072dd685aa
Fix #1964 - don't include trailing slashes in folder paths in COMMANDLINE_ARGS
2025-08-14 15:04:27 +05:30
6ca4f74a6f
Log the COMMANDLINE_ARGS sent to Forge
2025-08-14 10:29:56 +05:30
8de2536982
Disable Forge's behavior of renaming models to corrupted on any exception (while loading)
2025-08-08 15:27:40 +05:30
cce6a205a6
Use the launched python executable for installing new packages
2025-07-25 14:43:26 +05:30
0daf5ea9f6
WebUI: Temporarily apply the zombie process fix only on Windows. It seems to break on Linux (possibly Mac too)
2025-07-19 00:14:54 +05:30
491ee9ef1e
Fix a bug where 'Use these settings' wouldn't reset the text encoder field (from non-Flux/Chroma tasks)
2025-07-18 16:37:23 +05:30
1adf7d1a6e
v3.5.9
2025-07-18 16:15:12 +05:30
36f5ee97f8
WebUI: Stability fix, to avoid zombie processes hanging around occasionally after closing Easy Diffusion
2025-07-18 16:13:47 +05:30
ffaae89e7f
WebUI: Force it to start at the specified port, or fail if something's already running at that port. Avoids Gradio's silent behavior of automatically starting at the next available port, which throws the system into an inconsistent state if Forge is already running (as a user choice, or as a zombie process)
2025-07-18 13:36:41 +05:30
ec638b4343
Fix typo that broke copy/paste settings
2025-07-18 10:26:17 +05:30
3977565cfa
Allow custom CLI args for WebUI
2025-07-14 19:40:05 +05:30
09f7250454
Log the WebUI server error
2025-07-14 19:11:07 +05:30
4e07966e54
Log the WebUI server error
2025-07-14 19:10:39 +05:30
5548e422e9
Better logging for webui server errors
2025-07-14 18:54:39 +05:30
efd6dfaca5
Stability fix for Forge - garbage collect forcibly before reloading big models. caused a process crash on occasion
2025-07-14 18:40:15 +05:30
8da94496ed
Allow gguf text encoders and vae
2025-07-14 18:03:39 +05:30
78538f5cfb
Remove temporary logs
2025-07-14 13:25:23 +05:30
279ce2e263
v3.5.8
2025-07-14 13:21:29 +05:30
889a070e62
Support custom text encoders and Flux VAEs in the UI
2025-07-14 13:20:26 +05:30
497b996ce9
Stability patches for the WebUI backend
2025-07-14 12:59:04 +05:30
83b8028e0a
Ignore pickle scanning for .sft and .gguf files
2025-06-27 16:43:41 +05:30
7315584904
v3.5.7 - chroma support
2025-06-27 15:32:18 +05:30
9d86291b13
webui: chroma support; update Forge to the latest commit
2025-06-27 15:21:09 +05:30
2c359e0e39
webui: logging changes
2025-06-27 15:20:30 +05:30
4044d696e6
Use torchruntime with Forge
2025-03-11 11:03:48 +05:30
5285e8f5e8
Use safetensors as the default model instead of ckpt
2025-03-11 10:24:55 +05:30
1ec4547a68
v3.5.6
2025-02-17 14:53:00 +05:30
827ec785e1
Fix broken model merge
2025-02-17 14:52:13 +05:30
19c6e10eda
Use the correct size of the image when used as the input. Code credit: @AvidGameFan
2025-02-10 12:49:38 +05:30
55daf647a0
v3.5.5 / v3.0.13 - torchruntime
2025-02-10 10:36:43 +05:30
7577a1f66c
v3.5.4
2025-02-08 11:27:32 +05:30
a8b1bbe441
Fix a bug where the inpainting mask wasn't resized to the image size when using the WebUI/v3.5 backend. Thanks @AvidGameFan for their help in investigating and fixing this!
2025-02-08 11:26:03 +05:30
a1fb9bc65c
Revert "Merge pull request #1874 from AvidGameFan/inpaint_mask_size"
...
This reverts commit f737921eaa
, reversing
changes made to b12a6b9537
.
2025-02-07 10:46:03 +05:30
f737921eaa
Merge pull request #1874 from AvidGameFan/inpaint_mask_size
...
Remove inpaint size limit
2025-02-07 10:27:44 +05:30
b12a6b9537
changelog
2025-02-06 13:08:58 +05:30
2efef8043e
Remove hardcoded torch.cuda references from the webui backend code
2025-02-06 13:07:10 +05:30
9e21d681a0
Remove hardcoded references to torch.cuda; Use torchruntime and sdkit's device utilities instead
2025-02-06 13:06:51 +05:30
964aef6bc3
Fix a bug preventing the installation of the WebUI backend
2025-02-04 17:57:08 +05:30
07105d7cfd
Debug logging for webui root dir
2025-02-04 17:47:58 +05:30
35f752b36d
Move the half precision bug check logic to sdkit
2025-01-31 16:28:12 +05:30
e57599c01e
Fix for conda accidental jailbreak when using WebUI
2025-01-28 10:33:25 +05:30
4a62d4e76e
Workaround for when the context doesn't have a model_load_errors field; Not sure why it doesn't have it
2025-01-04 18:04:43 +05:30
0af3fa8e98
version bump for wmic deprecation
2025-01-04 13:28:50 +05:30
74b25bdcb1
Replace the use of wmic (deprecated) with a powershell call
2025-01-04 13:25:53 +05:30
e084a35ffc
Allow image editor to grow
...
Limiting the editor size seems to cause failures in inpainting, probably due to a mismatch with the requested image size.
2024-12-27 22:10:14 -05:00
ad06e345c9
Allow new resolutions with Use As Input
...
When Use As Input provides unlisted resolutions, add them first so that they'll be recognized as valid.
2024-12-24 23:36:12 -05:00
572b0329cf
Prevent invalid size settings for Use As Input
...
Images larger than the window size show up as blank sizes when setting the field. Retain previous (presumably valid) size rather than set an invalid one.
2024-12-23 21:46:25 -05:00
33ca04b916
Use as Input sets size
...
Force the generate size to match the image size when clicking Use as Input.
2024-12-22 21:22:37 -05:00
8cd6ca6269
Remove inpaint size limit
...
With the limit removed, inpaint can work with image sizes larger than 768 pixels.
2024-12-22 17:36:01 -05:00
49488ded01
v3.5.1
2024-12-17 18:29:39 +05:30