Convert M4V to JPG

M4V is Apple’s extension for iTunes Store video, and store purchases are encrypted with FairPlay. A browser cannot decode them, so there is no frame to read and the conversion refuses with a message saying nothing in the file could be converted.

What you produced yourself is fine. QuickTime, iMovie, Final Cut and Compressor exports, recordings from a camera or a phone, and downloads from course and conference platforms are all unprotected and give up frames normally. If a shelf of files converts except for a handful, the handful were bought.

The timestamp defaults to 0, which gives the first frame of the video, and the first frame of almost anything is black, a slate, a title card or a logo. As poster art that is worse than no image, because it looks like the file is broken rather than unillustrated.

Thirty seconds in is a reasonable blind guess for a talk or a lecture, and ten percent of the running time is the rule of thumb most media servers use for the same reason. What you get is the frame being displayed at that second — precisely, the last frame whose own start time falls at or before it — so a number a little past the opening titles is usually enough.

A poster frame is doing a job that has nothing to do with fidelity: it has to be recognisable at the size of a thumbnail in a grid, and it has to be distinguishable from the twenty files next to it. A face, a title slide with large type, a distinctive location or a strong colour all read at 200 pixels wide. A wide shot of a room does not.

Contrast matters more than sharpness for the same reason. A dark frame from a dim room becomes an indistinct smudge in a library that draws thumbnails at small sizes, while a slightly worse-composed frame with a clear subject reads instantly. Choose for the grid, not for the picture.

Video encoders exploit motion. They spend bits where the eye is looking and skimp where movement hides the difference, and at 25 frames a second nobody notices. Freeze one of those frames and look at it at full size and the economy becomes visible as softness, blocking around edges and smeared fine detail.

The fix is a moment where nothing is moving. A held shot, a pause between gestures, a static slide — those frames were expensive for the encoder and are correspondingly detailed. Trying two or three timestamps a fraction of a second apart routinely produces a noticeably better picture for no extra effort.

The image comes out at the video’s own pixel dimensions with no downscaling: a 1080p file gives 1920 by 1080, a 4K file gives 3840 by 2160. That is far more than a library grid will ever display, and it is the right default — scaling down is trivial and scaling up recovers nothing.

It also means the same extracted frame serves several purposes. The poster in the library, the header on a course page and the thumbnail in a list can all come from one file, resized by whatever displays them, rather than from three separate extractions at three sizes.

Poster frames are photographic, and JPG was built for photographs. The same frame as PNG is commonly several times the size for a difference nobody sees at thumbnail dimensions, and a library with a thousand posters in it feels that difference when it draws a page.

The exception is a frame that is mostly text or interface — a slide, a screen recording, a title card in fine type. JPEG’s compression puts a faint halo around hard edges, which is invisible on a face and irritating on lettering. For those, take a PNG instead.

Up to a hundred files at a time on the free tier, each converting independently and downloadable together as a ZIP. Since the work happens on your machine there is no queue, no upload, and no allowance to run out of two-thirds of the way through.

One caveat that matters in practice: every file in a batch is sampled at the same timestamp. A shelf of talks that all open with the same thirty-second title sequence works perfectly; a mixed shelf of clips and features does not, and is better done in two or three passes grouped by rough length.

Most media servers and catalogues match artwork to video by file name, and the extracted JPG arrives named after the source. Keep it that way, and if you extract more than one candidate per title, keep the second in a different folder rather than renaming it with a suffix that breaks the match.

It is also worth recording which second each poster came from, because a title you revisit in a year is a title you will want a better frame for, and finding the moment again is the slow part.

The frame is decoded and encoded inside the browser tab on your own processor, so the M4V never leaves the machine. That is easy to confirm by watching the network tab during a conversion, and the free tier takes files up to 100 MB.

On this pair it is a proportion argument as much as a privacy one. Sending a two-gigabyte video to a server so that a 400-kilobyte poster can come back is a strange amount of network for the result, and doing it a hundred times over for a library is a strange way to spend an afternoon.