Skip to main content
This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub
The WanTrackToVideo node converts motion tracking data into video sequences by processing track points and generating corresponding video frames. It takes tracking coordinates as input and produces video conditioning and latent representations that can be used for video generation. When no tracks are provided, it falls back to standard image-to-video conversion.

Inputs

ParameterData TypeRequiredRangeDescription
positiveCONDITIONINGYes-Positive conditioning for video generation
negativeCONDITIONINGYes-Negative conditioning for video generation
vaeVAEYes-VAE model for encoding and decoding
tracksSTRINGYes-JSON-formatted tracking data as a multiline string (default: ”[]“)
widthINTYes16 to MAX_RESOLUTIONOutput video width in pixels (default: 832, step: 16)
heightINTYes16 to MAX_RESOLUTIONOutput video height in pixels (default: 480, step: 16)
lengthINTYes1 to MAX_RESOLUTIONNumber of frames in the output video (default: 81, step: 4)
batch_sizeINTYes1 to 4096Number of videos to generate simultaneously (default: 1)
temperatureFLOATYes1.0 to 1000.0Temperature parameter for motion patching (default: 220.0, step: 0.1)
topkINTYes1 to 10Top-k value for motion patching (default: 2)
start_imageIMAGENo-Starting image for video generation
clip_vision_outputCLIPVISIONOUTPUTNo-CLIP vision output for additional conditioning
Note: When tracks contains valid tracking data, the node processes motion tracks to generate video. When tracks is empty, it switches to standard image-to-video mode. If start_image is provided, it initializes the first frame of the video sequence.

Outputs

Output NameData TypeDescription
positiveCONDITIONINGPositive conditioning with motion track information applied
negativeCONDITIONINGNegative conditioning with motion track information applied
latentLATENTGenerated video latent representation