Commit Graph

7779 Commits

Author SHA1 Message Date
lllyasviel d6f2e5bdd9 Disable Legacy Multidiffusion
Both Multidiffusion and Tiled VAE are integrated and this extension can be disabled
2024-02-09 19:07:50 -08:00
lllyasviel dca0a1d5d8 Integrate Multi Diffusion
Integrated Multi Diffusion, based on codes from pkuliyi2015 and shiimizu
Basic Diffusion test passed
ControlNet test passed
2024-02-09 18:37:32 -08:00
lllyasviel 4ec1015162 fix mask can be none 2024-02-09 18:25:44 -08:00
lllyasviel fb2e271668 support inpaint models from fooocus
put inpaint_v26.fooocus.patch in models\ControlNet, control SDXL models only
To get same algorithm as Fooocus, set "Stop at" (Ending Control Step) to 0.5
Fooocus always use 0.5 but in Forge users may use other values.
Results are best when stop at < 0.7. The model is not optimized with ending timesteps > 0.7
Supports inpaint_global_harmonious, inpaint_only, inpaint_only+lama.
In theory the inpaint_only+lama always outperform Fooocus in object removal task (but not all tasks).
2024-02-09 17:08:48 -08:00
lllyasviel d2af6d1b44 Contribution Guideline 2024-02-09 14:57:45 -08:00
lllyasviel 54edd29725 integrate latent modifier 2024-02-09 14:38:57 -08:00
lllyasviel 8c8f948666 integrated DynamicThresholding (CFG-Fix) 2024-02-09 14:11:59 -08:00
lllyasviel 8059533eaf fix lora not loaded for text encoder #142 2024-02-09 13:50:50 -08:00
Chenlei Hu 200f2b69ed Add back ControlNet model version filter (#131)
* Add back ControlNet model version filter

* Update choice after sd model changes
2024-02-09 16:34:09 -05:00
Chenlei Hu ac4a8820a5 Fix CQ tests (#141)
* Make test client run on cpu

* test on cpu

try fix device

try fix device

try fix device

* Use real SD1.5 model for testing

* ckpt nits

* Remove coverage calls
2024-02-09 16:33:05 -05:00
lllyasviel 472a510151 try fix control mode weighting #155 2024-02-09 13:23:49 -08:00
Andray e8f51579cc add button for refreshing extensions list 2024-02-09 13:12:14 -08:00
Chenlei Hu c06769c1fa Temporarily disable CQ unittests (#154) 2024-02-09 12:11:11 -05:00
Chenlei Hu ed60a99826 Fix import error (#146) 2024-02-08 23:02:33 -05:00
Chenlei Hu 61db0fba41 Ignore ruff directorys (#145) 2024-02-08 21:55:21 -05:00
Chenlei Hu 3c32cbb0af Revert "Fix more ruff lint (#139)" (#144)
This reverts commit e13072cb42.
2024-02-08 21:34:58 -05:00
Chenlei Hu 388ca351f4 Revert "Fix ruff linter (#137)" (#143)
This reverts commit 6b3ad64388.
2024-02-08 21:24:04 -05:00
Chenlei Hu b49742354d Fix CQ server arg (#140) 2024-02-08 20:50:03 -05:00
Chenlei Hu e13072cb42 Fix more ruff lint (#139) 2024-02-08 20:41:38 -05:00
Chenlei Hu 6b3ad64388 Fix ruff linter (#137)
* Fix ruff linter

* Remove unused imports

* Remove unused imports
2024-02-08 20:35:20 -05:00
Chenlei Hu 66c22490c3 Ignore ControlNet extension in eslint (#136) 2024-02-08 19:48:49 -05:00
Yuki Shindo 847d451505 Fix command line arguments format in webui-user.bat (#135)
* fix command line arguments format in webui-user.bat

* fix embeddings dir path
2024-02-08 19:47:42 -05:00
lllyasviel f06ba8e60b Significantly reduce thread abuse for faster model moving
This will move all major gradio calls into the main thread rather than random gradio threads.
This ensures that all torch.module.to() are performed in main thread to completely possible avoid GPU fragments.
In my test now model moving is 0.7 ~ 1.2 seconds faster, which means all 6GB/8GB VRAM users will get 0.7 ~ 1.2 seconds faster per image on SDXL.
2024-02-08 10:13:59 -08:00
lllyasviel 291ec743b6 use better context manager to fix potential problems 2024-02-08 02:00:54 -08:00
lllyasviel 760f727eb9 use better context manager to fix potential problems 2024-02-08 01:51:18 -08:00
lllyasviel 4c9db26541 support controlnet_model_function_wrapper for t2i-adapter 2024-02-07 21:42:56 -08:00
lllyasviel 50035ad414 fix outpaint with inpaint_global_harmonious 2024-02-07 20:31:22 -08:00
lllyasviel 49ec325f6a lin 2024-02-07 20:14:02 -08:00
lllyasviel a1670c536d Allow controlnet_model_function_wrapper
for animatediff to manage controlnet batch slicing window
2024-02-07 20:06:23 -08:00
Chenlei Hu 383aaca1eb Improve model filtering (#114) 2024-02-07 22:53:41 -05:00
lllyasviel 42dd258c8d revise attention alignment in stylealign #100
A mistake in 0day release is that the attention layers of cond and uncond items in a batch are aligned when they should not.
after align batch in cond and uncond separately they now works and give same results to legacy sd-webui-cnet
2024-02-07 19:11:53 -08:00
lllyasviel f63917a323 add codes discussed in #73 that may help ipadapters 2024-02-07 18:57:57 -08:00
lllyasviel ef781cabcb Backend: Allow control signal to be none for advanced weighting 2024-02-07 13:02:42 -08:00
lllyasviel c3a66b016b try solve dtype cast for #112 2024-02-07 12:39:55 -08:00
Chenlei Hu e1faf8327b Add back ControlNet HR option (#90)
* Add back ControlNet HR option

* nits

* protect kernel

* add enum

* fix

* fix

* Update controlnet.py

* restore controlnet.py

* hint ui

* Update controlnet.py

* fix

* Update controlnet.py

* Backend: better controlnet mask batch broadcasting

* Update README.md

* fix inpaint batch dim align #94

* fix sigmas device in rare cases #71

* rework sigma device mapping

* Add hr_option to infotext

---------

Co-authored-by: lllyasviel <lyuminzhang@outlook.com>
2024-02-07 11:09:52 -05:00
lllyasviel 257ac2653a rework sigma device mapping 2024-02-07 00:44:12 -08:00
lllyasviel d11c9d7506 fix sigmas device in rare cases #71 2024-02-06 23:12:35 -08:00
lllyasviel 4ea4a92fe9 fix inpaint batch dim align #94 2024-02-06 22:57:53 -08:00
lllyasviel 65f9c7d442 Update README.md 2024-02-06 21:46:54 -08:00
lllyasviel c185e39e59 Backend: better controlnet mask batch broadcasting 2024-02-06 20:13:09 -08:00
Chenlei Hu 1110183943 Fix ControlNet UI preset (#87) 2024-02-06 21:52:04 -05:00
lllyasviel e62631350a Update forge_version.py 2024-02-06 17:55:14 -08:00
lllyasviel e579fab4d0 try solve #71 2024-02-06 17:46:23 -08:00
lllyasviel 6301a6660e solve #73 2024-02-06 17:28:00 -08:00
lllyasviel 711844ecd8 solve #76 2024-02-06 17:25:28 -08:00
lllyasviel 70ae2a4bce fix controlnet ignored in batch #55 2024-02-06 17:12:57 -08:00
lllyasviel fc5c70a28d gradio fix 2024-02-06 14:32:42 -08:00
Chengsong Zhang b58b0bd425 batch mask (#44)
* mask batch, not working

* mask batch, working, infotext broken

* try remove old codes

* set CUDA_VISIBLE_DEVICES with args

* Revert "try remove old codes"

This reverts commit 63c527c373.

* Update controlnet_ui_group.py

* readme

* 🐛 Fix infotext

---------

Co-authored-by: lllyasviel <lyuminzhang@outlook.com>
Co-authored-by: huchenlei <chenlei.hu@mail.utoronto.ca>
2024-02-06 12:54:35 -06:00
lllyasviel 5bea443d94 add a note after fixing repeated loading bug on 4090 2024-02-06 08:13:29 -08:00
Chenlei Hu 79e4e46061 🐛 Fix SVD tab (#63) 2024-02-06 11:10:47 -05:00