Check your container and codec
B-Roll Archives currently scans files with these extensions: MOV, MP4, M4V. The extension identifies the container. The codec inside it also determines whether macOS can read the video and the app can prepare previews. An accepted extension alone does not guarantee decoding.
Check Clip metadata in the app for the recorded codec of a readable clip. See metadata and filters for where to find it.
Confirmed and unverified combinations
Recorded checks: October 6, 2026 · Apple Silicon on macOS 26.3.
| Status | Container and codec | Evidence |
|---|---|---|
| Confirmed in testing | MP4 — H.264 | Real footage and public Release acceptance. A supplied 2 h 6 min 1080p sample also passed dense indexing and recovery in the Release testing recorded below; this denser index is included in 0.3.0. |
| Confirmed in testing | MOV — HEVC/H.265 | Previously tested real footage. |
| Confirmed in testing | MOV — H.264 | Generated integration-test clips; metadata and preview extraction. |
| Confirmed in testing | MOV — ProRes 422 | Generated integration-test clips; metadata, rotation and preview extraction. |
| Confirmed in testing | MOV — ProRes 422 HQ | Real 3840 × 2160 footage at about 60 fps; Release-core harness checks of metadata, overview/range sheets, timestamp playback and visual-index pause/resume. |
| Confirmed in testing | MOV — ProRes 422 LT | Supplied 3808 × 2142 codec test-pattern clip at 50 fps; Release-core harness checks of metadata, overview/range sheets, timestamp playback and visual-index pause/resume. |
| Accepted, unverified | MP4 — HEVC/H.265 | One real sample produced overview previews but failed denser range extraction with Cannot Decode; full acceptance has not passed. |
| Accepted, unverified | M4V | The scanner accepts the extension; no recorded media validation. |
| Accepted, unverified | MOV, MP4, M4V — Other codecs and profiles | Other combinations, including other ProRes variants such as Proxy, 4444 and 4444 XQ, have not been verified. |
| Not scanned in v0 | AVI, MKV, MXF, WebM, MTS, M2TS, BRAW, R3D | Outside the current extension allowlist; these files are not discovered. |
On a narrow screen, scroll the table sideways to read the evidence.
“Confirmed in testing” covers the representative samples described in the evidence column. Real-footage checks and generated integration-test clips are identified separately. The generated MOV checks verify metadata and usable preview extraction; they do not establish native playback acceptance for every camera or codec profile.
The ProRes 422 check uses generated footage. ProRes 422 HQ has a real-scene footage check; ProRes 422 LT has a supplied codec test-pattern check. Those newer checks used the Release-core harness on candidate code, rather than an acceptance test of a newly published installer. They do not confirm Proxy, 4444, 4444 XQ, every bit depth, camera implementation or codec profile. M4V remains unverified even though the scanner discovers it.
Observed sample limitations
- An H.264 MP4 sample at 4320 × 7680 and 60 fps timed out during overview extraction. 8K acceptance is unverified; the H.264 confirmation above does not cover this sample.
- An HEVC MP4 sample at 1920 × 1080 and 60 fps produced 12 overview previews, but denser range extraction returned Cannot Decode. Its playback and full visual-index acceptance remain unverified.
The supplied 4K H.264 MP4 codec test-pattern clip passed at its recorded 25 fps; its filename says 50p. Compatibility records use the track metadata. Mixed-case .Mov and .Mp4 extensions were discovered successfully.
Long-recording case study
Included in B-Roll Archives 0.3.0 · checked October 6, 2026. This case study measures the denser visual index on one supplied continuous recording. It is a long-duration example, not a maximum supported duration or a guarantee for every format.
- Recording
- 2 h 6 min 39 s · MP4 / H.264 (avc1)
- Video
- 1920 × 1080 at about 25 fps
- Test Mac
- M4 Mac mini with 24 GiB RAM · macOS 26.3
- Index result
- 1,520 dense samples · 0 failures · 12 overview frames retained
The visual index samples approximately every five seconds. The first 60 samples took 5.94 seconds. After pausing and relaunching, the remaining 1,460 samples completed in 112.06 seconds. These timers include a one-second settling wait and exclude preceding model warm-up. Filesystem caches were warmed; these are observations on this Mac, not a speed guarantee.
| Measurement | Observed peak |
|---|---|
| Native app resumed indexing · peak process RSS | 418.42 MiB |
| Release-core resumed indexing · peak process RSS | 391.50 MiB |
| Release-core resumed indexing · peak physical footprint | 394.72 MiB |
| Harness whole-run peak RSS · includes warm-up and other checks | 568.67 MiB |
Shared decoder/compiler services and system pressure were monitored separately. Sampled memory pressure stayed normal, with no new VM swap-outs during the measured indexing phases. Process RSS and physical footprint are different measurements; they should not be added together. These values exclude some shared services and GPU allocations, so they are not an absolute whole-system RAM cap. The QA catalogue and generated cache occupied 49.37 MiB on disk, excluding the source video and packaged models.
Pause, quit, relaunch and resume passed through the native controls. Completed frames and vectors were preserved, and an unchanged rescan reused all cached work. A representative visual query found a matching marine-animal shot at 2:00:06.32; timestamp playback passed and the native app opened that matching moment. The source and private test copy retained their original file hashes.
This check covers one 1080p H.264 recording on macOS 26.3. Other long-form codecs, resolutions, lower-memory Macs, older macOS versions, NAS sources and large multi-clip libraries still need their own acceptance checks. Five-second sampling can miss brief shots, and this single-query result is not a broad search-relevance benchmark. The QA app was an isolated, ad-hoc-signed Release copy. These long-recording measurements were recorded before the separate signing, notarization and installation checks for the public 0.3.0 build 6 installer.
What these checks establish
Validation was performed on Apple Silicon with macOS 26.3. It does not verify every M1-or-later Mac or older macOS releases. See the download page for the current installer and platform requirements.
Formats listed as Not scanned in v0 are outside the extension allowlist. They are not pending clips in the preparation queue. No additional decoders or media support are added by this guide.
If your footage fails preparation
An unreadable codec is one possible cause. A disconnected source, missing permission, incomplete copy or corrupt file can also prevent preparation. Check the per-file status and try a known-readable clip from a fully copied, connected source.
See troubleshooting for recovery steps or contact support with your app/macOS version, file extension, recorded codec if available and exact error. Do not send private footage or licence keys. For common setup questions, start with the FAQ.