Frame Extract (FFMPEGA)

Extract individual frames and audio from a video at a specified frame rate, time range, and frame limit.

By AEmotionStudio·Created 7 months ago·Updated 3 months ago· 14

◄video_path►

◄fps1.0►

◄start_time0.00►

◄duration0.00►

◄max_frames100►

◄input_video_path—►

◄mask_points—►

CategoryFFMPEGA

Inputs (9)

Name	Type	Default	Description
video_path	STRING		Absolute path to the video file to extract frames from.
fps	FLOAT	1.00.1–60	Frames per second to extract. 1.0 = one frame every second, higher values extract more frames.
start_timeopt	FLOAT	0.000–3600	Start time in seconds for frame extraction.
durationopt	FLOAT	0.000–3600	Duration in seconds to extract frames from. 0 = full video from start_time.
max_framesopt	INT	1001–1000	Maximum number of frames to return. Limits output to prevent memory issues with long videos.
imagesopt	IMAGE		Optional upstream IMAGE input (e.g. from VHS or Save Video). Accepted for wiring but frames are always extracted from the video.
audioopt	AUDIO		Optional upstream AUDIO. If connected, this audio is output instead of extracting from the video.
input_video_pathopt	STRING		Optional upstream video path (e.g. from Save Video or Load Video Path). Overrides the video_path text field when connected.
mask_pointsopt	STRING		Optional upstream mask_points pass-through. When connected, overrides the locally drawn mask points.

Name	Type	Description
frames	IMAGE	Extracted video frames as a batched image tensor.
audio	AUDIO	Audio from the extracted segment in ComfyUI AUDIO format.
frame_count	INT	Number of frames extracted.
fps	FLOAT	Source video FPS (native frame rate, not extraction rate).
duration	FLOAT	Actual duration of the extracted segment in seconds.
video_path	STRING	Resolved absolute path to the video file.
mask_points	STRING	JSON-encoded point selection data from the Point Selector. Connect to FFMPEGA Agent's mask_points input for guided masking.
crop_data	STRING	JSON-encoded crop rectangle from the Crop Selector. Format: {"x":N, "y":N, "w":N, "h":N}.