 GiusTex
|
bc7c1ded58
|
made available comfyui clip
No more need to download tokenizers and text encoders, now we can use comfyui clip loader
|
2024-10-20 20:38:46 +02:00 |
|
|
GiusTex
|
a5e413b575
|
fixed cuda not available on device
|
2024-10-18 09:05:41 +02:00 |
|
|
GiusTex
|
60f445cf6f
|
added sequential_cpu_offload
It's low vram friendly!
|
2024-10-06 21:07:49 +02:00 |
|
|
GiusTex
|
a8d80e9fd9
|
fix new_width not found
|
2024-10-06 13:45:04 +02:00 |
|
|
GiusTex
|
43bf24a891
|
try fixing cannot access local variable 'new_width'
|
2024-10-05 15:59:23 +02:00 |
|
|
GiusTex
|
34c1bdad6e
|
quick fix when updating the prompt
text_encoders and tokenizers are not deleted when encoding prompt, and deleting them that way didn't free much vram anyway
|
2024-10-05 15:05:55 +02:00 |
|
|
GiusTex
|
68b87e79c1
|
Fix some tooltip typos
|
2024-09-29 01:50:03 +02:00 |
|
|
GiusTex
|
63fede2695
|
Updates and vram optimization tries
- Added Encode Diffusers Prompt node.
- Tried removing text_encoders and tokenizers before sampling
|
2024-09-29 01:26:52 +02:00 |
|
|
GiusTex
|
b1fadc411f
|
use diffusion_models folder and don't wait to remove "state_dict, model_file"
|
2024-09-28 15:49:18 +02:00 |
|
|
GiusTex
|
69e515c2b4
|
fix model files not stored in node chache when deleting them
|
2024-09-28 12:33:09 +02:00 |
|
|
GiusTex
|
c5e3c8cf14
|
fix delete vae error
|
2024-09-28 12:11:05 +02:00 |
|
|
GiusTex
|
227b3ae926
|
fix delete vae error
|
2024-09-28 12:09:10 +02:00 |
|
|
GiusTex
|
0b4d11cb15
|
fixed image output with 4 channels instead of 3 in diffusers image outpaint
|
2024-09-28 10:42:08 +02:00 |
|
|
GiusTex
|
245b61ad3a
|
Add files via upload
|
2024-09-26 23:45:36 +02:00 |
|