The term "sprite" is mostly used in games. A single image file that contains multiple sub-images is called a "sprite sheet". (This reduces overhead compared to loading a large number of image files.) In this case they happen to be video thumbnails.
I'll admit I was expecting an explanation of how to extract sprites from retro video game footage using FFmpeg, which I can see totally being possible with a bit of work.
Maybe thumbnail sprite sheet is a more accurate term? I would have thought thumbnail on it's own doesn't represent the fact we're stitching together all the thumbnails of the video into a single image.
I’m trying to remember the name of a UI control that could be backed by something like this. It was a sequence of thumbnails laid out in a row like a film strip. (For all I know, that was what it was called.)
We had to make do with Shape Tables in AppleSoft BASIC…good times drawing those out manually on graph paper. And then XOR-ing them onto the background. Amazing how good games were on the Apple ][ without dedicated graphics hardware.
I've used vcs[0] to make contact sheets from videos. It has a nice feature that if it detects the frame is blacked out, it will shift the capture time so every frame on the sheet has something to look at. Easy syntax to set capture at every delta or capture x number of frames evenly distanced, columns per row, ignore a set length of end time, and frame size.
This sounds like a more robust tool, however I wanted to keep our external tooling to a minimum. The contact sheet generation sits as part of our transcoding pipeline for uploaded media which already uses ffmpeg.
We're using this method as part of our user media upload pipeline at editsquare.com I'm guessing you're using ffmpeg wasm on your editor? One thing to note, that I didn't mention in the article, is generating a JPEG output is substantially quicker than a PNG.
