WanPhantomSubjectToVideo - ComfyUI Built-in Node Documentation - ComfyUI

The WanPhantomSubjectToVideo node prepares conditioning data and a latent for Wan video generation. It creates an empty latent video from the requested width, height, length, and batch size, and, when reference images are supplied, encodes them with the VAE and adds them to the conditionings as time-dimensional visual guidance.

Inputs

Parameter	Description	Data Type	Required	Range
`positive`	Positive conditioning input for guiding video generation	CONDITIONING	Yes	-
`negative`	Negative conditioning input to avoid certain characteristics	CONDITIONING	Yes	-
`vae`	VAE model used to encode the reference images when they are provided	VAE	Yes	-
`width`	Output video width in pixels (default: 832, must be a multiple of 16)	INT	Yes	16 to MAX_RESOLUTION
`height`	Output video height in pixels (default: 480, must be a multiple of 16)	INT	Yes	16 to MAX_RESOLUTION
`length`	Number of frames in the generated video (default: 81, must be a multiple of 4)	INT	Yes	1 to MAX_RESOLUTION
`batch_size`	Number of videos to generate simultaneously (default: 1)	INT	Yes	1 to 4096
`images`	Optional reference images used as time-dimensional visual guidance	IMAGE	No	-

Note: When images are provided, they are automatically upscaled to match the specified width and height, and only the first length images are used for processing. Each image is encoded with the vae and concatenated along the time dimension, and only the RGB channels of each image are used.

Outputs

Output Name	Description	Data Type
`positive`	Positive conditioning with time-dimensional concatenation of the encoded reference images when images are provided; otherwise the input `positive` is returned unchanged	CONDITIONING
`negative_text`	Negative conditioning with time-dimensional concatenation of the encoded reference images when images are provided; otherwise the input `negative` is returned unchanged	CONDITIONING
`negative_img_text`	Negative conditioning with zeroed time-dimensional concatenation when images are provided; otherwise the input `negative` is returned unchanged	CONDITIONING
`latent`	Zero-filled latent video tensor with 16 channels; its frame count is derived from `length` and its spatial dimensions from `height` and `width`	LATENT

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): a1853382f6e564f66262b69dd7b06cc58e26b93386a460a98e6fcc2ff6acf12b