WanAnimateToVideo - ComfyUI Built-in Node Documentation

WanAnimateToVideo prepares conditioning data and an initial latent for generating animated videos with Wan, using inputs such as a reference image, pose, face, background, and optional motion from a previous chunk. It also supports generating longer videos in chunks by reading and updating a video_frame_offset value. This node is marked as experimental.

Inputs

Parameter	Description	Data Type	Required	Range
`positive`	Positive conditioning for guiding the generation towards desired content.	CONDITIONING	Yes	-
`negative`	Negative conditioning for steering the generation away from unwanted content.	CONDITIONING	Yes	-
`vae`	VAE model used for encoding image and video inputs into latent space.	VAE	Yes	-
`width`	Width of the generated video in pixels (default: 832, step: 16).	INT	Yes	16 to MAX_RESOLUTION
`height`	Height of the generated video in pixels (default: 480, step: 16).	INT	Yes	16 to MAX_RESOLUTION
`length`	Number of frames to generate (default: 77, step: 4).	INT	Yes	1 to MAX_RESOLUTION
`batch_size`	Number of videos to generate in one batch (default: 1).	INT	Yes	1 to 4096
`clip_vision_output`	Optional CLIP vision output added to both positive and negative conditioning.	CLIP_VISION_OUTPUT	No	-
`reference_image`	Reference image used as the appearance starting point for the generated video. If not provided, a black image is used.	IMAGE	No	-
`face_video`	Video input providing facial expression guidance. It is resized to 512x512 and scaled to the range -1.0 to 1.0 internally.	IMAGE	No	-
`pose_video`	Video input providing pose and motion guidance.	IMAGE	No	-
`continue_motion_max_frames`	Maximum number of frames carried over from a previous motion sequence (default: 5, step: 4).	INT	Yes	1 to MAX_RESOLUTION
`background_video`	Background video used to fill the non-character parts of the frames.	IMAGE	No	-
`character_mask`	Mask defining character regions, used to separate the character from the background.	MASK	No	-
`continue_motion`	Previous motion frames to continue from, keeping temporal consistency with earlier generated chunks.	IMAGE	No	-
`video_frame_offset`	The amount of frames to seek in all the input videos. Used for generating longer videos by chunk. Connect to the video_frame_offset output of the previous node for extending a video. (default: 0, step: 1)	INT	Yes	0 to MAX_RESOLUTION

Parameter Constraints:

When continue_motion is provided, only its last continue_motion_max_frames frames are used.
Input videos (face_video, pose_video, background_video, character_mask) are offset by video_frame_offset before use. If the offset is greater than or equal to the input’s frame count, that input is ignored, except for a single-frame character_mask.
If character_mask has only one frame, that frame is repeated for every frame of the output.
When pose_video is shorter than length, its last frame is repeated to fill the remaining frames; the output length is not changed.
If clip_vision_output is provided, it is added to both positive and negative conditioning.
If reference_image is not provided, a black image (all zero values) is used as the default reference.
If continue_motion is not provided, the initial motion frames are filled with constant gray (0.5 intensity) frames.
When continue_motion is used, video_frame_offset is reduced by the number of carried-over frames before the next chunk offset is calculated, so overlapping frames are not processed twice.
background_video fills the motion frames after the reference-motion portion; it does not replace the reference image or the carried-over continue_motion frames.

Outputs

Output Name	Description	Data Type
`positive`	Modified positive conditioning with additional video context, including CLIP vision output, pose video latent, face video pixels, concatenated latent image, and concatenated mask.	CONDITIONING
`negative`	Modified negative conditioning with additional video context, including CLIP vision output, pose video latent, blank face pixels, concatenated latent image, and concatenated mask.	CONDITIONING
`latent`	Initial latent tensor (all zero samples) for the generated video, with the shape `[batch_size, 16, latent_length + trim_latent, latent_height, latent_width]`.	LATENT
`trim_latent`	Number of latent frames to trim from the beginning of the latent, corresponding to the reference image frames.	INT
`trim_image`	Number of image frames to trim from the beginning, corresponding to the reference motion frames.	INT
`video_frame_offset`	Updated frame offset to use for the next chunk, based on the input offset and the number of frames processed.	INT

This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub

Source fingerprint (SHA-256): a95bae4c7ae4ddc8a95bc9dafa2ca920b1d2166802615189537dce16949bfc03

​Inputs

​Outputs

Inputs

Outputs