Commit Graph
79 Commits
Author SHA1 Message Date
matt3o 53a06f11a2 add simple weight types to the basic IPAdapter node 2024-03-29 16:51:00 +01:00
matt3o 4b88fb404f added support for SDXL portrait in the unified loader 2024-03-28 20:18:07 +01:00
matt3o 67124914d9 add experimental style transfer weight type (sdxl) 2024-03-27 13:54:03 +01:00
matt3o 83bf9108ba fix custom mask in tiled batch 2024-03-27 09:28:58 +01:00
matt3o 477217b364 fix prepimage pad take2 2024-03-26 15:02:27 +01:00
matt3o eb60705e82 fix prepimage pad 2024-03-26 14:59:59 +01:00
matt3o 64eda9be97 fix lora and insightface loading in the unified loader 2024-03-26 11:55:17 +01:00
matt3o ece285d610 add embeds scaling 2024-03-25 20:17:51 +01:00
matt3o 13d7e12e07 fix the attn mask for the tiled ipadapter 2024-03-24 15:08:24 +01:00
matt3o 10150e4763 fallback for old torch versions 2024-03-24 07:41:25 +01:00
matt3o 643cb81e00 complete code rewrite. there will be blood 2024-03-23 20:09:30 +01:00
matt3o 6a411dcb2c fix timestepping on windows? 2024-02-13 17:30:34 +01:00
matt3o d87b7ce1c1 improve compatibility with InstantID 2024-02-13 17:01:18 +01:00
matt3o ba25c1f725 add compatibility with instantid 2024-02-11 10:31:53 +01:00
matt3o 90d3451cd9 add experimental tiled IPAdapter 2024-02-02 20:06:27 +01:00
matt3o 46241f3ba5 support for faceid portrait 2024-01-19 15:08:08 +01:00
matt3o 52434a9c90 support for SDXL Plus v2 models 2024-01-17 12:00:52 +01:00
matt3o 56a9fc08de increase quality of FaceID Plus/v2 models 2024-01-16 12:24:13 +01:00
matt3o 90f1475483 allow weights to be 0 2024-01-16 10:45:00 +01:00
matt3o c37485514b testing nothing to see here 2024-01-09 17:10:47 +01:00
matt3o 4b6d3c3518 better error reporting for insightface 2024-01-04 08:33:14 +01:00
matt3o bd712d85ac typo in node name 2023-12-31 09:58:59 +01:00
matt3o 3e80c92149 FaceID v2 2023-12-30 20:30:40 +01:00
matt3o a09f541184 Add FaceID Plus Support 2023-12-28 14:21:59 +01:00
matt3o 69e82da76b preliminary support for FaceID 2023-12-22 15:47:45 +01:00
matt3o a4e71ccbbf hopefully found the right combination for dtype 2023-12-17 20:02:43 +01:00
matt3o cb1c5e8a1a trying to find the right combination of dtypes :p 2023-12-17 18:10:41 +01:00
matt3o f21e22225b try to fix #130 mask resizing issue 2023-12-17 16:31:32 +01:00
matt3o f9d3769800 fix regression 2023-12-17 16:07:29 +01:00
matt3o 6a7e085efd tentatively remove autocast 2023-12-17 15:27:49 +01:00
matt3o 186b76eb58 fix #124 after comfyui changes on the clip vision encoder 2023-12-09 20:56:58 +01:00
matt3o 3e05e86805 add batch embeds node 2023-12-05 20:26:02 +01:00
matt3o 66d2f3133b better(?) mask downscaling, hopefully fix #97 2023-12-05 20:26:02 +01:00
matt3o 26ca61a025 trying to fix autocast on Mac 2023-12-02 10:37:07 +01:00
matt3o 50bde71f23 cleanup 2023-12-02 08:20:20 +01:00
matt3o 701f54d977 fix mps 2023-12-02 08:19:17 +01:00
matt3o ade7f2005c doesn't assume cuda 2023-12-01 10:38:38 +01:00
matt3o 77554c835d add unfold batch option 2023-11-29 18:51:24 +01:00
matt3o 6bb44bb0ab disable timestepping if not available 2023-11-28 09:43:06 +01:00
matt3o 1ed4f93d1b timestepping, fix #83 2023-11-26 20:20:59 +01:00
matt3o c28a04466b Merge laksjdjf 2023-11-26 10:06:02 +01:00
matt3o ebd946f947 add multiple attention masks 2023-11-24 19:33:54 +01:00
matt3o d6cfa11bef add global models directory, fix #82 2023-11-23 15:04:17 +01:00
matt3o 2255ae7660 fix compatibility with deep shrink 2023-11-20 20:23:57 +01:00
matt3o ee85b4cff6 Add padding to prep image, fix #79 2023-11-16 17:05:25 +01:00
laksjdjf 6f0e573c3e Update IPAdapterPlus.py 2023-11-16 20:39:01 +09:00
laksjdjf 2881bc58f8 fix typo 2023-11-16 20:12:47 +09:00
matt3o 4cb9e97583 remove left over 2023-11-11 09:09:26 +01:00
laksjdjf f36ed1572a support full model 2023-11-11 08:18:09 +09:00
matt3o bf0508bd59 attention masking 2023-11-08 20:54:08 +01:00