VeedRemoveVideoBackgroundNode - ComfyUI Built-in Node Documentation

Cut the subject out of a video with VEED and return the frames plus an alpha mask. Use the two outputs for compositing, or send them through Join Image with Alpha and Save WEBM to write a transparent video. Choose "standard" for the best edges on any background, "fast" when speed matters, or "green screen" for footage shot against a green backdrop.

Inputs

Parameter	Description	Data Type	Required	Range
`model`	Segmentation model to use: `"standard"` for any background with the best edges and output up to 3840x2160, `"fast"` for a quicker result up to 1920x1080, or `"green screen"` to key out a green backdrop up to 3840x2160. Larger videos are downscaled. Selecting an option reveals its parameters.	DYNAMIC_COMBO	Yes	`"standard"` `"fast"` `"green screen"`
`video`	The video to cut the subject out of. Up to 60 seconds.	VIDEO	Yes	Up to 60 seconds
`refine_foreground_edges`	Cleaner edges on hair, fur and fine detail. Turning it off lowers the price. Available for `"standard"` and `"fast"` (default: True).	BOOLEAN	Yes	false true
`subject_is_person`	Turn off when the subject is an object or animal. Available for `"standard"` and `"fast"` (default: True).	BOOLEAN	Yes	false true
`spill_suppression_strength`	How strongly green cast is removed from the subject. Raise it when green spots remain; lower it when the subject’s colors shift. Available for `"green screen"` (default: 0.8).	FLOAT	Yes	0.0 to 1.0 (step 0.01)

The clip is limited to 60 seconds; longer videos are rejected with an error. The mask is 1 where the background was removed, so use Invert Mask before Image Composite Masked when you want to keep the subject instead.

Outputs

Output Name	Description	Data Type
`images`	The video frames with the background removed.	IMAGE
`mask`	1 where the background was removed, per frame.	MASK

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): 1bf52a380ab09437a5a57f70215856d4ea0055f78ce8b42fe2a2b87d686e8d26

​Inputs

​Outputs

Inputs

Outputs