using FFmpeg: {e}")
def pingpong_images(self, images=None, video_path="", use_python_ffmpeg=False):
"""Generate a ping-pong sequence from images or a video file, prioritizing images if provided."""
temp_dir = "temp_pingpong"
if os.path.exists(temp_dir):
shutil.rmtree(temp_dir)
os.makedirs(temp_dir)
try:
# Save each image in the temporary directory
num_images = images.shape[0]
for i in range(num_images):
img_tensor = images[i]
img_pil = Image.fromarray((img_tensor.cpu().numpy() * 255).astype('uint8'))
img_path = os.path.join(temp_dir, f"image_{i:04d}.png")
img_pil.save(img_path)
if images is not None:
num_frames = images.shape[0]
for i in range(num_frames):
img_tensor = images[i]
img_pil = Image.fromarray((img_tensor.cpu().numpy() * 255).astype('uint8'))
img_path = os.path.join(temp_dir, f"frame_{i:04d}.png")
img_pil.save(img_path)
elif video_path and os.path.exists(video_path):
self.extract_frames(video_path, temp_dir, use_python_ffmpeg)
else:
raise ValueError("Either images or a valid video_path must be provided")
# Create the pingpong sequence
pingpong_list = list(range(num_images)) + list(range(num_images - 2, 0, -1))
frame_files = sorted(glob.glob(os.path.join(temp_dir, "frame_*.png")))
num_frames = len(frame_files)
if num_frames == 0:
raise RuntimeError("No frames available to process")
pingpong_list = list(range(num_frames)) + list(range(num_frames - 2, 0, -1))
# Process images in batches
batch_size = 10
pingpong_tensors = []
for i in range(0, len(pingpong_list), batch_size):
batch = pingpong_list[i:i+batch_size]
batch = pingpong_list[i:i + batch_size]
batch_tensors = []
for j in batch:
img_path = os.path.join(temp_dir, f"image_{j:04d}.png")
img_path = os.path.join(temp_dir, f"frame_{j:04d}.png")
img_pil = Image.open(img_path)
img_np = np.array(img_pil).astype(np.float32) / 255.0
img_tensor = torch.from_numpy(img_np)
batch_tensors.append(img_tensor)
# Close the image to free up memory
img_pil.close()
# Stack the batch tensors
batch_tensor = torch.stack(batch_tensors)
pingpong_tensors.append(batch_tensor)
# Clear unnecessary variables
del batch_tensors
torch.cuda.empty_cache()
# Concatenate all batches
pingpong_tensor = torch.cat(pingpong_tensors, dim=0)
finally:
# Clean up the temporary directory
shutil.rmtree(temp_dir)
return (pingpong_tensor,)

113
video_preview.py
View File

@@ -3,18 +3,28 @@ import shutil
import time
import hashlib
from pathlib import Path
import subprocess
import numpy as np
import cv2
import tempfile
SUPPORTED_EXTENSIONS = {'.mp4', '.webm', '.ogg', '.mov', '.mkv'}
# Supported extensions for video inputs
SUPPORTED_VIDEO_EXTENSIONS = {'.mp4', '.webm', '.ogg', '.mov', '.mkv'}
class VideoPreview:
@classmethod
def INPUT_TYPES(cls):
return {
"required": {
"video_path": ("STRING", {"forceInput": True}),
"fps_for_IMAGES": ("FLOAT", {"default": 24.0, "min": 1.0, "max": 60.0}),
"autoplay": ("BOOLEAN", {"default": False}),
"mute": ("BOOLEAN", {"default": True}),
"loop": ("BOOLEAN", {"default": False}),
},
"optional": {
"video_path": ("STRING", {"forceInput": True, "default": ""}),
"IMAGES": ("IMAGE", {"default": None}),
}
}
RETURN_TYPES = ()
@@ -22,31 +32,84 @@ class VideoPreview:
CATEGORY = "Bjornulf"
OUTPUT_NODE = True
def preview_video(self, video_path, autoplay, mute):
def preview_video(self, fps_for_IMAGES, autoplay, mute, loop, video_path="", IMAGES=None):
try:
if not video_path or not isinstance(video_path, str):
raise ValueError("Invalid video path provided")
video_path = os.path.abspath(video_path)
if not os.path.exists(video_path):
raise FileNotFoundError(f"Video file not found: {video_path}")
ext = Path(video_path).suffix.lower()
if ext not in SUPPORTED_EXTENSIONS:
raise ValueError(f"Unsupported video format: {ext}. Supported formats: {', '.join(SUPPORTED_EXTENSIONS)}")
# Destination directory for preview videos
dest_dir = os.path.join("output", "Bjornulf", "preview_video")
os.makedirs(dest_dir, exist_ok=True)
file_hash = hashlib.md5(open(video_path,'rb').read()).hexdigest()[:8]
timestamp = int(time.time())
base_name = Path(video_path).stem
dest_name = f"{base_name}_{timestamp}_{file_hash}{ext}"
dest_path = os.path.join(dest_dir, dest_name)
# Determine which input is provided
if video_path and isinstance(video_path, str) and video_path.strip():
video_path = os.path.abspath(video_path)
if not os.path.exists(video_path):
raise FileNotFoundError(f"Video file not found: {video_path}")
if not os.path.exists(dest_path):
shutil.copy2(video_path, dest_path)
ext = Path(video_path).suffix.lower() # e.g., '.mp4', '.webm'
if ext not in SUPPORTED_VIDEO_EXTENSIONS:
raise ValueError(f"Unsupported video format: {ext}. Supported formats: {', '.join(SUPPORTED_VIDEO_EXTENSIONS)}")
final_video_path = video_path
# Generate unique filename with original extension
file_hash = hashlib.md5(open(final_video_path, 'rb').read()).hexdigest()[:8]
timestamp = int(time.time())
base_name = "video_preview" # More descriptive than "image_sequence"
dest_name = f"{base_name}_{timestamp}_{file_hash}{ext}" # Keeps original extension
dest_path = os.path.join(dest_dir, dest_name)
shutil.copy2(final_video_path, dest_path)
print(f"Video copied to: {dest_path}")
elif IMAGES is not None and len(IMAGES) > 0:
# Use a unique temporary directory for this run
with tempfile.TemporaryDirectory(prefix="bjornulf_temp_video_") as temp_dir:
# Convert image tensors to files in the unique temp directory
image_files = []
for i, img_tensor in enumerate(IMAGES):
# Convert tensor (H, W, C) in range [0, 1] to numpy array in range [0, 255]
img_np = (img_tensor.numpy() * 255).astype(np.uint8)
# Ensure RGB format for OpenCV (ComfyUI IMAGES are typically RGB)
img_path = os.path.join(temp_dir, f"frame_{i:04d}.png")
cv2.imwrite(img_path, cv2.cvtColor(img_np, cv2.COLOR_RGB2BGR))
image_files.append(img_path)
if not image_files:
raise ValueError("No valid IMAGES provided to create a video.")
# Create temporary video using FFmpeg
output_video = os.path.join(temp_dir, "temp_video.mp4")
pattern = os.path.join(temp_dir, "frame_%04d.png")
cmd = [
"ffmpeg",
"-framerate", str(fps_for_IMAGES),
"-i", pattern,
"-c:v", "libx264",
"-pix_fmt", "yuv420p",
"-y", # Overwrite output file if it exists
output_video
]
subprocess.run(cmd, check=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
if not os.path.exists(output_video):
raise RuntimeError("Failed to create temporary video from IMAGES.")
final_video_path = output_video
# Generate unique destination filename
file_hash = hashlib.md5(open(final_video_path, 'rb').read()).hexdigest()[:8]
timestamp = int(time.time())
base_name = "image_sequence"
dest_name = f"{base_name}_{timestamp}_{file_hash}.mp4"
dest_path = os.path.join(dest_dir, dest_name)
# Copy the video to the preview directory
if not os.path.exists(dest_path):
shutil.copy2(final_video_path, dest_path)
else:
raise ValueError("Either 'video_path' or 'IMAGES' must be provided.")
# Successful return with video data
return {
"ui": {
"video": [dest_name, "Bjornulf/preview_video"],
@@ -54,15 +117,17 @@ class VideoPreview:
"width": 512,
"height": 512,
"autoplay": autoplay,
"mute": mute
"mute": mute,
"loop": loop
}
}
}
except Exception as e:
# Error case: return an empty list for "video" to prevent iteration error
return {
"ui": {
"error": str(e),
"video": None
"video": [], # Changed from None to []
"error": str(e)
}
}

48
web/js/API_civitai.js
View File

@@ -26,4 +26,52 @@ app.registerExtension({
});
}
}
});
app.registerExtension({
name: "Bjornulf.LoadCivitAILinks",
async nodeCreated(node) {
if (node.comfyClass === "Bjornulf_LoadCivitAILinks") {
// Add a refresh button widget
const refreshButton = node.addWidget(
"button",
"Refresh File List",
null,
() => {
fetch("/get_civitai_links_files", {
method: "POST",
})
.then((response) => response.json())
.then((data) => {
if (data.success) {
// Update the dropdown with the new file list
const dropdownWidget = node.widgets.find(w => w.name === "selected_file");
if (dropdownWidget) {
dropdownWidget.options.values = ["Not selected", ...data.files];
dropdownWidget.value = "Not selected";
app.ui.dialog.show(
"[LoadCivitAILinks] File list refreshed successfully!"
);
}
} else {
app.ui.dialog.show(
`[LoadCivitAILinks] Failed to refresh file list: ${
data.error || "Unknown error"
}`
);
}
})
.catch((error) => {
console.error(
"[LoadCivitAILinks] Error fetching links files:",
error
);
app.ui.dialog.show(
"[LoadCivitAILinks] An error occurred while refreshing the file list."
);
});
}
);
}
},
});

81
web/js/audio_preview.js Normal file
View File

@@ -0,0 +1,81 @@
import { api } from '../../../scripts/api.js';
import { app } from "../../../scripts/app.js";
// Function to display the audio preview
function displayAudioPreview(component, filename, category, autoplay, mute, loop) {
let audioWidget = component._audioWidget;
// Create the audio widget if it doesn't exist
if (!audioWidget) {
const container = document.createElement("div");
// Add the DOM widget to the component
audioWidget = component.addDOMWidget("Bjornulf", "preview", container, {
serialize: false,
hideOnZoom: false,
getValue() {
return container.value;
},
setValue(v) {
container.value = v;
},
});
// Initialize widget properties
audioWidget.value = { hidden: false, paused: false, params: {} };
audioWidget.parentElement = document.createElement("div");
audioWidget.parentElement.className = "audio_preview";
audioWidget.parentElement.style.width = "100%";
audioWidget.parentElement.style.marginBottom = "50px";
container.appendChild(audioWidget.parentElement);
// Create the audio element
audioWidget.audioElement = document.createElement("audio");
audioWidget.audioElement.controls = true;
audioWidget.audioElement.style.width = "100%";
// Hide the audio player on error
audioWidget.audioElement.addEventListener("error", () => {
audioWidget.parentElement.hidden = true;
});
audioWidget.parentElement.hidden = audioWidget.value.hidden;
audioWidget.parentElement.appendChild(audioWidget.audioElement);
component._audioWidget = audioWidget; // Store for reuse
}
// Set audio source and properties
const params = {
"filename": filename,
"subfolder": category,
"type": "temp",
"rand": Math.random().toString().slice(2, 12) // Cache-busting random parameter
};
const urlParams = new URLSearchParams(params);
audioWidget.audioElement.src = `api/view?${urlParams.toString()}`;
audioWidget.audioElement.autoplay = autoplay && !audioWidget.value.paused && !audioWidget.value.hidden;
audioWidget.audioElement.loop = loop;
}
// Register the extension
app.registerExtension({
name: "Bjornulf.AudioPreview",
async beforeRegisterNodeDef(nodeType, nodeData, appInstance) {
if (nodeData?.name === "Bjornulf_AudioPreview") {
// Define behavior when the node executes
nodeType.prototype.onExecuted = function(data) {
const autoplay = this.widgets.find(w => w.name === "autoplay")?.value ?? false;
const loop = this.widgets.find(w => w.name === "loop")?.value ?? false;
displayAudioPreview(this, data.audio[0], data.audio[1], autoplay, loop);
};
// Override computeSize to set a fixed height
nodeType.prototype.computeSize = function() {
const size = LiteGraph.LGraphNode.prototype.computeSize.call(this);
size[1] = 150; // Fixed height of 150px
return size;
};
}
}
});

53
web/js/fix_face.js Normal file
View File

@@ -0,0 +1,53 @@
import { app } from "../../../scripts/app.js";
app.registerExtension({
name: "Bjornulf.FixFace",
async nodeCreated(node) {
if (node.comfyClass === "Bjornulf_FixFace") {
const updateInputs = () => {
const initialWidth = node.size[0];
const numFacesWidget = node.widgets.find(w => w.name === "number_of_faces");
if (!numFacesWidget) return;
const numFaces = numFacesWidget.value;
// Initialize node.inputs if it doesn't exist
if (!node.inputs) {
node.inputs = [];
}
// Filter existing FACE_SETTINGS inputs
const existingInputs = node.inputs.filter(input => input.name.startsWith('FACE_SETTINGS_'));
// Add or remove inputs based on number_of_faces
if (existingInputs.length < numFaces) {
for (let i = existingInputs.length + 1; i <= numFaces; i++) {
const inputName = `FACE_SETTINGS_${i}`;
if (!node.inputs.find(input => input.name === inputName)) {
node.addInput(inputName, "FACE_SETTINGS");
}
}
} else {
node.inputs = node.inputs.filter(input => !input.name.startsWith('FACE_SETTINGS_') ||
parseInt(input.name.split('_')[2]) <= numFaces);
}
node.setSize(node.computeSize());
node.size[0] = initialWidth; // Keep width fixed
};
// Move number_of_faces widget to the top and set callback
const numFacesWidget = node.widgets.find(w => w.name === "number_of_faces");
if (numFacesWidget) {
node.widgets = [numFacesWidget, ...node.widgets.filter(w => w !== numFacesWidget)];
numFacesWidget.callback = () => {
updateInputs();
app.graph.setDirtyCanvas(true);
};
}
// Initial update after node creation
setTimeout(updateInputs, 0);
}
}
});

292
web/js/images_compare.js
View File

@@ -1,122 +1,135 @@
import { app } from "../../../scripts/app.js";
import { api } from "../../../scripts/api.js";
// Helper function to construct image URLs for ComfyUI
function imageDataToUrl(data) {
return api.apiURL(
`/view?filename=${encodeURIComponent(data.filename)}&type=${data.type}&subfolder=${data.subfolder}${app.getPreviewFormatParam()}`
`/view?filename=${encodeURIComponent(data.filename)}&type=${
data.type
}&subfolder=${data.subfolder}` + app.getPreviewFormatParam()
);
}
app.registerExtension({
name: "Bjornulf.FourImageViewer",
async nodeCreated(node) {
// Ensure this applies only to the specific node type
if (node.comfyClass !== "Bjornulf_FourImageViewer") return;
const marginTop = 90;
const verticalOffset = -30;
const minSize = 256; // Minimum size for the node
const maxSize = 2048; // Maximum size for the node
node.size = [512, 512 + marginTop];
node.images = new Array(4).fill(null);
const padding = 10;
// Constants for layout and sizing
const marginTop = 90; // Space at the top for node UI elements
const verticalOffset = -30; // Adjustment for canvas positioning
const minSize = 256; // Minimum size of the node
const maxSize = 2048; // Maximum size of the node
const padding = 10; // Padding around the image area
// Add resize handles
// Initialize node properties
node.size = [512, 512 + marginTop];
node.data = {
images: new Array(4).fill(null), // Array to hold up to 4 images
sliderX: null, // X position of the vertical slider
sliderY: null, // Y position of the horizontal slider
};
// Enable resizing
node.flags |= LiteGraph.RESIZABLE;
node.onResize = function(size) {
// Ensure square aspect ratio (excluding marginTop)
const minDimension = Math.max(minSize, Math.min(size[0], size[1] - marginTop));
// Handle resizing to maintain square aspect ratio
node.onResize = function (size) {
const minDimension = Math.max(
minSize,
Math.min(size[0], size[1] - marginTop)
);
const maxDimension = Math.min(maxSize, minDimension);
size[0] = maxDimension;
size[1] = maxDimension + marginTop;
// Update slider positions proportionally
const fullImgWidth = size[0] - 3 * padding;
const fullImgHeight = size[1] - padding - marginTop;
// Only update sliders if they exist (node has been initialized)
if (node.hasOwnProperty('sliderX')) {
// Update slider positions proportionally
if (node.data.hasOwnProperty("sliderX")) {
const oldWidth = node.size[0] - 3 * padding;
const oldHeight = node.size[1] - padding - marginTop;
// Calculate relative positions (0 to 1)
const relativeX = (node.sliderX - padding) / oldWidth;
const relativeY = (node.sliderY - marginTop) / oldHeight;
// Update slider positions
node.sliderX = padding + (fullImgWidth * relativeX);
node.sliderY = marginTop + (fullImgHeight * relativeY);
const relativeX = (node.data.sliderX - padding) / oldWidth;
const relativeY = (node.data.sliderY - marginTop) / oldHeight;
node.data.sliderX = padding + fullImgWidth * relativeX;
node.data.sliderY = marginTop + fullImgHeight * relativeY;
} else {
// Initial slider positions
node.sliderX = padding + fullImgWidth / 2;
node.sliderY = marginTop + fullImgHeight / 2;
node.data.sliderX = padding + fullImgWidth / 2;
node.data.sliderY = marginTop + fullImgHeight / 2;
}
node.size = size;
return size;
};
// Full area where images get drawn
const fullImgWidth = node.size[0] - 3 * padding;
const fullImgHeight = node.size[1] - padding - marginTop;
node.sliderX = padding + fullImgWidth / 2;
node.sliderY = marginTop + fullImgHeight / 2;
node.onMouseDown = function(e) {
// Handle mouse down to move sliders
node.onMouseDown = function (e) {
const rect = node.getBounding();
const [clickX, clickY] = [
e.canvasX - rect[0],
e.canvasY - rect[1] + verticalOffset
e.canvasY - rect[1] + verticalOffset,
];
const imgWidth = rect[2] - 3 * padding;
const imgHeight = rect[3] - padding - marginTop;
const xStart = padding;
const xEnd = xStart + imgWidth;
const yStart = marginTop;
const yEnd = yStart + imgHeight;
if (clickX >= xStart && clickX <= xEnd && clickY >= yStart && clickY <= yEnd) {
const hasImage2 = node.images[1] !== null;
const hasImage3 = node.images[2] !== null;
const hasImage4 = node.images[3] !== null;
const xStart = padding;
const xEnd = xStart + (node.size[0] - 3 * padding);
const yStart = marginTop;
const yEnd = yStart + (node.size[1] - padding - marginTop);
if (
clickX >= xStart &&
clickX <= xEnd &&
clickY >= yStart &&
clickY <= yEnd
) {
const hasImage2 = node.data.images[1] !== null;
const hasImage3 = node.data.images[2] !== null;
const hasImage4 = node.data.images[3] !== null;
// Lock sliderY to bottom if only two images are present
if (hasImage2 && !hasImage3 && !hasImage4) {
node.sliderY = yEnd;
node.data.sliderY = yEnd;
}
node.sliderX = Math.max(xStart, Math.min(clickX, xEnd));
node.sliderY = hasImage3 || hasImage4
? Math.max(yStart, Math.min(clickY, yEnd))
: yEnd;
node.data.sliderX = Math.max(xStart, Math.min(clickX, xEnd));
node.data.sliderY =
hasImage3 || hasImage4
? Math.max(yStart, Math.min(clickY, yEnd))
: yEnd;
app.graph.setDirtyCanvas(true, true);
return true;
}
return false;
};
node.onExecuted = async function(message) {
node.images = new Array(4).fill(null);
// Load images when the node is executed
node.onExecuted = async function (message) {
node.data.images = new Array(4).fill(null);
for (let i = 1; i <= 4; i++) {
const images = message[`images_${i}`] || [];
if (images.length) {
const imgData = images[0];
const img = new Image();
img.src = imageDataToUrl(imgData);
await new Promise(resolve => (img.onload = img.onerror = resolve));
node.images[i - 1] = img;
await new Promise((resolve) => (img.onload = img.onerror = resolve));
node.data.images[i - 1] = img;
}
}
app.graph.setDirtyCanvas(true, true);
};
node.onDrawForeground = function(ctx) {
const padding = 10;
// Render images and sliders
node.onDrawForeground = function (ctx) {
const xStart = padding;
const xEnd = xStart + (node.size[0] - 3 * padding);
const yStart = marginTop;
const yEnd = yStart + (node.size[1] - padding - marginTop);
const fullImgWidth = node.size[0] - 3 * padding;
const fullImgHeight = node.size[1] - padding - marginTop;
// Calculate fitted rectangle for image display
function getFittedDestRect(dx, dy, dWidth, dHeight, targetRatio) {
let newWidth = dWidth;
let newHeight = dWidth / targetRatio;
@@ -128,18 +141,34 @@ app.registerExtension({
const offsetY = dy + (dHeight - newHeight) / 2;
return [offsetX, offsetY, newWidth, newHeight];
}
// Draw a cropped image within specified bounds
function drawCroppedImage(img, dx, dy, dWidth, dHeight) {
if (!img) return;
let targetRatio = dWidth / dHeight;
if (node.images[0] && node.images[0].naturalWidth && node.images[0].naturalHeight) {
targetRatio = node.images[0].naturalWidth / node.images[0].naturalHeight;
if (
node.data.images[0] &&
node.data.images[0].naturalWidth &&
node.data.images[0].naturalHeight
) {
targetRatio =
node.data.images[0].naturalWidth /
node.data.images[0].naturalHeight;
}
const [ndx, ndy, ndWidth, ndHeight] = getFittedDestRect(dx, dy, dWidth, dHeight, targetRatio);
const [ndx, ndy, ndWidth, ndHeight] = getFittedDestRect(
dx,
dy,
dWidth,
dHeight,
targetRatio
);
const imgRatio = img.naturalWidth / img.naturalHeight;
let sx = 0, sy = 0, sWidth = img.naturalWidth, sHeight = img.naturalHeight;
let sx = 0,
sy = 0,
sWidth = img.naturalWidth,
sHeight = img.naturalHeight;
if (imgRatio > targetRatio) {
sWidth = img.naturalHeight * targetRatio;
sx = (img.naturalWidth - sWidth) / 2;
@@ -147,64 +176,139 @@ app.registerExtension({
sHeight = img.naturalWidth / targetRatio;
sy = (img.naturalHeight - sHeight) / 2;
}
ctx.drawImage(img, sx, sy, sWidth, sHeight, ndx, ndy, ndWidth, ndHeight);
ctx.drawImage(
img,
sx,
sy,
sWidth,
sHeight,
ndx,
ndy,
ndWidth,
ndHeight
);
}
const connectedImages = node.images.slice(1).filter(img => img !== null).length;
const connectedImages = node.data.images
.slice(1)
.filter((img) => img !== null).length;
if (connectedImages === 0) {
if (node.images[0]) {
drawCroppedImage(node.images[0], xStart, yStart, fullImgWidth, fullImgHeight);
// Single image display
if (node.data.images[0]) {
drawCroppedImage(
node.data.images[0],
xStart,
yStart,
fullImgWidth,
fullImgHeight
);
}
} else if (connectedImages === 1 && node.images[1]) {
const splitX = node.sliderX;
} else if (connectedImages === 1 && node.data.images[1]) {
// Two images with vertical split
const splitX = node.data.sliderX;
ctx.save();
ctx.beginPath();
ctx.rect(xStart, yStart, splitX - xStart, fullImgHeight);
ctx.clip();
drawCroppedImage(node.images[0], xStart, yStart, fullImgWidth, fullImgHeight);
drawCroppedImage(
node.data.images[0],
xStart,
yStart,
fullImgWidth,
fullImgHeight
);
ctx.restore();
ctx.save();
ctx.beginPath();
ctx.rect(splitX, yStart, xEnd - splitX, fullImgHeight);
ctx.clip();
drawCroppedImage(node.images[1], xStart, yStart, fullImgWidth, fullImgHeight);
drawCroppedImage(
node.data.images[1],
xStart,
yStart,
fullImgWidth,
fullImgHeight
);
ctx.restore();
} else {
// Three or four images with quadrants
const drawQuadrant = (imgIndex, clipX, clipY, clipW, clipH) => {
if (!node.images[imgIndex]) return;
if (!node.data.images[imgIndex]) return;
ctx.save();
ctx.beginPath();
ctx.rect(clipX, clipY, clipW, clipH);
ctx.clip();
drawCroppedImage(node.images[imgIndex], xStart, yStart, fullImgWidth, fullImgHeight);
drawCroppedImage(
node.data.images[imgIndex],
xStart,
yStart,
fullImgWidth,
fullImgHeight
);
ctx.restore();
};
drawQuadrant(0, xStart, yStart, node.sliderX - xStart, node.sliderY - yStart);
drawQuadrant(1, node.sliderX, yStart, xEnd - node.sliderX, node.sliderY - yStart);
if (node.images[3] === null) {
drawQuadrant(2, xStart, node.sliderY, xEnd - xStart, yEnd - node.sliderY);
drawQuadrant(
0,
xStart,
yStart,
node.data.sliderX - xStart,
node.data.sliderY - yStart
);
drawQuadrant(
1,
node.data.sliderX,
yStart,
xEnd - node.data.sliderX,
node.data.sliderY - yStart
);
if (node.data.images[3] === null) {
drawQuadrant(
2,
xStart,
node.data.sliderY,
xEnd - xStart,
yEnd - node.data.sliderY
);
} else {
drawQuadrant(2, xStart, node.sliderY, node.sliderX - xStart, yEnd - node.sliderY);
drawQuadrant(3, node.sliderX, node.sliderY, xEnd - node.sliderX, yEnd - node.sliderY);
drawQuadrant(
2,
xStart,
node.data.sliderY,
node.data.sliderX - xStart,
yEnd - node.data.sliderY
);
drawQuadrant(
3,
node.data.sliderX,
node.data.sliderY,
xEnd - node.data.sliderX,
yEnd - node.data.sliderY
);
}
}
// Draw sliders
ctx.strokeStyle = "#FFF";
ctx.lineWidth = 1;
if (connectedImages > 0) {
ctx.beginPath();
ctx.moveTo(node.sliderX, yStart);
ctx.lineTo(node.sliderX, yEnd);
if (node.data.images[3] === null && node.data.images[2] !== null) {
ctx.moveTo(node.data.sliderX, yStart);
ctx.lineTo(node.data.sliderX, node.data.sliderY);
} else {
ctx.moveTo(node.data.sliderX, yStart);
ctx.lineTo(node.data.sliderX, yEnd);
}
if (connectedImages >= 2) {
ctx.moveTo(xStart, node.sliderY);
ctx.lineTo(xEnd, node.sliderY);
ctx.moveTo(xStart, node.data.sliderY);
ctx.lineTo(xEnd, node.data.sliderY);
}
ctx.stroke();
}
};
},
});
});

47
web/js/load_text.js Normal file
View File

@@ -0,0 +1,47 @@
import { app } from "../../../scripts/app.js";
app.registerExtension({
name: "Bjornulf.LoadTextFromFolder",
async nodeCreated(node) {
if (node.comfyClass === "Bjornulf_LoadTextFromFolder") {
// Add a refresh button widget
// Assuming this is inside your node's setup function
const refreshButton = node.addWidget(
"button",
"Refresh File List",
null,
() => {
fetch("/get_text_files", {
method: "POST",
})
.then((response) => response.json())
.then((data) => {
if (data.success) {
// Update the dropdown with the new file list
node.widgets[0].options.values = data.files; // Assuming the dropdown is the first widget
node.widgets[0].value = data.files[0] || ""; // Set default value
app.ui.dialog.show(
"[LoadTextFromFolder] File list refreshed successfully!"
);
} else {
app.ui.dialog.show(
`[LoadTextFromFolder] Failed to refresh file list: ${
data.error || "Unknown error"
}`
);
}
})
.catch((error) => {
console.error(
"[LoadTextFromFolder] Error fetching text files:",
error
);
app.ui.dialog.show(
"[LoadTextFromFolder] An error occurred while refreshing the file list."
);
});
}
);
}
},
});

58
web/js/math_node.js Normal file
View File

@@ -0,0 +1,58 @@
import { app } from "../../../scripts/app.js";
app.registerExtension({
name: "Bjornulf.MathNode",
async nodeCreated(node) {
if (node.comfyClass === "Bjornulf_MathNode") {
const updateInputs = () => {
const initialWidth = node.size[0];
const numInputsWidget = node.widgets.find(w => w.name === "num_inputs");
if (!numInputsWidget) return;
const numInputs = numInputsWidget.value;
// Initialize node.inputs if it doesn't exist
if (!node.inputs) {
node.inputs = [];
}
// Filter existing value inputs
const existingInputs = node.inputs.filter(input => input.name.startsWith("value_"));
// Add new inputs if needed
if (existingInputs.length < numInputs) {
for (let i = existingInputs.length + 1; i <= numInputs; i++) {
const inputName = `value_${i}`;
if (!node.inputs.find(input => input.name === inputName)) {
node.addInput(inputName); // Type is defined in Python
}
}
}
// Remove excess inputs if too many
else {
node.inputs = node.inputs.filter(input =>
!input.name.startsWith("value_") ||
parseInt(input.name.split("_")[1]) <= numInputs
);
}
// Adjust node size while preserving width
node.setSize(node.computeSize());
node.size[0] = initialWidth;
};
// Ensure num_inputs widget is at the top and set its callback
const numInputsWidget = node.widgets.find(w => w.name === "num_inputs");
if (numInputsWidget) {
node.widgets = [numInputsWidget, ...node.widgets.filter(w => w !== numInputsWidget)];
numInputsWidget.callback = () => {
updateInputs();
app.graph.setDirtyCanvas(true);
};
}
// Perform initial input update after node creation
setTimeout(updateInputs, 0);
}
}
});

80
web/js/note_image.js
View File

@@ -1,44 +1,44 @@
import { app } from "../../../scripts/app.js";
// import { app } from "../../../scripts/app.js";
app.registerExtension({
name: "Bjornulf.ImageNoteLoadImage",
async nodeCreated(node) {
// Ensure the node is of the specific class
if (node.comfyClass !== "Bjornulf_ImageNoteLoadImage") return;
console.log("node created");
// app.registerExtension({
// name: "Bjornulf.ImageNoteLoadImage",
// async nodeCreated(node) {
// // Ensure the node is of the specific class
// if (node.comfyClass !== "Bjornulf_ImageNoteLoadImage") return;
// console.log("node created");
// Store the initial node size
let prevSize = [...node.size];
let stableCount = 0;
const minStableFrames = 3; // Number of frames the size must remain stable
// // Store the initial node size
// let prevSize = [...node.size];
// let stableCount = 0;
// const minStableFrames = 3; // Number of frames the size must remain stable
// Function to check if the node's size has stabilized
const checkSizeStable = () => {
if (node.size[0] === prevSize[0] && node.size[1] === prevSize[1]) {
stableCount++;
if (stableCount >= minStableFrames) {
// Size has been stable, simulate a resize to trigger layout update
const originalSize = [...node.size];
node.setSize([originalSize[0] + 1, originalSize[1]]); // Slightly increase width
setTimeout(() => {
node.setSize(originalSize); // Revert to original size
app.graph.setDirtyCanvas(true, true); // Trigger canvas redraw
}, 0);
} else {
// Size is stable but not for enough frames yet, check again
requestAnimationFrame(checkSizeStable);
}
} else {
// Size changed, reset counter and update prevSize
prevSize = [...node.size];
stableCount = 0;
requestAnimationFrame(checkSizeStable);
}
};
// // Function to check if the node's size has stabilized
// const checkSizeStable = () => {
// if (node.size[0] === prevSize[0] && node.size[1] === prevSize[1]) {
// stableCount++;
// if (stableCount >= minStableFrames) {
// // Size has been stable, simulate a resize to trigger layout update
// const originalSize = [...node.size];
// node.setSize([originalSize[0] + 1, originalSize[1]]); // Slightly increase width
// setTimeout(() => {
// node.setSize(originalSize); // Revert to original size
// app.graph.setDirtyCanvas(true, true); // Trigger canvas redraw
// }, 0);
// } else {
// // Size is stable but not for enough frames yet, check again
// requestAnimationFrame(checkSizeStable);
// }
// } else {
// // Size changed, reset counter and update prevSize
// prevSize = [...node.size];
// stableCount = 0;
// requestAnimationFrame(checkSizeStable);
// }
// };
// Start checking after a short delay to allow node initialization
setTimeout(() => {
requestAnimationFrame(checkSizeStable);
}, 5000);
}
});
// // Start checking after a short delay to allow node initialization
// setTimeout(() => {
// requestAnimationFrame(checkSizeStable);
// }, 5000);
// }
// });

8
web/js/ollama_talk.js
View File

@@ -7,10 +7,10 @@ app.registerExtension({
async nodeCreated(node) {
if (node.comfyClass === "Bjornulf_OllamaTalk") {
// Set seed widget to hidden input
const seedWidget = node.widgets.find((w) => w.name === "seed");
if (seedWidget) {
seedWidget.type = "HIDDEN";
}
// const seedWidget = node.widgets.find((w) => w.name === "seed");
// if (seedWidget) {
// seedWidget.type = "HIDDEN";
// }
// Function to update the Reset Button text
const updateResetButtonTextNode = () => {

186
web/js/style_selector.js Normal file
View File

@@ -0,0 +1,186 @@
// // style_selector.js
// import { app } from "../../../scripts/app.js";
// app.registerExtension({
// name: "Bjornulf.StyleSelector",
// async nodeCreated(node) {
// // Only apply to the Bjornulf_StyleSelector node
// if (node.comfyClass !== "Bjornulf_StyleSelector") return;
// // Find the widgets for LOOP_random_LIST and LOOP_style_LIST
// const loopRandomWidget = node.widgets.find(w => w.name === "LOOP_random_LIST");
// const loopStyleWidget = node.widgets.find(w => w.name === "LOOP_style_LIST");
// // Check if widgets exist to avoid errors
// if (!loopRandomWidget || !loopStyleWidget) {
// console.error("LOOP widgets not found in Bjornulf_StyleSelector node.");
// return;
// }
// // Function to toggle the other widget off when one is turned on
// const toggleExclusive = (widgetToToggle, otherWidget) => {
// if (widgetToToggle.value === true) {
// otherWidget.value = false;
// }
// };
// // Add event listeners to handle toggling
// loopRandomWidget.callback = () => {
// toggleExclusive(loopRandomWidget, loopStyleWidget);
// };
// loopStyleWidget.callback = () => {
// toggleExclusive(loopStyleWidget, loopRandomWidget);
// };
// // Ensure initial state has only one enabled (if both are true, disable one)
// if (loopRandomWidget.value && loopStyleWidget.value) {
// loopStyleWidget.value = false; // Default to disabling LOOP_style_LIST
// }
// // Find the category and style widgets
// const categoryWidget = node.widgets.find(w => w.name === "category");
// const styleWidget = node.widgets.find(w => w.name === "style");
// // Define categories and styles (must match the Python file)
// const BRANCHES = {
// "Painting": [
// "Renaissance", "Baroque", "Rococo", "Neoclassicism",
// "Romanticism", "Realism", "Impressionism", "Post-Impressionism",
// "Expressionism", "Fauvism", "Cubism", "Futurism", "Dadaism",
// "Surrealism", "Abstract Expressionism", "Pop Art", "Op Art",
// "Minimalism"
// ],
// "Photography": [
// "Black and White", "Color", "Vintage", "Sepia Tone", "HDR",
// "Long Exposure", "Macro", "Portrait", "Landscape", "Street",
// "Fashion", "Analog Film", "Cinematic"
// ],
// "Digital Art": [
// "Digital Painting", "Vector Art", "Pixel Art", "Fractal Art",
// "Algorithmic Art", "Glitch Art"
// ],
// "3D Rendering": [
// "Low Poly", "Voxel", "Isometric", "Ray Tracing"
// ],
// "Illustration": [
// "Line Art", "Cartoon", "Comic Book", "Manga", "Anime",
// "Technical Illustration", "Botanical Illustration",
// "Architectural Rendering", "Concept Art", "Storyboard Art"
// ],
// };
// // Function to update the style dropdown based on the selected category
// const updateStyles = () => {
// const selectedCategory = categoryWidget.value;
// const styles = BRANCHES[selectedCategory] || [];
// styleWidget.options.values = styles;
// if (styles.length > 0) {
// styleWidget.value = styles[0]; // Set to the first style
// } else {
// styleWidget.value = ""; // Fallback if no styles
// }
// node.setDirtyCanvas(true); // Refresh the UI
// };
// // Initialize the style dropdown
// updateStyles();
// // Update the style dropdown whenever the category changes
// categoryWidget.callback = updateStyles;
// }
// });
import { app } from "../../../scripts/app.js";
app.registerExtension({
name: "Bjornulf.StyleSelector",
async nodeCreated(node) {
if (node.comfyClass !== "Bjornulf_StyleSelector") return;
// Find loop widgets
const loopRandomWidget = node.widgets.find(w => w.name === "LOOP_random_LIST");
const loopStyleWidget = node.widgets.find(w => w.name === "LOOP_style_LIST");
const loopSequentialWidget = node.widgets.find(w => w.name === "LOOP_SEQUENTIAL");
// Function to toggle the other widget off when one is turned on
const toggleExclusive = (widgetToToggle, otherWidget) => {
if (widgetToToggle.value === true) {
otherWidget.value = false;
}
};
// Add event listeners to handle toggling
loopRandomWidget.callback = () => {
toggleExclusive(loopRandomWidget, loopStyleWidget);
};
loopStyleWidget.callback = () => {
toggleExclusive(loopStyleWidget, loopRandomWidget);
};
// Ensure initial state has only one enabled (if both are true, disable one)
if (loopRandomWidget.value && loopStyleWidget.value) {
loopStyleWidget.value = false; // Default to disabling LOOP_style_LIST
}
// Add reset button for style list counter
const styleResetButton = node.addWidget(
"button",
"Reset Style Counter",
null,
async () => {
try {
const response = await fetch("/reset_style_list_counter", { method: "POST" });
const data = await response.json();
if (data.success) {
app.ui.dialog.show("[Style Selector] Style counter reset successfully.");
} else {
app.ui.dialog.show("[Style Selector] Failed to reset style counter.");
}
} catch (error) {
app.ui.dialog.show("[Style Selector] Error resetting style counter.");
}
}
);
// Add reset button for model list counter
const modelResetButton = node.addWidget(
"button",
"Reset Model Counter",
null,
async () => {
try {
const response = await fetch("/reset_model_list_counter", { method: "POST" });
const data = await response.json();
if (data.success) {
app.ui.dialog.show("[Style Selector] Model counter reset successfully.");
} else {
app.ui.dialog.show("[Style Selector] Failed to reset model counter.");
}
} catch (error) {
app.ui.dialog.show("[Style Selector] Error resetting model counter.");
}
}
);
// Function to update visibility of reset buttons
const updateButtonVisibility = () => {
const sequentialEnabled = loopSequentialWidget.value;
styleResetButton.type = sequentialEnabled && loopStyleWidget.value ? "button" : "hidden";
modelResetButton.type = sequentialEnabled && loopRandomWidget.value ? "button" : "hidden";
};
// Initial update of button visibility
setTimeout(updateButtonVisibility, 0);
// Update visibility when widgets change
loopSequentialWidget.callback = updateButtonVisibility;
loopStyleWidget.callback = () => {
toggleExclusive(loopStyleWidget, loopRandomWidget);
updateButtonVisibility();
};
loopRandomWidget.callback = () => {
toggleExclusive(loopRandomWidget, loopStyleWidget);
updateButtonVisibility();
};
}
});

3
web/js/tts_config.js
View File

@@ -31,7 +31,8 @@ app.registerExtension({
// Function to update voices based on selected language
const updateVoicesForLanguage = async (selectedLanguage) => {
try {
const response = await fetch('/bjornulf_xtts_get_voices', {
// const response = await fetch('/bjornulf_xtts_get_voices', {
const response = await fetch('/bjornulf_TTS_get_voices', {
method: 'POST',
headers: {
'Content-Type': 'application/json',

72
web/js/video_preview.js
View File

@@ -1,12 +1,16 @@
import { api } from '../../../scripts/api.js';
import { app } from "../../../scripts/app.js";
function displayVideoPreview(component, filename, category, autoplay, mute) {
// Function to display the video preview
function displayVideoPreview(component, filename, category, autoplay, mute, loop) {
let videoWidget = component._videoWidget;
// Create the video widget if it doesn't exist
if (!videoWidget) {
// Create the widget if it doesn't exist
var container = document.createElement("div");
const container = document.createElement("div");
const currentNode = component;
// Add the DOM widget to the component
videoWidget = component.addDOMWidget("videopreview", "preview", container, {
serialize: false,
hideOnZoom: false,
@@ -17,73 +21,83 @@ function displayVideoPreview(component, filename, category, autoplay, mute) {
container.value = v;
},
});
// Define how the widget computes its size
videoWidget.computeSize = function(width) {
if (this.aspectRatio && !this.parentElement.hidden) {
let height = (currentNode.size[0] - 20) / this.aspectRatio + 10;
if (!(height > 0)) {
height = 0;
}
return [width, height];
return [width, height > 0 ? height : 0];
}
return [width, -4];
};
// Initialize widget properties
videoWidget.value = { hidden: false, paused: false, params: {} };
videoWidget.parentElement = document.createElement("div");
videoWidget.parentElement.className = "video_preview";
videoWidget.parentElement.style['width'] = "100%";
videoWidget.parentElement.style.width = "100%";
container.appendChild(videoWidget.parentElement);
// Create the video element
videoWidget.videoElement = document.createElement("video");
videoWidget.videoElement.controls = true;
videoWidget.videoElement.loop = false;
videoWidget.videoElement.muted = false;
videoWidget.videoElement.style['width'] = "100%";
videoWidget.videoElement.style.width = "100%";
// Update aspect ratio when metadata is loaded
videoWidget.videoElement.addEventListener("loadedmetadata", () => {
videoWidget.aspectRatio = videoWidget.videoElement.videoWidth / videoWidget.videoElement.videoHeight;
adjustSize(component);
});
// Hide the video on error
videoWidget.videoElement.addEventListener("error", () => {
videoWidget.parentElement.hidden = true;
adjustSize(component);
});
videoWidget.parentElement.hidden = videoWidget.value.hidden;
videoWidget.videoElement.autoplay = !videoWidget.value.paused && !videoWidget.value.hidden;
videoWidget.videoElement.hidden = false;
videoWidget.parentElement.appendChild(videoWidget.videoElement);
component._videoWidget = videoWidget; // Store the widget for future reference
component._videoWidget = videoWidget; // Store for reuse
}
// Update the video source
let params = {
// Set video source and properties
const params = {
"filename": filename,
"subfolder": category,
"type": "output",
"rand": Math.random().toString().slice(2, 12)
"rand": Math.random().toString().slice(2, 12) // Cache-busting random parameter
};
const urlParams = new URLSearchParams(params);
if(mute) videoWidget.videoElement.muted = true;
else videoWidget.videoElement.muted = false;
if(autoplay) videoWidget.videoElement.autoplay = !videoWidget.value.paused && !videoWidget.value.hidden;
else videoWidget.videoElement.autoplay = false;
// videoWidget.videoElement.src = `http://localhost:8188/api/view?${urlParams.toString()}`;
videoWidget.videoElement.src = `api/view?${urlParams.toString()}`;
adjustSize(component); // Adjust the component size
const urlParams = new URLSearchParams(params);
videoWidget.videoElement.src = `api/view?${urlParams.toString()}`;
videoWidget.videoElement.muted = mute;
videoWidget.videoElement.autoplay = autoplay && !videoWidget.value.paused && !videoWidget.value.hidden;
videoWidget.videoElement.loop = loop;
// Adjust the component size after setting the video
adjustSize(component);
}
// Function to adjust the component size
function adjustSize(component) {
component.setSize([component.size[0], component.computeSize([component.size[0], component.size[1]])[1]]);
const newSize = component.computeSize([component.size[0], component.size[1]]);
component.setSize([component.size[0], newSize[1]]);
component?.graph?.setDirtyCanvas(true);
}
// Register the extension
app.registerExtension({
name: "Bjornulf.VideoPreview",
async beforeRegisterNodeDef(nodeType, nodeData, appInstance) {
if (nodeData?.name == "Bjornulf_VideoPreview") {
nodeType.prototype.onExecuted = function (data) {
if (nodeData?.name === "Bjornulf_VideoPreview") {
nodeType.prototype.onExecuted = function(data) {
// Retrieve widget values with defaults
const autoplay = this.widgets.find(w => w.name === "autoplay")?.value ?? false;
const mute = this.widgets.find(w => w.name === "mute")?.value ?? true;
displayVideoPreview(this, data.video[0], data.video[1], autoplay, mute);
const loop = this.widgets.find(w => w.name === "loop")?.value ?? false;
// Display the video preview with the retrieved values
displayVideoPreview(this, data.video[0], data.video[1], autoplay, mute, loop);
};
}
}

156
write_text_advanced.py
View File

@@ -5,8 +5,12 @@ import csv
from itertools import cycle
#{red|blue}
#{left|right|middle|group=LR}+{left|right|middle|group=LR}+{left|right|middle|group=LR}
#{left|right|middle|group=LMR}+{left|right|middle|group=LMR}+{left|right|middle|group=LMR}
#{A(80%)|B(15%)|C(5%)}
#2 {apple|orange|banana|static_group=FRUIT}s, one {apple|orange|banana|static_group=FRUIT} on the left, one {apple|orange|banana|static_group=FRUIT} on the right
#Double layer variable : <<CHAR>>
#CHAR = JESSICA
#JESSICA = en/jess.wav
class WriteTextAdvanced:
@classmethod
@@ -28,6 +32,7 @@ class WriteTextAdvanced:
CATEGORY = "Bjornulf"
def find_variables(self, text):
"""Identify nested curly brace sections in the text."""
stack = []
variables = []
for i, char in enumerate(text):
@@ -40,10 +45,12 @@ class WriteTextAdvanced:
'end': i + 1,
'nesting': nesting
})
# Sort by descending nesting level and position to process inner variables first
variables.sort(key=lambda x: (-x['nesting'], -x['end']))
return variables
def parse_option(self, part):
"""Parse options within curly braces, handling CSV and weighted choices."""
if part.startswith('%csv='):
try:
filename = part.split('=', 1)[1].strip()
@@ -56,49 +63,66 @@ class WriteTextAdvanced:
return (option.strip(), float(weight.split('%)')[0]))
return part.strip()
def process_content(self, content, seed):
random.seed(seed)
parts = []
def process_content(self, content, base_seed, position):
"""Process content within curly braces, handling groups and random choices."""
# Use position to vary the seed for independent random choices
random.seed(base_seed + position)
parts = [p.strip() for p in content.split('|')]
options = []
weights = []
group_defined = False
group_name = None
static_group = None
cycling_group = None
for p in content.split('|'):
p = p.strip()
if p.startswith('group='):
group_name = p.split('=', 1)[1].strip()
group_defined = True
continue
parsed = self.parse_option(p)
if isinstance(parsed, list): # CSV data
parts.extend(parsed)
weights.extend([1]*len(parsed))
elif isinstance(parsed, tuple): # Weighted option
parts.append(parsed[0])
weights.append(parsed[1])
for p in parts:
if p.startswith('static_group='):
static_group = p.split('=', 1)[1].strip()
elif p.startswith('group='):
cycling_group = p.split('=', 1)[1].strip()
else:
parts.append(parsed)
weights.append(1)
if group_defined:
return {'type': 'group', 'name': group_name, 'options': parts}
if any(w != 1 for w in weights):
total = sum(weights)
if total == 0: weights = [1]*len(parts)
return random.choices(parts, weights=[w/total for w in weights])[0]
return random.choice(parts) if parts else ''
parsed = self.parse_option(p)
if isinstance(parsed, list): # CSV data
options.extend(parsed)
weights.extend([1] * len(parsed))
elif isinstance(parsed, tuple): # Weighted option
options.append(parsed[0])
weights.append(parsed[1])
else:
options.append(parsed)
weights.append(1)
if static_group and cycling_group:
raise ValueError("Cannot specify both static_group and group in the same curly brace section.")
if static_group:
return {'type': 'static_group', 'name': static_group, 'options': options, 'weights': weights}
elif cycling_group:
return {'type': 'cycling_group', 'name': cycling_group, 'options': options, 'weights': weights}
else:
if options:
if any(w != 1 for w in weights):
total = sum(weights)
if total == 0:
weights = [1] * len(options)
return random.choices(options, weights=[w / total for w in weights])[0]
else:
return random.choice(options)
return ''
def write_text_special(self, text, variables="", seed=None):
"""Main function to process text with special syntax."""
if seed is None or seed == 0:
seed = int(time.time() * 1000)
random.seed(seed)
# Handle variables
# Process text: remove comments and empty lines
text_lines = [line.strip() for line in text.split('\n')
if line.strip() and not line.strip().startswith('#')]
text = '\n'.join(text_lines)
# Replace predefined variables
var_dict = {}
for line in variables.split('\n'):
var_lines = [line.strip() for line in variables.split('\n')
if line.strip() and not line.strip().startswith('#')]
for line in var_lines:
if '=' in line:
key, value = line.split('=', 1)
var_dict[key.strip()] = value.strip()
@@ -107,23 +131,36 @@ class WriteTextAdvanced:
# Process nested variables
variables = self.find_variables(text)
static_groups = {}
cycling_groups = {}
substitutions = []
groups = {}
for var in variables:
start, end = var['start'], var['end']
content = text[start+1:end-1]
processed = self.process_content(content, seed)
content = text[start + 1:end - 1]
# Pass the position (start) to vary the seed for non-group choices
processed = self.process_content(content, seed, start)
if isinstance(processed, dict):
if processed['type'] == 'group':
group_name = processed['name']
if group_name not in groups:
groups[group_name] = []
groups[group_name].append({
if processed['type'] == 'static_group':
name = processed['name']
if name not in static_groups:
static_groups[name] = []
static_groups[name].append({
'start': start,
'end': end,
'options': processed['options']
'options': processed['options'],
'weights': processed['weights']
})
elif processed['type'] == 'cycling_group':
name = processed['name']
if name not in cycling_groups:
cycling_groups[name] = []
cycling_groups[name].append({
'start': start,
'end': end,
'options': processed['options'],
'weights': processed['weights']
})
else:
substitutions.append({
@@ -132,15 +169,35 @@ class WriteTextAdvanced:
'sub': processed
})
# Handle groups
for group_name, matches in groups.items():
# Handle static groups: choose one value per group name
random.seed(seed) # Reset seed for consistent static group behavior
for name, matches in static_groups.items():
if not matches or not matches[0]['options']:
continue
options = matches[0]['options']
weights = matches[0]['weights']
if any(w != 1 for w in weights):
total = sum(weights)
if total == 0:
weights = [1] * len(options)
chosen = random.choices(options, weights=[w / total for w in weights])[0]
else:
chosen = random.choice(options) if options else ''
for m in matches:
substitutions.append({
'start': m['start'],
'end': m['end'],
'sub': chosen
})
# Handle cycling groups: cycle through shuffled options
random.seed(seed) # Reset seed for consistent cycling group behavior
for name, matches in cycling_groups.items():
if not matches or not matches[0]['options']:
continue
options = matches[0]['options']
permuted = random.sample(options, len(options))
perm_cycle = cycle(permuted)
for m in matches:
substitutions.append({
'start': m['start'],
@@ -148,7 +205,7 @@ class WriteTextAdvanced:
'sub': next(perm_cycle)
})
# Apply regular substitutions
# Apply substitutions in reverse order
substitutions.sort(key=lambda x: -x['start'])
result_text = text
for sub in substitutions:
@@ -158,4 +215,5 @@ class WriteTextAdvanced:
@classmethod
def IS_CHANGED(s, text, variables="", seed=None):
"""Check if inputs have changed."""
return (text, variables, seed)