Action cameras are usually pointed at things that are hard to photograph any other way - mid-jump, underwater, on a helmet at speed. Video captures the moment reliably; a photo often does not. Which makes pulling stills out of the footage afterwards the natural workflow.

It works well, and there are three characteristics of action camera footage that change how you should go about it.

High frame rates are an advantage

Action cameras shoot fast. 60, 120 and 240 frames per second are all common, and this is the single biggest reason extracted stills from a GoPro often beat photographs of the same moment.

High frame rate requires a short shutter. A camera capturing 240 frames per second cannot expose each one for longer than roughly 1/240th of a second, and short exposures freeze motion. So individual frames from high-frame-rate footage are genuinely sharp, not merely acceptable - considerably sharper than frames from the same scene shot at 24 or 30 fps, where each frame carries motion blur by design.

The practical implication: if you know you want stills, shoot at the highest frame rate the resolution allows. You get more moments to choose from and each one is crisper.

Frame rate	Max shutter per frame	Motion blur
24-30 fps	~1/30s	Noticeable on fast motion
60 fps	~1/60s	Reduced
120 fps	~1/120s	Minimal
240 fps	~1/240s	Effectively frozen

HEVC is the usual obstacle

GoPro records HEVC in most high-resolution and high-frame-rate modes, and so do DJI and Insta360. Anything at 4K, anything above 60 fps, and all HDR modes are typically HEVC rather than H.264.

HEVC halves the file size at equivalent quality, which matters enormously on a device writing to a microSD card. The cost is browser support: Safari handles it reliably, Chrome and Edge depend on hardware decoding being present, and Firefox generally does not.

GoPro cameras have a setting for this. Under Preferences then Video Compression, choosing H.264 + HEVC records H.264 wherever the mode allows it and falls back to HEVC only for modes that require it. This costs storage and removes most compatibility problems.

Try Safari first if you have a Mac - it decodes HEVC natively.
Otherwise transcode to H.264 with HandBrake, using the Fast 1080p30 or Fast 2160p60 preset.
For future shoots, set the camera to prefer H.264 where possible.

Try FrameRipper - free, no upload

Extract frames from any video directly in your browser. No sign-up, no file size limits.

Open FrameRipper

Fisheye distortion

Action cameras use very wide lenses, and the widest modes bend straight lines noticeably toward the edges of the frame. On video this reads as immersive. On a still photograph it often just looks wrong, particularly if there is a horizon or a building in shot.

The cleanest fix is at capture time. GoPro’s Linear and Linear + Horizon Lock modes correct distortion in-camera, and DJI and Insta360 have equivalents. Frames extracted from footage shot in these modes need no correction at all.

For footage already shot wide, GoPro Player applies lens correction and re-exports. Lightroom and Photoshop both have GoPro lens profiles that correct extracted stills directly. Correcting the video before extraction is more efficient if you want many frames; correcting individual stills is easier for one or two.

What resolution to expect

Extracted frames come out at the video’s native resolution, which on modern action cameras is high enough for real photographic use.

Video mode	Frame resolution	Megapixels	Print size at 300dpi
1080p	1920x1080	2.1 MP	6 x 3.5 in
2.7K	2704x1520	4.1 MP	9 x 5 in
4K	3840x2160	8.3 MP	12.8 x 7.2 in
5.3K	5312x2988	15.9 MP	17.7 x 10 in

A 4K frame at 8.3 megapixels is comparable to a mid-range camera photo and prints comfortably at A4. Extracting from 4K or 5.3K rather than 1080p makes a large difference if the still matters.

Finding the moment

Action footage rewards dense sampling, because the moment you want is often a fraction of a second wide and there is no way to know in advance exactly when it falls.

A two-pass approach works well. Sample the whole clip sparsely - one frame every two seconds - to find the section containing the action. Then re-extract densely around it. On a thirty-second clip, 300 frames gives you one every tenth of a second, which at 120 fps footage means you are choosing between genuinely distinct moments rather than near-duplicates.

Because extraction is fast and over-sampling costs only disk space you immediately reclaim, err dense. Missing the peak of a jump by two frames is far more annoying than deleting 250 images.

Format and colour notes

JPEG at high quality is the right default for action stills - these are photographic images with lots of fine detail, exactly what JPEG handles well, and PNG would be several times the size for no visible gain. Use PNG only if you will be editing the frames repeatedly.

One thing to watch: if you shot in a flat or log colour profile - GoPro’s Flat, DJI’s D-Log - extracted frames will look washed out and desaturated. That is correct behaviour, not a fault. Flat profiles deliberately compress contrast to preserve grading latitude, and they are meant to be graded.

Grade the video before extracting rather than correcting each still afterwards. It is one operation instead of fifty, and the result is consistent across every frame.