Video Frame Extractor Keep the moments that matter

Open a local file in the Video Frame Extractor, pause on the moment you want, and choose Save current frame. The still appears in the candidate area, where you can inspect it and download a PNG or JPEG. You can repeat this without leaving the page.

Yes. Enter a time in seconds, such as 12.5, or use 00:00:12.500, then click Go to time and save the current frame. Seeking selects a decoded picture near that position. Requested time and captured time are kept separately because they can differ.

Set a start time, end time, and interval. For example, 0 to 10 seconds at a 2-second interval requests 0, 2, 4, 6, 8, and 10 seconds when the video is longer than 10 seconds. The exact video endpoint is excluded. Review the estimate before collecting.

No. This Video Frame Extractor samples by time and supports manual snapshots; it does not enumerate every encoded frame. Browser seeking, video presentation callbacks, and display timing cannot guarantee an exhaustive sequence. Use a dedicated decoder workflow if you need every frame or frame-index accuracy.

Choose Original display size to use the browser-reported video dimensions, within the output limits. The default long edge is 1280 pixels and never upscales. The image comes from decoded video, not a screen capture. PNG avoids another lossy image encoding step but cannot restore detail missing from the video.

These names describe containers; the video codec inside must also be supported by your browser and operating system. The tool attempts actual decoding and reports loading, unsupported-source, or decoding failures. A damaged file or an audio-only track can also prevent image extraction. Try another browser or convert a local copy.

When fresh browser frame metadata is available, the Video Frame Extractor records its presentation time. Otherwise it labels the playback position as approximate, including an approx prefix in timestamped filenames. The ZIP manifest records both requested and captured times plus the time source; milliseconds do not imply frame-perfect precision.

Keep up to 100 candidate images and 128 MiB of encoded images in one collection. Each output is limited to 16 million pixels and 8192 pixels per side. The minimum interval is 0.1 seconds. Cancel stops further collection and retains completed pictures. Shorter ranges and smaller outputs use less memory.

Yes. Local video decoding, candidate selection, and downloads run in your browser without uploading the video or requiring an account. Opening a new video clears the candidates. Sending selected images to another tool uses temporary storage in the same browser; download a copy before leaving if you want to keep your work.

This version accepts local video files only. It does not fetch YouTube, social media, or remote video links. If you have a local copy you are allowed to use, open that file and collect its pictures in the same workspace.

First check whether the same moment is black in the video preview; intros and transitions can contain real black frames. Try a nearby time. Inspect portrait orientation and displayed proportions before a batch. Browser handling of rotation metadata and HDR color may vary; if preview and export disagree, try another browser or a standard SDR copy.

Send selected images to the image combiner for an overview or to the sprite sheet tool for a regular atlas. These transfers allow up to 32 million source pixels and 64 MiB in total, with 20 MiB per image. For animation editing, download the pictures and import them into the separate pixel art animation maker.