SeedVR2TemporalMerge - ComfyUI Built-in Node Documentation

This node recombines temporal chunks of latent video data that were previously split by the Split SeedVR2 Latent node. It uses a Hann window crossfade over the overlapping regions to create a smooth transition between chunks, or performs a simple concatenation when no overlap is specified.

Inputs

Parameter	Description	Data Type	Required	Range
`latents`	The sampled temporal chunks in sequence order.	LATENT	Yes	List of latents
`temporal_overlap`	The temporal_overlap output of Split SeedVR2 Latent. 0 = plain concatenation.	INT	Yes	0 to 16384 (default: 0)

Note: The temporal_overlap parameter must be wired from the temporal_overlap output of the Split SeedVR2 Latent node. All input latents must have the same batch size, number of channels, height, and width. Only the final chunk may have a shorter temporal dimension than the others.

Outputs

Output Name	Description	Data Type
`latent`	The recombined full-length latent.	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): 647653fc651462f62bcc9dca2919cfcaa4b476b4c679b0a1111698facab83f28

SeedVR2TemporalChunk - ComfyUI Built-in Node Documentation

EmptyChromaRadianceLatentImage - ComfyUI Built-in Node Documentation

​Inputs

​Outputs

Inputs

Outputs