nodes/kijai-ComfyUI-SUPIR
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
150 Commits 2 Branches 0 Tags
d5b69963db96e7b363dce9d7384975f82989a91b
Commit Graph
150 Commits
This Branch
This Branch
All Branches
Author SHA1 Message Date
snomiao d5b69963db chore(publish): update GitHub Actions workflow for node publishing 
- Add permissions for issue writing
- Update action version to v1 for publish-node-action
- Add condition to run job only for 'kijai' repository owner
 2025-01-20 21:36:40 +00:00
kijai 53fc4f82f1 clean some unnecessary dependencies 2024-07-07 02:44:51 +03:00
kijai 0597c9cdb6 Update pyproject.toml 2024-07-06 12:23:56 +03:00
kijai 59ca341c0a workaround for open-clip-torch update 
fixes the attn_mask shape error
 2024-07-06 12:22:30 +03:00
kijai a8780cd188 use more comfy ops 2024-07-06 12:20:21 +03:00
Jukka Seppänen 2726b6ef34 Update requirements.txt 2024-07-06 11:52:45 +03:00
kijai 006754633a Support SDXL latents directly 
Bypassing need for SUPIR VAE
 2024-06-27 16:40:45 +03:00
Jukka Seppänen c257cce555 Merge pull request #124 from haohaocreates/publish 
Add Github Action for Publishing to Comfy Registry
 2024-06-27 12:29:20 +03:00
kijai eda2cf504a Update pyproject.toml 2024-06-27 12:28:18 +03:00
Jukka Seppänen e1770b002a Merge pull request #125 from haohaocreates/pyproject 
Add pyproject.toml for Custom Node Registry
 2024-06-27 12:27:28 +03:00
haohaocreates 3320502310 Update pyproject.toml description 2024-05-21 10:56:33 -04:00
haohaocreates 91893f123d chore(pyproject): Add pyproject.toml for Custom Node Registry 2024-05-21 14:46:39 +00:00
haohaocreates 1db91b1510 chore(publish): Add Github Action for Publishing to Comfy Registry 2024-05-21 14:46:38 +00:00
kijai bbb2bd7e8f Add version of the loader node to load CLIP separately 2024-05-20 11:12:07 +03:00
Kijai 5518e00a14 Remove redundant req 2024-04-23 13:04:10 +03:00
kijai 656e55e815 accelerate fixes 2024-04-14 11:27:17 +03:00
kijai c97893850e Make accelerate model loading target choseable 2024-04-14 02:01:54 +03:00
kijai 7596566a89 Accelerate 2024-04-13 17:48:47 +03:00
kijai 49afedaddd Description for sampler 2024-04-08 21:10:46 +03:00
kijai 3a911c0cbd Merge branch 'main' of https://github.com/kijai/ComfyUI-SUPIR 2024-04-07 02:43:25 +03:00
kijai 84e67594d0 Initial DESCRIPTIONs 2024-04-07 02:43:23 +03:00
Jukka Seppänen 472dbce4dc Update README.md 2024-04-07 02:34:15 +03:00
kijai d20c5190a1 Update nodes_v2.py 2024-04-05 10:59:10 +03:00
kijai 30f60a3b03 clip decode on gpu 2024-04-05 10:51:32 +03:00
Jukka Seppänen e62f3c5dac Create LICENSE 2024-03-30 00:29:23 +02:00
Jukka Seppänen 91ecd6f8de Delete LICENSE 2024-03-30 00:28:57 +02:00
Jukka Seppänen ba7fceeaac Update LICENSE 2024-03-30 00:28:33 +02:00
kijai 2ce3ced581 Tiled sampler consequential runs workaround 
Copy the conds so they are not permanently modified by the Tiled samplers
 2024-03-26 10:34:17 +02:00
Kijai 3f340a6b96 update example 2024-03-25 18:12:55 +02:00
Kijai 203bc93933 Update nodes_v2.py 2024-03-25 18:06:11 +02:00
Kijai eb7d1e31ae Update nodes_v2.py 2024-03-25 17:54:24 +02:00
Kijai 61cbb153e6 Update nodes_v2.py 2024-03-25 17:47:48 +02:00
kijai 5c54bb1437 Better xformers check 
Now relies in the comfy xformers check and can thus be disabled with the command line argument --disable-xformers
 2024-03-25 10:46:36 +02:00
kijai 2cd64cdfce cleanup 2024-03-25 10:17:45 +02:00
kijai 80d20731a4 Update nodes_v2.py 2024-03-25 01:29:31 +02:00
kijai dced82b070 Smarter model loading and support comfy latents 
- Less spiky memory usage when using v2 model loader node
- Support latents from other comfy nodes, when using latents from SUPIR nodes the behaviour is same as before: fully random noise
 2024-03-24 15:45:10 +02:00
kijai e933093166 fix decoder dtype 2024-03-22 08:48:48 +02:00
kijai fb6f99e7d5 clip decoder output 2024-03-22 08:37:04 +02:00
Jukka Seppänen 4a10dd4da5 Update README.md 2024-03-22 00:40:05 +02:00
kijai adc82313eb Make model loader node without sdxl_model selection for clarity 2024-03-22 00:37:21 +02:00
Kijai 61942e250c Support using comfy model loading, including loras 
These inputs are optional, when used the sdxl_model dropdown is ignored, I don't want to remove it as of yet as it would break old workflows.
 2024-03-21 16:06:08 +02:00
Kijai 98291ac559 Update nodes_v2.py 2024-03-20 18:45:46 +02:00
Kijai d7ab37bed0 Update nodes_v2.py 2024-03-20 18:42:47 +02:00
Kijai 14480f926a Default to fp16 unet with 'auto' 2024-03-19 16:25:54 +02:00
kijai dbc22b88fb batch fixes 2024-03-19 11:02:54 +02:00
kijai 734054ea64 Update nodes_v2.py 2024-03-19 10:19:32 +02:00
Kijai e460953d7c Fix resize and batch captioning 2024-03-18 17:44:14 +02:00
Kijai 033f83ad17 Batch support for video 2024-03-18 16:49:57 +02:00
kijai 2bd6421b6f Flip cfg linear scale behaviour for v2 sampler 
Makes more sense to me, scales cfg linearly from cfg_scale_start at first step to  cfg_scale_end at last
 2024-03-18 01:48:19 +02:00
kijai b887b9e4be clean prints 2024-03-17 17:55:24 +02:00