Extract frames from a video

PNG keeps every pixel, so text stays sharp; JPG makes much smaller files from camera footage.

A frame shows the video exactly as it was recorded at that moment. If the camera or the subject was moving, the frame is blurred; the frame next to it can be much sharper. Get close with the slider, then step with the frame buttons and watch the preview until the edges are crisp. The time beside the buttons shows where you are, to a hundredth of a second.

In a screen recording, wait for the screen to settle: just after a click, a page may still be loading or a menu fading in. A frame or two later usually shows it finished. For a slide from a recorded talk, pick a frame after the slide has fully appeared and before the speaker moves on; save it as PNG so small text stays readable.

To turn a recording into a written guide with screenshots, the video to blog tool writes the steps from the recording, and video to text gives you the transcript.

You may already have something that can do this. These are free:

VLC (Windows, Mac and Linux)

1. Open the video in VLC and pause it on the frame you want.
2. Choose Video, then Take Snapshot.
3. Find the PNG in your Pictures folder.

Free and saves the frame at the video’s full size. One frame at a time.

A screenshot (Mac and Windows)

1. Pause the video and make the player full screen.
2. On a Mac press Shift-Command-4 and drag over the picture; on Windows press Windows logo key + Shift + S.
3. The image is saved to your desktop (Mac) or Screenshots folder (Windows).

Quick, but you get your screen’s pixels, not the video’s: the size depends on your display and the player’s controls can get in the way.

FFmpeg (Any computer)

1. One frame at 1 minute 5 seconds: ffmpeg -ss 65 -i in.mp4 -frames:v 1 frame.png.
2. One frame every 10 seconds: ffmpeg -i in.mp4 -vf fps=1/10 frame-%03d.jpg.
3. Only new scenes: ffmpeg -i in.mp4 -vf "select=gt(scene\,0.3)" -fps_mode vfr scene-%03d.png.

Free and exact, from the command line. Its scene detection compares each frame with the one before, so a slow fade can give several frames.

This page needs nothing installed, works on any computer or phone with a current browser, and saves at the video’s own size.

• One frame at a time. To save many frames at once, such as one every second or every frame, use FFmpeg as shown above.
• It reads any video your browser can decode: H.264, VP8, VP9 and AV1 almost everywhere, HEVC (H.265) in Safari and in Chrome and Edge on most computers. The page says so if yours can’t.
• An HDR video’s frames are saved in standard colour, and look flatter than the video.
• Frames are saved at the size the video is shown at, with any rotation applied. JPG has no transparency, and neither does a video.
• It needs WebCodecs, which current Chrome, Edge, Safari and Firefox have; the page says so before you start if yours does not. It works on a phone too.

• A compressed video stores only a few frames whole, called key frames or I-frames. The rest are stored as changes from frames around them, so a player must decode from the last key frame to show any other one. Source, checked 9 November 2026.
• JPEG is the usual choice for photos and other still images where a little loss is fine. PNG reproduces the picture exactly, so text and lines stay sharp, at a larger file size. Source, checked 9 November 2026.
• In VLC, Video, then Take Snapshot saves the frame on screen as a PNG, by default in your Pictures folder. Source, checked 9 November 2026.
• FFmpeg’s fps filter keeps frames at a fixed rate, and its select filter can keep only new scenes; FFmpeg’s own guide says a scene value of 0.3 to 0.5 is generally a sane choice. Source, checked 9 November 2026.
• On a Mac, Shift-Command-4 turns the pointer into a crosshair to capture part of the screen; screenshots are saved to the desktop by default. Source, checked 9 November 2026.
• On Windows, Windows logo key + Shift + S opens Snipping Tool’s capture bar; snips are saved to your Screenshots folder. Source, checked 9 November 2026.