Commits · d191c4f9edee21f675d00d9d6feb02a32d63942e · chenpangpang / ComfyUI · GitLab

24 Jul, 2023 2 commits
- Add a ControlNetApplyAdvanced node. · d191c4f9
  comfyanonymous authored Jul 24, 2023
```
The controlnet can be applied to the positive or negative prompt only by
connecting it correctly.
```
  d191c4f9
- Add a way to set which range of timesteps the cond gets applied to. · 0240946e
  comfyanonymous authored Jul 24, 2023
  
  0240946e
23 Jul, 2023 1 commit
- Try to fix memory issue with lora. · 22f29d66
  comfyanonymous authored Jul 22, 2023
  
  22f29d66
22 Jul, 2023 4 commits
- Nodes can now patch the unet function. · 67be7eb8
  comfyanonymous authored Jul 22, 2023
  
  67be7eb8
- Del the right object when applying lora. · 12a6e931
  comfyanonymous authored Jul 22, 2023
  
  12a6e931
- Support controlnet in diffusers format. · 78e7958d
  comfyanonymous authored Jul 21, 2023
  
  78e7958d
- Fix issue with lora in some cases when combined with model merging. · 09386a36
  comfyanonymous authored Jul 21, 2023
  
  09386a36
21 Jul, 2023 1 commit
- Properly support SDXL diffusers unet with UNETLoader node. · 58b2364f
  comfyanonymous authored Jul 21, 2023
  
  58b2364f
20 Jul, 2023 1 commit
- Print errors and continue when lora weights are not compatible. · 01150186
  comfyanonymous authored Jul 20, 2023
  
  01150186
19 Jul, 2023 4 commits
- Fix typo. · 0b284f65
  comfyanonymous authored Jul 19, 2023
  
  0b284f65
- Fix ddim issue with older torch versions. · e032ca61
  comfyanonymous authored Jul 19, 2023
  
  e032ca61
- Add MX450 and MX550 to list of cards with broken fp16. · 18885f80
  comfyanonymous authored Jul 19, 2023
  
  18885f80
- It's actually possible to torch.compile the unet now. · 9ba44099
  comfyanonymous authored Jul 18, 2023
  
  9ba44099
18 Jul, 2023 1 commit
- Add key to indicate checkpoint is v_prediction when saving. · 51d54775
  comfyanonymous authored Jul 18, 2023
  
  51d54775
17 Jul, 2023 7 commits
- Fix device print on old torch version. · ff6b047a
  comfyanonymous authored Jul 17, 2023
  
  ff6b047a
- Enable --cuda-malloc by default on torch 2.0 and up. · 9871a15c
  comfyanonymous authored Jul 17, 2023
```
Add --disable-cuda-malloc to disable it.
```
  9871a15c
- --windows-standalone-build now enables --cuda-malloc · 55d0fca9
  comfyanonymous authored Jul 17, 2023
  
  55d0fca9
- Add a command line argument to enable backend:cudaMallocAsync · 1679abd8
  comfyanonymous authored Jul 17, 2023
  
  1679abd8
- Only calculate randn in some samplers when it's actually being used. · 3a150bad
  comfyanonymous authored Jul 17, 2023
  
  3a150bad
- Fix regression with ddim and uni_pc when batch size > 1. · ee8f8ee0
  comfyanonymous authored Jul 17, 2023
  
  ee8f8ee0
- Refactor of sampler code to deal more easily with different model types. · 3ded1a3a
  comfyanonymous authored Jul 17, 2023
  
  3ded1a3a
16 Jul, 2023 1 commit
- Lower lora ram usage when in normal vram mode. · 5f573626
  comfyanonymous authored Jul 16, 2023
  
  5f573626
15 Jul, 2023 3 commits
- Speed up lora loading a bit. · 490771b7
  comfyanonymous authored Jul 15, 2023
  
  490771b7
- Fix CLIPSetLastLayer not reverting when removed. · 50b1180d
  comfyanonymous authored Jul 15, 2023
  
  50b1180d
- Reduce floating point rounding errors in loras. · 6fb084f3
  comfyanonymous authored Jul 15, 2023
  
  6fb084f3
14 Jul, 2023 1 commit
- Add a node to merge CLIP models. · 91ed2815
  comfyanonymous authored Jul 14, 2023
  
  91ed2815
13 Jul, 2023 1 commit
- Prevent the clip_g position_ids key from being saved in the checkpoint. · b2f03164
  comfyanonymous authored Jul 12, 2023
```
This is to make it match the official checkpoint.
```
  b2f03164
12 Jul, 2023 2 commits
- Fix potential tensors being on different devices issues. · 46dc050c
  comfyanonymous authored Jul 12, 2023
  
  46dc050c
- fix mps miss import · 3e2309f1
  KarryCharon authored Jul 12, 2023
  
  3e2309f1
10 Jul, 2023 1 commit
- Support SDXL embedding format with 2 CLIP. · 606a5370
  comfyanonymous authored Jul 10, 2023
  
  606a5370
09 Jul, 2023 5 commits
- Don't patch weights when multiplier is zero. · 6ad0a6d7
  comfyanonymous authored Jul 09, 2023
  
  6ad0a6d7
- latent2rgb matrix for SDXL. · d5323d16
  comfyanonymous authored Jul 09, 2023
  
  d5323d16
- Empty cache after model unloading for normal vram and lower. · 0ae81c03
  comfyanonymous authored Jul 09, 2023
  
  0ae81c03
- Support loading clip_g from diffusers in CLIP Loader nodes. · d3f59982
  comfyanonymous authored Jul 09, 2023
  
  d3f59982
- Fix merging not working when model2 of model merge node was a merge. · a9a4ba75
  comfyanonymous authored Jul 08, 2023
  
  a9a4ba75
07 Jul, 2023 1 commit
- Add arguments to run the VAE in fp16 or bf16 for testing. · e7bee85d
  comfyanonymous authored Jul 06, 2023
  
  e7bee85d
06 Jul, 2023 2 commits
- Fix bug with weights when prompt is long. · 608fcc25
  comfyanonymous authored Jul 06, 2023
  
  608fcc25
- Disable autocast in unet for increased speed. · ddc6f12a
  comfyanonymous authored Jul 05, 2023
  
  ddc6f12a
05 Jul, 2023 2 commits
- Fix loras not working when loading checkpoint with config. · 603f02d6
  comfyanonymous authored Jul 05, 2023
  
  603f02d6
- Support loading unet files in diffusers format. · af7a4991
  comfyanonymous authored Jul 05, 2023
  
  af7a4991