Video to Images
100% private — runs on your device, never uploaded. Works offline once loaded.
Pull the individual pictures out of a video: every frame for rotoscoping or machine-learning training data, one a second for a storyboard or a contact sheet, or a handful for thumbnails to choose between. Choose the rate and the format — JPG for small files, PNG when you need the exact pixels — and the frames come back as a numbered ZIP. A minute of 30 fps video is 1,800 images, so start with one a second unless you truly need them all. Nothing is uploaded.
How many frames you actually want
Full rate means every frame: 30 a second, 1,800 a minute. That is what rotoscoping, frame-by-frame analysis and machine-learning datasets need, and it produces a lot of files — a minute of 1080p as PNG can run past a gigabyte. One frame a second is what you want for a storyboard, a contact sheet, or picking a thumbnail, and it keeps the output manageable.
The frames are numbered in order, so they can be re-assembled into video, imported as an image sequence, or stepped through in any viewer.
JPG or PNG
JPG re-compresses every frame, which is invisible for browsing and choosing but compounds if the frames are going back into a video or into an image pipeline. PNG stores exactly what was decoded, at several times the size. Use JPG to look through, PNG when the pixels matter.
One caveat for datasets: the frames come from a compressed video, so they already carry whatever artefacts the original encode introduced. PNG preserves those faithfully rather than removing them.
Frequently asked questions
Can I extract frames from just part of the video?
Trim the clip first, then extract from the trimmed copy.
How do I turn images back into a video?
The Video Studio timeline accepts an image sequence, and the GIF maker builds an animation from stills.
Advertisement