Commits · 13e6d5366e87ae76f517e1d79349e51fe92087b2 · chenpangpang / ComfyUI

"lightx2v/vscode:/vscode.git/clone" did not exist on "58f84489dab4780480c00c9cf9a2567a8ff69f26"

16 Dec, 2023 1 commit
- Switch clip vision to manual cast. · 13e6d536
  comfyanonymous authored Dec 16, 2023
```
Make it use the same dtype as the text encoder.
```
  13e6d536
15 Dec, 2023 1 commit
- Set clip vision model in eval mode so it works without inference mode. · 719fa086
  comfyanonymous authored Dec 15, 2023
  
  719fa086
12 Dec, 2023 1 commit

comfyanonymous authored Dec 11, 2023

comfy.ops -> comfy.ops.disable_weight_init

This should make it more clear what they actually do.

Some unused code has also been removed.

77755ab8

09 Dec, 2023 1 commit
- Use own clip vision model implementation. · 174eba8e
  comfyanonymous authored Dec 09, 2023
  
  174eba8e
08 Dec, 2023 1 commit
- Make --gpu-only put intermediate values in GPU memory instead of cpu. · 9ac0b487
  comfyanonymous authored Dec 08, 2023
  
  9ac0b487
26 Oct, 2023 1 commit
- Faster clip image processing. · 723847f6
  comfyanonymous authored Oct 26, 2023
  
  723847f6
18 Oct, 2023 1 commit
- Fix some potential issues. · 430a8334
  comfyanonymous authored Oct 18, 2023
  
  430a8334
12 Sep, 2023 1 commit
- Don't leave very large hidden states in the clip vision output. · ed587306
  comfyanonymous authored Sep 12, 2023
  
  ed587306
08 Sep, 2023 1 commit
- Add a penultimate_hidden_states to the clip vision output. · e85be36b
  comfyanonymous authored Sep 08, 2023
  
  e85be36b
28 Aug, 2023 2 commits
- Put clip vision outputs on the CPU. · 4e89b2c2
  comfyanonymous authored Aug 28, 2023
  
  4e89b2c2
- Load clipvision model to GPU for faster performance. · a094b45c
  comfyanonymous authored Aug 28, 2023
  
  a094b45c
18 Aug, 2023 1 commit
- Add support for clip g vision model to CLIPVisionLoader. · 76d53c46
  comfyanonymous authored Aug 18, 2023
  
  76d53c46
16 Aug, 2023 2 commits
- Fix clip vision issue with old transformers versions. · 58f0c616
  comfyanonymous authored Aug 16, 2023
  
  58f0c616
- Fix potential issue with batch size and clip vision. · ae270f79
  comfyanonymous authored Aug 16, 2023
  
  ae270f79
14 Aug, 2023 1 commit
- CLIPVisionEncode can now encode multiple images. · 9cc12c83
  comfyanonymous authored Aug 14, 2023
  
  9cc12c83
23 Jun, 2023 1 commit
- Fix error with ClipVision loader node. · 9e37f4c7
  comfyanonymous authored Jun 23, 2023
  
  9e37f4c7
22 Jun, 2023 1 commit
- Support base SDXL and SDXL refiner models. · f87ec10a
  comfyanonymous authored Jun 22, 2023
```
Large refactor of the model detection and loading code.
```
  f87ec10a
19 Jun, 2023 1 commit
- pop clip vision keys after loading them. · cd930d4e
  comfyanonymous authored Jun 18, 2023
  
  cd930d4e
15 Jun, 2023 1 commit
- Properly disable weight initialization in clip models. · bb1f45d6
  comfyanonymous authored Jun 14, 2023
  
  bb1f45d6
14 Jun, 2023 1 commit
- Don't initialize CLIPVision weights to default values. · fa2cca05
  comfyanonymous authored Jun 14, 2023
  
  fa2cca05
09 Apr, 2023 1 commit
- Fix for new transformers version. · 92eca60e
  comfyanonymous authored Apr 09, 2023
  
  92eca60e
02 Apr, 2023 1 commit

Add support for unCLIP SD2.x models. · 809bcc8c

comfyanonymous authored Apr 01, 2023

See _for_testing/unclip in the UI for the new nodes.

unCLIPCheckpointLoader is used to load them.

unCLIPConditioning is used to add the image cond and takes as input a
CLIPVisionEncode output which has been moved to the conditioning section.

809bcc8c