Video Contact Sheet

Summarize a whole video on one sheet: 12 to 30 frames selected automatically from the clip or a range you set, each labelled with its elapsed time. Download the sheet as a PNG or as one page of PDF.

The whole video on one sheet

Drop a video here

MP4, MOV or WebM. Your video stays on your device.

Title: —
File: —
Duration: —
Resolution: —
Range: —
Images: —

Verified facts

Inputs: MP4, MOV, WEBM
Output formats: PNG, PDF
Processing: In your browser. No media upload.
Access: Free · No signup
Frame limit: 30 per run

MP4, MOV and WebM are containers. Decoding depends on the codec and browser/device support; not every file in these containers will open.

How to use it

Choose a local video or try the sample. The tool scans the whole clip for likely visual cuts and builds an overview with a target of 20 images.
Choose a target of 12–30 images. For part of a video, set a time range and update the overview. Short sources can produce fewer distinct frames.
Choose a grid, wide or portrait layout. Add a title if you want one; the filename, full video duration, resolution and selected range identify the source. Every image shows its elapsed source time.
Optionally show estimated shot numbers, then download a large PNG or a one-page PDF. Layout and export settings keep the completed automatic selection.

Limitations

A sheet targets 12 to 30 frames, 20 to begin with, spread across the whole clip or a selected frame range. A source with fewer distinct frames than the target yields fewer cells.
Likely cuts are estimated by comparing up to 600 low-resolution samples on how red, green and blue are distributed across quarters of the frame. This is a heuristic on pixels. It does not recognise scenes or subjects, and camera motion, flashes, fades and dissolves can read as a cut or hide one.
Every result reports the interval of its scan: the longest gap in time between two sampled frames. A shot shorter than that gap can fall between samples and be missed, so a sheet is a view of the clip rather than a list of its shots. On a variable frame rate the gaps are uneven, and the reported interval is the widest one rather than an average.
Shot numbers are ids for the segments one scan estimated. They are numbered inside the range that was scanned and change when that range or the sampling changes. Asking for more or fewer images renumbers nothing by itself. They are not production shot numbers.
Sampling and cells use the source presentation timestamps, including variable frame rates. Each cell is decoded at its own size and checked against the requested timestamp; a mismatch fails instead of showing a neighbouring frame.
Every cell carries elapsed time from the first source frame as HH:MM:SS.mmm. This is not embedded camera timecode or SMPTE timecode.
The sheet is written at up to 4096 pixels on its longest edge and 12 million pixels in total. A width above either limit is reduced before rendering, and the reduction is reported.
The PDF holds the sheet as one page of image. Its text is part of that image, so it cannot be selected or searched, and print output is left to the reader.
Sources are read up to 2 GB and 240,000 frames.
A frame is read up to 16 megapixels and 16,384 pixels on its longest edge, measured on the display size and on the coded raster. A file past one of these limits is named in the error instead of being opened. These are protective limits, not measured memory ceilings.
The sheet is drawn by the browser in SDR. Each cell keeps the display aspect ratio of the source at the cell dimensions. HDR sources may look different.
Sheets are not saved between visits.
MP4, MOV and WebM are containers. Decoding depends on the codec and browser/device support; not every file in these containers will open.

An overview of the sequence

Use the sheet to review composition, color, shot variety, continuity and coverage across a sequence. It gives you a visual reference for your own judgement; it does not automatically score these qualities or diagnose continuity errors.

Use the Video Frame Extractor to find a source frame and export it as a separate still at the source display resolution. This tool summarizes the video on one page, so each image occupies only part of the sheet.

How automatic selection works

The analysis compares small decoded images across the full clip or selected range, with a maximum of 600 samples. Large visual differences suggest possible cuts. The tool shows the longest interval between sampled frames. Longer clips receive a coarser scan.

Frames are selected from the estimated shots and placed in chronological order. Long shots can contribute more than one image. If there are more estimated shots than the target image count, the sheet selects a spread across the sequence rather than showing every shot.

Motion, flashes and lighting changes can look like cuts. Gradual fades and short shots between samples can be missed. Similar-looking cuts can also be missed. The tool compares a limited set of images. It does not recognise the meaning of a scene or produce an edit decision list.

Time labels and shot numbers

Each image is labelled HH:MM:SS.mmm: elapsed source time from the first non-negative presentation timestamp. A selected range keeps the original source offset. These labels are not embedded camera timecode, SMPTE timecode or drop-frame counting.

Optional shot numbers such as 001 and 002 identify estimated segments within the analysed range. Multiple images from one segment can share a number; a limited overview may skip numbers. The numbering restarts when a different range is analysed and is not a production shot list.

PNG or PDF

Output	What you receive
PNG	One large raster image containing the header, selected frames and labels. Output dimensions are shown before download; a tall sheet may be scaled down to stay within the limits.
PDF	The same raster sheet fitted, without cropping, onto one A4 page. Page orientation follows the sheet. Text is part of the image, not searchable or editable vector text; small labels can be hard to read at print size.

Privacy and processing

Video analysis and PNG/PDF creation happen locally in this browser. The tool does not upload your media to Avo servers. The page and optional sample clip still use the network; local processing is not an offline or no-analytics claim.

Nothing is saved between visits. Changing the video discards the working overview after confirmation and releases its resources; downloaded files remain wherever you saved them.

Questions

Will it find every shot?

No. The scan samples the video and estimates cuts from visual changes. Very short shots, fades and similar-looking cuts may be missed; motion and flashes may create extra candidates. The overview is for visual review, not frame-accurate cut detection.

Can I inspect a single scene or sequence?

Yes. Set a source time range and update the overview. Selection and estimated shot numbering then apply to that range; image timestamps still refer to the original video.

Can I send a frame directly to the Frame Extractor?

Not yet. The tools do not currently share a local media session. Open the original video separately in the Frame Extractor for precise navigation and individual still exports.

Why will my MOV or MP4 not open?