Video Frame Sampler – Extract Evenly-Spaced Frames from Video URL

by zero-gif-zeroclick.vercel.app·Website ↗API ↗

Video Frame Sampler – Extract Evenly-Spaced Frames from Video URL is a paid API for AI agents from zero-gif-zeroclick.vercel.app, paid per call via x402, $0.01/call, status unknown (last checked 2026-10-23, last successful call 2026-07-02).

Extracts evenly-spaced image frames from a direct video file URL so a multimodal model can visually analyze video content

How this service works

Extract evenly-spaced frames from a video URL so your own multimodal model can 'watch' it. Body: { video_url, fps?: 1 (max 4), max_frames?: 60 (max 100), width?: 384 (max 768) }. Returns { frames: [{ url, t_seconds, index }], video_duration }. REQUIRED: video_url must be a direct .mp4/.webm/.mov/.mkv file URL — YouTube/Vimeo/Reddit/X page URLs and HLS .m3u8 streams do NOT work; pre-resolve with yt-dlp or use Wikimedia Commons / Internet Archive / Pexels which serve direct files. BEHAVIOR: samples evenly across the WHOLE video, no start/end window. For a long video where you only care about a specific moment, do one wide sample (max_frames 24-30) to localize the moment, then optionally a denser sample (max_frames 100, fps 4) to zoom in. TIMESTAMPS: each frame's t_seconds is in the JSON response, not burned into the image — read the JSON, do not OCR. LIMITS: source ≤ 10 min, ≤ 500 MB. Effective fps is capped to max_frames/duration so you never get more than max_frames frames.

Output

Returns a JSON object containing an array of frames, each with a hosted image URL, the timestamp in seconds (t_seconds), and a sequential index, plus the total video_duration in seconds. The frames are evenly distributed across the full video duration.