Save an IMAGE batch as H.264/H.265/ProRes/DNxHR/FFV1/WebM/GIF/WebP or 16-bit PNG / EXR sequences. Streams chunk-by-chunk within the RAM budget, muxes audio…

Experimental nodes! This is a production suite for ComfyUI — SAM2.1/SAM3 segmentation, alpha matting, inpaint crop/stitch, VFX tools (EXR I/O, LUTs, render passes), video mask propagation, temporal anchors, VAE merging, spline editors, and auto-versioned output.

ComfyUI-CustomNodePacks

format

filename_prefix

quality

images

audio

video_info

colorspace_in

colorspace_out

naming

naming_resolved

subfolder

save_output

latent

tile_mode

tile_size

Save an IMAGE batch as H.264/H.265/ProRes/DNxHR/FFV1/WebM/GIF/WebP or 16-bit PNG / EXR sequences. Streams chunk-by-chunk within the RAM budget, muxes audio, applies OCIO colour transforms, and shows a browser preview with timeline scrub after each run.

SaveVideoC2C

filenames

preview

Save Video (C2C)

Name	Type	Default	Description
format	COMBO		16 options: MP4 H.264, MP4 H.265 10-bit, MOV ProRes 422, MOV ProRes 422 HQ, MOV ProRes 4444, MOV ProRes 4444 XQ, +10
fps	FLOAT	24.0000.001–240	Frame rate for the written file. When video_info is wired and fps is left at 24, loaded_fps from the loader is used.
filename_prefix	STRING	C2C/video	—
quality	INT	800–100	Encoder quality where the format supports it (CRF / qscale / ProRes profile). Hidden for lossless formats.
imagesopt	IMAGE		The frames to save. Or wire latent + vae instead to decode and save as one stream.
audioopt	AUDIO		—
video_infoopt	VHS_VIDEOINFO		—
colorspace_inopt	COMBO	sRGB - Display	3 options: sRGB - Display, Linear Rec.709 (sRGB), ACEScg
colorspace_outopt	COMBO	same as input	4 options: same as input, sRGB - Display, Linear Rec.709 (sRGB), ACEScg
namingopt	COMBO	default	3 options: default, ComfyUI counter, Folder Version
naming_resolvedopt	STRING		Set by the UI from c2c.saveVideo.naming.
subfolderopt	STRING		—
save_outputopt	BOOLEAN	true	—
latentopt	LATENT		Stream save: a video latent decoded straight into the file, in order, keeping the VAE's causal state - the full frame batch is never held in RAM. Wire this and vae instead of images.
vaeopt	VAE		The VAE for latent (stream save).
tile_modeopt	COMBO	auto	Stream save, Wan VAEs: auto tiles only when the frame would not fit the GPU (2K on 8 GB -> 512 px). Each tile is decoded over the whole clip in one causal pass and blended on disk, so the clip never sits in RAM. off: no tiles. manual: tile_size.
tile_sizeopt	INT	51264–2048	Tile size in pixels when tile_mode is manual.

Save Video (C2C)

Inputs (17)

Outputs (2)