ComfyUI Extension
ComfyUI-LaVIT
Nodes:VideoLaVITLoader, VideoLaVITT2V, VideoLaVITI2V, VideoLaVITI2VLong, VideoLaVITT2VLong, VideoLaVITI2I
chaojie/ComfyUI-LaVIT
Nodes12
On cloudLocal install
CategoryStreamingT2V, LaVIT
Stars12
Updated2 years ago
Nodes (12)
VHS_FILENAMES_STRING_LaVIT
Turn VHS filename output into a plain path you can actually use
StreamingT2V
VideoLaVITI2I
This 'image-to-image' node is text-to-image — and it's broken as shipped
LaVIT
VideoLaVITI2V
Image-to-video, the Video-LaVIT way
LaVIT
VideoLaVITI2VLong
Stitch 24-frame clips into one continuous take
LaVIT
VideoLaVITLoader
Load Video-LaVIT before any generation node will talk to you
LaVIT
VideoLaVITT2V
Prompt it, get 24 frames out
LaVIT
VideoLaVITT2VLong
The multi-clip T2V node
LaVIT
VideoLaVITUnderstandingImage
Ask a still image a question and get a text answer out
LaVIT
VideoLaVITUnderstandingLoader
Load the vision half of Video-LaVIT — the model that answers questions
LaVIT
VideoLaVITUnderstandingVideo
Ask a video file what's happening in it
LaVIT
VideoLaVITVideoDetokenizerLoader
Load the piece that turns tokens back into pixels
LaVIT
VideoLaVITVideoReconstruction
Rebuild a video from one frame and its motion vectors
LaVIT
Readme
ComfyUI-LaVIT
workflow
T2V
<img src="wf_t2v.png" raw=true>I2V
<img src="wf_i2v.png" raw=true>T2V Long Video
<img src="wf_t2v_long.png" raw=true>Video Reconstruction
<img src="wf_videoreconstruction.png" raw=true>Model Weights
huggingface-cli download --resume-download rain1011/Video-LaVIT-v1 --local-dir ~/ComfyUI/models/diffusers/Video-LaVIT-v1 --local-dir-use-symlinks False
Notes
Width and height must be multiples of 64