Extract Frames
from Video

YouTube and TikTok thumbnails are the most common case. Scrub to the exact expression or composition you want, export as PNG at full source resolution, then bring into Photoshop or Figma for text overlay. A thumbnail pulled from the source is sharper than one re-extracted from a compressed upload, since you skip the second round of codec loss.

B-roll and editorial stills — journalists, documentary editors, and news producers pull frames from press conference footage or interview clips for use as standalone photographs in articles, social posts, and pitch decks. The frame extractor avoids the awkward screenshot-the-player workflow, which captures UI chrome and reduces resolution.

Frame-by-frame analysis — sports coaches, motion designers, animators, and VFX artists need to see exactly what happens between two motion-blurred frames. Step forward or back one frame (about 1/25 s), one second or five seconds with the buttons under the video to inspect a tennis serve, study an animation pose, or grab a hard-to-spot continuity error from a rough cut.

Reaction stills and meme generation — extract the single perfect facial expression from a podcast clip or a film scene to use as a chat reaction, blog header, or shareable image. The freeze-frame is often more usable than the moving clip itself in static contexts (email, PDF, print).

Video codecs do not store every frame independently. Most frames are P-frames (predicted from the preceding frame) or B-frames (interpolated between past and future frames). Only I-frames (a.k.a. keyframes) are fully self-contained. To decode any frame, the decoder must walk back to the previous I-frame and replay everything in between.

H.264 typically places an I-frame every 2 seconds (~48–60 frames at 24/30 fps). Seeking is fast and frame-accurate because the GOP (Group of Pictures) is short. H.265 / HEVC uses longer GOPs (10+ seconds) to improve compression; seeking is slower because more frames must be replayed. AV1 can use even longer GOPs and is markedly slower to decode in the browser without hardware acceleration.

ConverterUp uses the browser's own video decoder: you seek to a moment, and the frame currently shown is drawn to a canvas at the video's native resolution. The step buttons move by about 1/25 of a second, which lands on a new frame for 24–30 fps footage. H.264 MP4 and WebM work in every modern browser; HEVC/H.265 depends on your browser and hardware.

Variable frame rate (VFR) clips — common in screen recordings, OBS captures, and phone videos with adaptive frame rates — have frames of different durations. A fixed step may occasionally land on the same frame twice or skip one; if that happens, nudge again with the one-frame buttons until the preview changes.

Capture as many frames as you need: move to each moment and click Capture Frame. Every capture is added to the Captured Frames gallery with its timestamp, where you can preview, download or delete it — useful for storyboards, contact sheets and thumbnail picks.

For a storyboard, step through the video in 5-second jumps and capture at each stop; for a precise moment, use the one-frame buttons. Scene cuts are the natural points to capture, since they give visually distinct frames.

Choose PNG for lossless, full-quality frames or JPG for smaller files. Remember that PNG frames from HD or 4K video are large — a dozen 4K PNGs can easily reach 100 MB, so use JPG if size matters more than perfect quality.

Download All as ZIP bundles every captured frame into one archive, with filenames that include the frame number and its timestamp (for example frame-001-00-01-23.png), so alphabetical order matches the order in the video.