alphard/stable-diffusion.cpp
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix: improve VAE tiling (#372)

Browse Source 
* fix and improve: VAE tiling
- properly handle the upper left corner interpolating both x and y
- refactor out lerp
- use smootherstep to preserve more detail and spend less area blending

* actually fix vae tile merging

Co-authored-by: stduhpf <stephduh@live.fr>

* remove the now unused lerp function

---------

Co-authored-by: stduhpf <stephduh@live.fr>
This commit is contained in:
Erik Scholz 2024-08-27 18:21:12 +02:00 committed by GitHub
parent f4c937cb94
commit e71ddcedad
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 26 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

35
ggml_extend.hpp
View File

@ -353,6 +353,12 @@ __STATIC_INLINE__ void ggml_split_tensor_2d(struct ggml_tensor* input,
}
}
// unclamped -> expects x in the range [0-1]
__STATIC_INLINE__ float ggml_smootherstep_f32(const float x) {
GGML_ASSERT(x >= 0.f && x <= 1.f);
return x * x * x * (x * (6.0f * x - 15.0f) + 10.0f);
}
__STATIC_INLINE__ void ggml_merge_tensor_2d(struct ggml_tensor* input,
struct ggml_tensor* output,
int x,
@ -361,6 +367,10 @@ __STATIC_INLINE__ void ggml_merge_tensor_2d(struct ggml_tensor* input,
int64_t width = input->ne[0];
int64_t height = input->ne[1];
int64_t channels = input->ne[2];
int64_t img_width = output->ne[0];
int64_t img_height = output->ne[1];
GGML_ASSERT(input->type == GGML_TYPE_F32 && output->type == GGML_TYPE_F32);
for (int iy = 0; iy < height; iy++) {
for (int ix = 0; ix < width; ix++) {
@ -368,19 +378,26 @@ __STATIC_INLINE__ void ggml_merge_tensor_2d(struct ggml_tensor* input,
float new_value = ggml_tensor_get_f32(input, ix, iy, k);
if (overlap > 0) { // blend colors in overlapped area
float old_value = ggml_tensor_get_f32(output, x + ix, y + iy, k);
if (x > 0 && ix < overlap) { // in overlapped horizontal
ggml_tensor_set_f32(output, old_value + (new_value - old_value) * (ix / (1.0f * overlap)), x + ix, y + iy, k);
continue;
}
if (y > 0 && iy < overlap) { // in overlapped vertical
ggml_tensor_set_f32(output, old_value + (new_value - old_value) * (iy / (1.0f * overlap)), x + ix, y + iy, k);
continue;
}
}
const float x_f_0 = (x > 0) ? ix / float(overlap) : 1;
const float x_f_1 = (x < (img_width - width)) ? (width - ix) / float(overlap) : 1 ;
const float y_f_0 = (y > 0) ? iy / float(overlap) : 1;
const float y_f_1 = (y < (img_height - height)) ? (height - iy) / float(overlap) : 1;
const float x_f = std::min(std::min(x_f_0, x_f_1), 1.f);
const float y_f = std::min(std::min(y_f_0, y_f_1), 1.f);
ggml_tensor_set_f32(
output,
old_value + new_value * ggml_smootherstep_f32(y_f) * ggml_smootherstep_f32(x_f),
x + ix, y + iy, k
);
} else {
ggml_tensor_set_f32(output, new_value, x + ix, y + iy, k);
}
}
}
}
}
__STATIC_INLINE__ float ggml_tensor_mean(struct ggml_tensor* src) {