b6f6588417
add licenses
2023-04-19 21:29:52 -07:00
f4cd2d51f6
use ControlNetHED with Apache License 2.0
2023-04-19 16:04:58 -07:00
238c8cd8f1
improve inpaint mask quality
2023-04-19 01:19:40 -07:00
483450add6
fix 90% amd problems
2023-04-19 00:37:20 -07:00
d62badd9d1
add sanity check to log so that reading github issue is easier
2023-04-18 09:06:55 -07:00
3d3555b960
add sanity check to log so that reading github issue is easier
2023-04-18 09:05:34 -07:00
ea2c16bf74
add sanity check to log so that reading github issue is easier
2023-04-18 08:54:26 -07:00
804d32ee3f
add options
2023-04-18 06:16:59 -04:00
57b95a3a78
better module
2023-04-18 00:21:29 -07:00
78ad101279
include downloads for clip vison fix2
2023-04-17 23:44:56 -07:00
c0059da198
include downloads for clip vison
2023-04-17 23:42:06 -07:00
1d07f5ebd7
change annotator paths so that users can remove all when corrupted
2023-04-17 22:48:00 -07:00
42158b1b43
major update - ControlNet 1.1 GUI
2023-04-17 16:53:20 -07:00
385b67c2e3
Merge upstream changes and resolve merge conflicts.
2023-04-17 09:35:49 -07:00
efee5fe399
remove buggy pycocotools
2023-04-15 11:40:35 -07:00
f5eb9bd080
i
2023-04-15 08:11:52 +00:00
920c948ec0
fix: remove large txt file (of)
2023-04-15 08:05:29 +00:00
b7a2058419
fix: deprec thr-based inversion (and #736 ), cleanup code
2023-04-15 07:41:26 +00:00
b1c0bc62e0
Add {lineart, lineart_anime, normalbae, oneformer, shuffle, zoe} annotator
2023-04-13 18:01:29 +08:00
603449ce48
Clarify error message and remove PIL import.
2023-03-31 12:07:44 -07:00
132bcb4dc9
Since we're already diverging from the original, might as well remove vestigial annotation stuff.
2023-03-31 11:39:37 -07:00
51cbd171f4
Default to 0.5 confidence. Handle an edge case where result set is null rather than empty.
2023-03-31 11:33:15 -07:00
4adb9a38b9
Make min_confidence a required parameter. Change the default max faces to 10.
2023-03-31 11:26:54 -07:00
42a25e0b5c
Update UI to allow number of faces and confidence to be selected.
2023-03-31 10:57:10 -07:00
3cd07a6ee8
Rename mediapipe_laion_face to mediapipe_face.
2023-03-31 10:48:17 -07:00
f4d728539c
Move the copy from controlnet's internals back into laion_face_common to see if we can alleviate the negative stride without changing the codebase.
2023-03-31 10:31:55 -07:00
a2c1561c1f
Fix processor script + init to correctly generate processed face.
2023-03-30 17:12:03 -07:00
78833dce35
Add MediaPipe Face conditioning as a step.
2023-03-30 16:30:41 -07:00
5dfdc772d4
fix: resize issue with t2i-adapter/color alignment
2023-03-08 03:52:21 +00:00
84274a1388
fix: move pidinet to correct device
2023-03-07 02:33:28 +00:00
6a425582e9
fix: add more fallbacks for mmpkg ( #510 )
2023-03-06 13:18:28 +00:00
b49d3375ca
Add binary preprocessor
2023-03-06 01:09:34 +09:00
eab5bc08cb
Merge pull request #479 from catboxanon/new-t2i-adapters
...
Support for new T2I adapters (color, style)
2023-03-05 23:50:54 +09:00
cdc807ad55
feat: add color annotator, minor fix
2023-03-05 04:54:18 +00:00
9acfb6d89e
feat: add clip_vision annotator, support non-image input
2023-03-05 04:27:36 +00:00
2ff1154f46
fix: module registration issue ( #488 )
2023-03-04 16:30:56 +00:00
eb84fd9c59
fix: add mmcv fallback - revert 6dea4af
2023-03-04 15:07:06 +00:00
6dea4af076
fix: remove mmcv packages, prepare for new annotator/models
2023-03-04 10:14:10 +00:00
63223e5261
fix(uniformer) fix model loading path
2023-02-23 22:34:52 +01:00
2397a6c884
Add files via upload
2023-02-23 08:43:57 +00:00
9831aa3006
fix: move leres model to correct device ( #293 )
2023-02-22 15:20:10 +00:00
f4f51abd3c
fix: add fallback path
2023-02-22 14:20:25 +00:00
ca88ccd8ed
Merge branch 'main' into model-path
2023-02-21 22:37:45 -07:00
caa56b7dc6
minor fix and #276
2023-02-22 04:07:12 +00:00
698e136093
Merge branch 'main' into model-path
2023-02-20 21:48:29 -07:00
3ccc1fd01e
fix: move depth_boosting to settings
2023-02-21 03:25:29 +00:00
b1dbf0a048
cleanup
2023-02-21 11:53:17 +09:00
0b55af6cee
Merge branch 'main' into main
2023-02-21 11:50:11 +09:00
85156dcdf9
fix: use CUDA only if available for leres
2023-02-20 20:09:39 -06:00
bcec70c378
Fix imports
2023-02-20 18:39:17 -05:00
c6fc96eef9
feat: add guess_mode and cfg_based guidance
2023-02-20 15:20:59 +00:00
030922a3d2
Add depth boost
...
https://github.com/compphoto/BoostingMonocularDepth
2023-02-19 20:52:00 -05:00
b62873196c
Merge branch 'main' into model-path
2023-02-18 20:47:41 -07:00
d32907b90a
Add LICENSE
2023-02-18 20:16:53 -05:00
453424fe21
Separate annotator folders
2023-02-18 15:39:15 -07:00
842f2a5ab5
Update leres path
2023-02-18 14:54:30 -07:00
7dd91595d1
Merge branch 'Mikubill:main' into model-path
2023-02-18 14:50:47 -07:00
5620c1cd5a
Add LeReS depth thresholds and cleanup
...
Adds near and background removal/thresholds
Adds CPU model loading
Increase default model res to 512
Pass image dimensions to model
Remove duplicate functions in utils.py
2023-02-18 14:02:13 -05:00
a8b1e79b9b
Remove unused ckpts folder
2023-02-17 21:53:28 -07:00
e1ff318d72
Move annotators
2023-02-17 15:48:27 -07:00
5048c4c9f1
Add support for LeReS
2023-02-17 14:27:29 -05:00
eca48855f8
merge conflict
2023-02-17 15:03:39 +00:00
3484a30b94
Merge pull request #94 from eltociear/patch-1
...
chore: fix typo in metrics.py
2023-02-17 23:35:48 +09:00
0146699545
feat: impl t2i-adapter
2023-02-17 11:40:19 +00:00
7a427668c8
Revert "use relative path for download directory"
2023-02-17 01:26:13 +09:00
4b8dea44af
use relative path for download directory
2023-02-17 00:07:18 +09:00
47092d69a2
Patch out prettytable requirement
2023-02-16 16:38:47 +02:00
df1c200779
chore: fix typo in metrics.py
...
Wether -> Whether
2023-02-16 21:15:29 +09:00
601192b7f3
Use CPU, not MPS for depth/normal map preprocessor
...
Use CPU instead of MPS for the depth/normal map preprocessor model to workaround the bugs in MPS that produce corrupted depth and normal maps.
Also included here is a change to `unload_midas_model()` to fix not sending the model to CPU.
2023-02-16 01:49:24 -05:00
c7c9f936d0
fix: unload models when switching preprocessor
2023-02-15 13:03:35 +00:00
5d04d8cc59
Add support for other torch devices to uniformer
...
Also fixes the model name for the uniformer model so it isn't redownloaded every time.
2023-02-14 22:46:36 -05:00
843c20a706
Add support for using the web UI torch device
2023-02-14 05:48:58 -05:00
d896a7c288
feat: add support for uniformer, skip reload model when not changed
2023-02-14 10:38:09 +00:00
570bb2c787
feat: add midas support
2023-02-13 17:37:17 +00:00
644e2a27ae
fix: mlsd issue, add resize mode radio( #3 )
2023-02-13 16:16:39 +00:00
4fdf3e58d2
fix: path issue on hed model
2023-02-13 14:05:53 +00:00
21056faa2a
feat: add samples
2023-02-13 09:58:17 +00:00
eed7e5dd51
feat: make openpose mode works
2023-02-13 09:10:52 +00:00
565447cfdb
feat: make hed mode works
2023-02-13 08:21:10 +00:00
1da97f4e2b
Add files via upload
2023-02-13 15:00:54 +09:00
lvmin
ljleb
Joseph Catrambone
Mikubill
RcINS
Yoshitada Kurozumi
AO
Nerdy Rodent
missionfloyd
wbh1129
Chris Mekarski
Luo Boming
Aarni Koskela
Ikko Eltociear Ashimine
brkirch