From 8be200eaccbfec58f73fcacb0bb952b41ef02be9 Mon Sep 17 00:00:00 2001 From: Weiqi Gao Date: Wed, 17 Dec 2025 13:24:44 +0800 Subject: [PATCH 1/3] Document usage of tae for VRAM reduction using wan Added details on using tae to reduce VRAM requirements with links to relevant safetensors --- docs/wan.md | 11 +++++++++++ 1 file changed, 11 insertions(+) diff --git a/docs/wan.md b/docs/wan.md index ce15ba589..7d69225bc 100644 --- a/docs/wan.md +++ b/docs/wan.md @@ -39,6 +39,17 @@ - safetensors: https://huggingface.co/Comfy-Org/Wan_2.1_ComfyUI_repackaged/blob/main/split_files/vae/wan_2.1_vae.safetensors - wan_2.2_vae (for Wan2.2 TI2V 5B only) - safetensors: https://huggingface.co/Comfy-Org/Wan_2.2_ComfyUI_Repackaged/blob/main/split_files/vae/wan2.2_vae.safetensors +
+ Using tae (taehv) to reduce VRAM requirement (but with poorer results) + + - taew2_1 (for all the wan model except Wan2.2 TI2V 5B) + - safetensors: https://github.com/madebyollin/taehv/blob/main/safetensors/taew2_1.safetensors + - taew2_2 (for Wan2.2 TI2V 5B only) + - safetensors: https://github.com/madebyollin/taehv/blob/main/safetensors/taew2_2.safetensors + + > To use these tae, simply replace the `--vae xxx.safetensors` with `--tae xxx.safetensors` in the commands below. If it still out of VRAM, add `--vae-conv-direct` to your command (but will be slower) + +
- Download umt5_xxl - safetensors: https://huggingface.co/Comfy-Org/Wan_2.1_ComfyUI_repackaged/blob/main/split_files/text_encoders/umt5_xxl_fp16.safetensors - gguf: https://huggingface.co/city96/umt5-xxl-encoder-gguf/tree/main From f14f77b6ada8aaabbe601f710235e361c98f8cad Mon Sep 17 00:00:00 2001 From: Weiqi Gao Date: Wed, 17 Dec 2025 23:26:51 +0800 Subject: [PATCH 2/3] Move tae usage to taesd.md --- docs/taesd.md | 24 +++++++++++++++++++++++- docs/wan.md | 14 +++----------- 2 files changed, 26 insertions(+), 12 deletions(-) diff --git a/docs/taesd.md b/docs/taesd.md index 5160b7934..1a14bfc33 100644 --- a/docs/taesd.md +++ b/docs/taesd.md @@ -14,4 +14,26 @@ curl -L -O https://huggingface.co/madebyollin/taesd/resolve/main/diffusion_pytor ```bash sd-cli -m ../models/v1-5-pruned-emaonly.safetensors -p "a lovely cat" --taesd ../models/diffusion_pytorch_model.safetensors -``` \ No newline at end of file +``` + +### Qwen-Image and wan + +sd.cpp also supports TAEHV (#937), which can be used for Qwen-Image and wan. + +For Qwen-Image and wan2.1 and wan2.2-A14B, download the wan2.1 tae [safetensors weights](https://github.com/madebyollin/taehv/blob/main/safetensors/taew2_1.safetensors) + +Or curl + +```bash +curl -L -O https://github.com/madebyollin/taehv/raw/refs/heads/main/safetensors/taew2_1.safetensors +``` + +For wan2.2-TI2V-5B, use the wan2.2 tae [safetensors weights](https://github.com/madebyollin/taehv/blob/main/safetensors/taew2_2.safetensors) + +Of curl + +```bash +curl -L -O https://github.com/madebyollin/taehv/raw/refs/heads/main/safetensors/taew2_2.safetensors +``` + +Then simply replace the `--vae xxx.safetensors` with `--tae xxx.safetensors` in the commands. If it still out of VRAM, add `--vae-conv-direct` to your command though might be slower. diff --git a/docs/wan.md b/docs/wan.md index 7d69225bc..6f5749c8c 100644 --- a/docs/wan.md +++ b/docs/wan.md @@ -39,17 +39,9 @@ - safetensors: https://huggingface.co/Comfy-Org/Wan_2.1_ComfyUI_repackaged/blob/main/split_files/vae/wan_2.1_vae.safetensors - wan_2.2_vae (for Wan2.2 TI2V 5B only) - safetensors: https://huggingface.co/Comfy-Org/Wan_2.2_ComfyUI_Repackaged/blob/main/split_files/vae/wan2.2_vae.safetensors -
- Using tae (taehv) to reduce VRAM requirement (but with poorer results) - - - taew2_1 (for all the wan model except Wan2.2 TI2V 5B) - - safetensors: https://github.com/madebyollin/taehv/blob/main/safetensors/taew2_1.safetensors - - taew2_2 (for Wan2.2 TI2V 5B only) - - safetensors: https://github.com/madebyollin/taehv/blob/main/safetensors/taew2_2.safetensors - - > To use these tae, simply replace the `--vae xxx.safetensors` with `--tae xxx.safetensors` in the commands below. If it still out of VRAM, add `--vae-conv-direct` to your command (but will be slower) - -
+ + > Wan models vae requires really much VRAM! If you do not have enough VRAM, please try tae instead, though the results may be poorer. For tae usage, please refer to [taesd](taesd.md) + - Download umt5_xxl - safetensors: https://huggingface.co/Comfy-Org/Wan_2.1_ComfyUI_repackaged/blob/main/split_files/text_encoders/umt5_xxl_fp16.safetensors - gguf: https://huggingface.co/city96/umt5-xxl-encoder-gguf/tree/main From ec560287806e5a59f8e21e9ee24110b72e89567d Mon Sep 17 00:00:00 2001 From: Weiqi Gao Date: Wed, 17 Dec 2025 23:31:27 +0800 Subject: [PATCH 3/3] Fix typo --- docs/taesd.md | 38 +++++++++++++++++++------------------- 1 file changed, 19 insertions(+), 19 deletions(-) diff --git a/docs/taesd.md b/docs/taesd.md index 1a14bfc33..a41c64d43 100644 --- a/docs/taesd.md +++ b/docs/taesd.md @@ -16,24 +16,24 @@ curl -L -O https://huggingface.co/madebyollin/taesd/resolve/main/diffusion_pytor sd-cli -m ../models/v1-5-pruned-emaonly.safetensors -p "a lovely cat" --taesd ../models/diffusion_pytorch_model.safetensors ``` -### Qwen-Image and wan - -sd.cpp also supports TAEHV (#937), which can be used for Qwen-Image and wan. - -For Qwen-Image and wan2.1 and wan2.2-A14B, download the wan2.1 tae [safetensors weights](https://github.com/madebyollin/taehv/blob/main/safetensors/taew2_1.safetensors) - -Or curl - -```bash -curl -L -O https://github.com/madebyollin/taehv/raw/refs/heads/main/safetensors/taew2_1.safetensors -``` - -For wan2.2-TI2V-5B, use the wan2.2 tae [safetensors weights](https://github.com/madebyollin/taehv/blob/main/safetensors/taew2_2.safetensors) - -Of curl - -```bash -curl -L -O https://github.com/madebyollin/taehv/raw/refs/heads/main/safetensors/taew2_2.safetensors -``` +### Qwen-Image and wan (TAEHV) + +sd.cpp also supports [TAEHV](https://github.com/madebyollin/taehv) (#937), which can be used for Qwen-Image and wan. + +- For **Qwen-Image and wan2.1 and wan2.2-A14B**, download the wan2.1 tae [safetensors weights](https://github.com/madebyollin/taehv/blob/main/safetensors/taew2_1.safetensors) + + Or curl + + ```bash + curl -L -O https://github.com/madebyollin/taehv/raw/refs/heads/main/safetensors/taew2_1.safetensors + ``` + +- For **wan2.2-TI2V-5B**, use the wan2.2 tae [safetensors weights](https://github.com/madebyollin/taehv/blob/main/safetensors/taew2_2.safetensors) + + Or curl + + ```bash + curl -L -O https://github.com/madebyollin/taehv/raw/refs/heads/main/safetensors/taew2_2.safetensors + ``` Then simply replace the `--vae xxx.safetensors` with `--tae xxx.safetensors` in the commands. If it still out of VRAM, add `--vae-conv-direct` to your command though might be slower.