Choose a span, then review the image sequence
Start with the source file you intend to document. Use the video preview if the browser can decode it, or take the start time from your editing timeline. Enter a start and a length in seconds; the second field is a duration, not an end timestamp.
- 1
Set the start and length
Choose one local video. Set the start to the beginning of the passage you want and the length to the amount of footage to process. For example, a start of 8 and a length of 3 requests the span from 8 to 11 seconds. The job stops at the source end if it arrives first.
- 2
Pick a cadence, size and format
The frame-count limit shown before a run is based on the requested duration and rate. Reduce either if it exceeds 120. Pick a longest-edge cap and choose JPG for a compact batch or PNG for cleaner fine detail. The JPG quality control changes the encoding of the images, not the sampling schedule.
- 3
Inspect images and download
Move through the numbered output with the arrows or choose a filename from the list. Download the current still for a quick handoff, or save the ZIP and extract it to retain the full numbered sequence. The preview shows the exported images rather than a separately generated thumbnail.
Sampling is not the same as copying every source frame
A video has presentation timestamps as well as a nominal frame rate. Screen recordings and phone clips can show frames for unequal lengths of time. The extractor samples that timeline onto a regular cadence rather than treating every decoded picture as one output. Nearby samples can therefore use the same underlying picture, and a low rate can miss a brief event entirely.
Numbered filenames show the export order, not a guaranteed source frame number or timecode. Use them to organize reference images, then consult the original clip if an exact timestamp matters. If a transition lands between two samples, shorten the range and increase the rate around that transition instead of filling a ZIP with unrelated frames.
How the output settings interact
The available rates run from 0.25 to 10 pictures per second. The 480, 720 and 1280 pixel caps constrain the longest edge. The tool produces even dimensions and converts the displayed proportions to square pixels. JPG uses its own lossy quality choices; PNG stores the converted RGB image without lossy PNG compression. The ZIP groups the already encoded images without resizing or compressing them a second time. A run can return fewer pictures than the displayed limit, especially when it reaches the end of the source.
Good candidates for batch frame extraction
Check the stills at their intended size
Review the first and last image for coverage. Open a sample at actual size and look at small labels, high-contrast borders and areas with motion blur. Blur captured by the camera is part of the source; choosing PNG cannot remove it. For screenshots or archival references, keep the original video so you can return to the same passage later.
- Choose the sampling rate for the action, not the advertised source FPS. One image per second may cover a gradual change; a rapid gesture might need five or ten per second over a much shorter window.
- Set the longest-edge cap high enough to read important details. On a portrait video the cap applies to height, not width. Resizing does not recover detail missing from the original.
- Inspect the actual export before sending it. JPG can soften small text and colored edges, while PNG avoids an additional lossy image encode but can use much more space on detailed footage.
Stills and ZIP are created locally
The decoding engine runs in a browser worker. The initial engine download is separate from your source video, which the tool does not upload. The original is not changed. Reset abandons the current selection or job without deleting any files already saved on your device.
Extracted pictures can still expose everything visible in the recording, including account details, private messages or sensitive screens. Numbered filenames do not anonymize the image content. Open and review the exact stills you plan to share, especially when distributing a full ZIP.
Practical starting points
For a screen recording where a menu opens around 8 seconds, start slightly earlier and sample five times per second for two seconds. Select the frame with the fully expanded menu, not the one caught mid-animation.
For a tutorial that holds each instruction on screen for several seconds, try 0.25 FPS. It captures roughly one picture every four seconds and produces fewer repetitive stills to review.
For a product rotation, compare consecutive frames in the preview and export the couple that show the front and side most clearly. Keep the numbered ZIP for repeatable feedback.
Scope and practical limits
One local video and its first non-cover-art video track are used per run. Length is limited to 120 seconds and the requested rate and length cannot exceed 120 images. The browser keeps both the source and working output in memory; detailed PNG frames may hit a 64 MiB encoded-image limit before a ZIP can be made. Use a smaller cap or shorter window when that happens. A file can be readable by the extractor even when native browser playback is unavailable.
- Use a single-frame extractor if you already know the exact display time and want only one image. A cadence-based sequence is meant for reviewing a set, not identifying the source frame nearest a particular timestamp.
- Use a frame-accurate editing or archival workflow for every native frame, HDR preservation, ancillary data, or original timecode. These exported files are newly encoded square-pixel reference images.
Video frame extraction questions
Why are consecutive images sometimes identical?
The output uses a regular sampling rate. If the source holds a frame or contains fewer distinct pictures than the requested cadence, several output positions may show the same picture. Exporting more files does not interpolate new motion.
Will PNG preserve the video without any changes?
No. PNG compression does not degrade the image it receives, but the pixels may already have been decoded, resized and converted to RGB. The export is a still-image derivative, not a lossless copy of the compressed video stream or its HDR metadata.
Can I get all original frames from a long video?
This tool deliberately limits each job and exports at a chosen cadence. It is useful for a manageable selection, not an exhaustive frame dump. For review of a long clip, work in several short windows around the relevant events.
Does the ZIP contain a video or audio?
No. It contains the numbered JPG or PNG images only. To keep sound and motion, use a video trimming or conversion tool instead.