Wan 2.1 FLF model support?

[matabear-wyx]

Loading models from /data/yuxiong/pretrained-weights/Wan2.1-FLF2V-14B-720P on GPU 0
Loading models from: ['/data/yuxiong/pretrained-weights/Wan2.1-FLF2V-14B-720P/diffusion_pytorch_model-00001-of-00007.safetensors', '/data/yuxiong/pretrained-weights/Wan2.1-FLF2V-14B-720P/diffusion_pytorch_model-00002-of-00007.safetensors', '/data
/yuxiong/pretrained-weights/Wan2.1-FLF2V-14B-720P/diffusion_pytorch_model-00003-of-00007.safetensors', '/data/yuxiong/pretrained-weights/Wan2.1-FLF2V-14B-720P/diffusion_pytorch_model-00004-of-00007.safetensors', '/data/yuxiong/pretrained-weights/
Wan2.1-FLF2V-14B-720P/diffusion_pytorch_model-00005-of-00007.safetensors', '/data/yuxiong/pretrained-weights/Wan2.1-FLF2V-14B-720P/diffusion_pytorch_model-00006-of-00007.safetensors', '/data/yuxiong/pretrained-weights/Wan2.1-FLF2V-14B-720P/diffus
ion_pytorch_model-00007-of-00007.safetensors']
We cannot detect the model type. No models are loaded.

It would be nice if we have official Wan 2.1 FLF model support here, since Diffsynth already supported first last frame generation for base I2V model

[Artiprocher]

@matabear-wyx Please install from source, not pip.

[ForeverAurorak]

Hi, I installed it from the source code, but the problem still exists.

[KevinXu-01]

Hi, I installed it from the source code, but the problem still exists.

Hi. You can clone this repo to local,
cd DiffSynth-Studio-main
pip install -v -e .
and then run
python examples/wanvideo/wan_14B_flf2v.py