Account

What new tool do you want?

Video & Audio

Video Frame Extractor & Thumbnail Grabber

Save any frame of a video as a high-quality image, or extract frames every N seconds — plus see the video's resolution, duration and frame size.

How to use

  1. 1Add a video and scrub to the moment you want.
  2. 2Press Save frame, or extract a frame every few seconds as a ZIP.

Frequently asked questions

What resolution are the frames?

Full video resolution (e.g. 1920×1080 for 1080p video).

Which formats work?

Any video your browser can play — MP4 (H.264), WebM, and MOV in Safari.

How do I get an exact frame?

Pause and use the ◀ ▶ buttons to step about one frame at a time.

Why this uses the browser's native video element instead of FFmpeg

Unlike most tools in this category, frame extraction here doesn't invoke FFmpeg at all — it plays the video in a standard HTML5 <video> element, seeks to the desired timestamp, and draws the current video frame directly onto a canvas with drawImage(), then encodes that canvas as a PNG or JPEG. That's a much lighter approach for exactly this job: grabbing still frames doesn't need a full video-processing engine, just the browser's own decoder (which it already has, since it's playing the video) and a canvas snapshot — which is why this tool feels noticeably snappier than the FFmpeg-based tools that have to load a ~32MB WebAssembly engine first.

Why frame-stepping is an approximation, not an exact frame count

The ◀ ▶ buttons step by exactly 1/30th of a second regardless of the video's actual frame rate — a reasonable approximation for the very common 30fps case, but if your source footage is actually 24fps, 25fps or 60fps, each button press won't land precisely on a true frame boundary. For video formats and browsers that support more precise frame-stepping APIs this would be more exact, but as a cross-browser, cross-format solution, the fixed 1/30s step is a pragmatic compromise — close enough to 'step forward roughly one frame' for the vast majority of consumer video without needing to parse the exact frame rate out of every possible container format first.

Why batch extraction caps out at 500 frames

Extracting 'every N seconds' works by seeking the video element to each timestamp in sequence, capturing a canvas snapshot at each stop, and zipping all of them together client-side — entirely in browser memory, with no server to offload the accumulating image data to. The hardcoded 500-frame ceiling exists specifically to stop a very long video combined with a very short interval (a 2-hour video at a 1-second interval would be 7,200 frames) from silently consuming all of a tab's available memory mid-extraction — if you hit that cap, the fix is deliberately in your control: increase the interval between frames rather than extracting every single one.

A frame at 12.5 seconds downloads as clip-12_50s.png

Saving one frame uses the full width and height of the video. JPEG is saved at about 95% quality. PNG is the format already selected, and that quality figure does not change a PNG. The filename uses the time on screen, to two decimal places, with the dot written as an underscore. A frame at 12.5 seconds from clip.mp4 is clip-12_50s.png, or clip-12_50s.jpg if you switch to JPEG. A batch uses a different pattern: clip-0001.png, clip-0002.png, and so on, four digits starting at 0001, inside clip-frames.zip. If a step would land past the end of the video, it stays on the last moment the player can show.

A blank interval means every 5 seconds, and the last step can fall short of the end

The interval box opens at 5 seconds. Leaving it blank, or typing something that is not a number, is read as 5. Anything under a tenth of a second, including 0.05, is raised to 0.1 seconds, which is as often as a batch will go. The number of frames is the length divided by that interval, rounded down, plus one, and the first frame is always at the start. A 10-second video every 5 seconds is 3 frames, at 0, 5, and 10 seconds. The same video every 3 seconds is 4 frames, at 0, 3, 6, and 9 — not at 10, because 9 is the last step that still fits the count. The hint beside the button shows that count before you extract.

Related tools

More: Video & Audio