Does compressing iPhone video keep Dolby Vision?
Usually, no. Most apps that shrink video either turn iPhone HDR into plain SDR, or keep the HDR picture and quietly drop the part that makes it Dolby Vision. A few keep everything. Which one you get is rarely written on the box, so here is what to look for.
What is in an iPhone Dolby Vision video
The iPhone 12 and every later model except the iPhone SE record HDR video in Dolby Vision unless you turn it off in Settings → Camera → Record Video (the switch is called HDR Video or just HDR, depending on the model). Apple’s technical note TN3145 describes the format: Dolby Vision profile 8.4, which is two things in one file.
- An HLG HDR picture. 10-bit HEVC in the wide BT.2020 color space, with the HLG transfer curve. Anything that understands HLG plays this as HDR, Dolby Vision or not.
- Dolby Vision metadata for every frame. A small block of data next to each frame that tells a Dolby Vision screen how to fit that frame’s highlights and colors into what the screen can actually show.
That design is forgiving: lose the metadata and you still have a good HDR video.
What compression can do to it
It becomes SDR. The copy has ordinary video colors and brightness. Bright skies and reflections flatten out, and the clip looks dull next to the original on an iPhone or a recent Mac. This is what happens when a tool converts to SDR or drops the HDR tags, and it can’t be undone.
It stays HDR, without Dolby Vision. The copy keeps 10-bit HLG, but the per-frame metadata is gone, and a Dolby Vision screen falls back to its own guess about each frame. This is where most careful tools stop. HEVCut, for example, says openly that Dolby Vision’s dynamic metadata isn’t kept.
It stays Dolby Vision. The copy keeps the HLG picture and the metadata of every frame, and the file says so. This takes more work, and few tools do it.
The difference shows most where iPhone HDR shines: snow, sunsets, city lights, anything with bright highlights. And the metadata is a tiny part of the file, so there is no reason to give it up to save space.
The copy that looks too bright
There is a fourth outcome, and it fools people because it looks like an improvement. The copy plays noticeably brighter than the original.
Camera HDR video carries more than the picture and the Dolby Vision data, and Apple’s players use some of it to decide how bright to show a clip. TN3145 warns that removing the ambient viewing environment information from camera video results in “extra bright imagery”. iPhones also record a separate track with the brightness of the scene, moment by moment. In our own testing, dropping that track alone is enough: macOS shows the copy noticeably brighter than the original, even though the pixels are the same.
So if a compressed copy looks punchier than the original, don’t keep it on that basis. Something was lost on the way.
How to check a file
You can see whether a video is Dolby Vision with ffprobe, the free inspector that comes with FFmpeg:
ffprobe -hide_banner yourclip.mov
In the output, look at the video stream. An iPhone Dolby Vision clip, and a copy that kept it, show arib-std-b67 (that is HLG) and a line like this:
DOVI configuration record: version: 1.0, profile: 8, level: 4, rpu flag: 1, el flag: 0, bl flag: 1, compatibility id: 4
Profile 8 with compatibility id 4 is the iPhone’s 8.4. The level depends on resolution and frame rate, so yours may differ. rpu flag: 1 says there is per-frame metadata. If the line is missing, the copy isn’t Dolby Vision. If the stream no longer says arib-std-b67 (or smpte2084, the other HDR curve), it isn’t HDR either.
One caveat: that line is a declaration. A file can carry it and still have missing or mismatched metadata inside. The only real test is decoding both files and comparing the metadata frame by frame, which is what KeepFeather does for every copy.
Why most tools don’t keep it
Keeping Dolby Vision through compression takes several things to go right at once:
- The encoder has to receive the original metadata for each frame and write it next to the new frame. Many encoders, and the apps built on them, drop it unless told otherwise.
- The file has to declare the Dolby Vision layer in a specific box. FFmpeg doesn’t write that box in the MOV format iPhones use, so another tool has to.
- Frame timing has to line up exactly, including the variable frame rate iPhones record, or the metadata ends up on the wrong frames.
- The brightness information above has to come along.
- Someone has to check the result, because a file that looks fine in a quick preview can still be missing half of it.
HandBrake can pass Dolby Vision through since version 1.7, but only with its 10-bit H.265 encoders, not with its default preset, and it doesn’t check the result. More in KeepFeather vs HandBrake.
What KeepFeather does
KeepFeather compresses iPhone Dolby Vision 8.4 with the x265 encoder, carries the metadata of every frame, declares the layer in the file, keeps the scene brightness track, checks that macOS can play the copy, and then decodes the original and the copy and compares the Dolby Vision metadata frame by frame. A copy that doesn’t match is never imported into Photos.
It only does this for the case it can verify: iPhone-style profile 8.4. Dolby Vision from other sources, such as profile 5 or 8.1, would lose its Dolby Vision layer, so those clips are skipped unless you turn them on. HDR clips that aren’t Dolby Vision stay 10-bit HDR with their original transfer curve.
You can see the difference on your own clips before you commit: Compare plays the original next to up to five compressed versions, in sync, with the same brightness handling as the final copy.