Recovery methodsTechnical

Recovering Fragmented Video from Memory Cards

A large camera recording can span many noncontiguous filesystem extents — recovering only the first file signature or one contiguous region can produce a clip that looks plausible at the start and fails partway through.

Large recordings fragment on FAT/exFAT

Long video recordings on camera and drone memory cards are frequently many gigabytes in size, and on FAT/exFAT they can become fragmented across the card due to prior deletions, the layout of remaining free space, the presence of multiple earlier clips, and the specific allocation behavior of the camera's own recording software. Fragmentation is the normal state for a long recording on a well-used card, not an unusual edge case.

Signature carving stops at the first gap

A signature carver typically recovers the file's start plus whatever contiguous region follows the matched header, but has no reliable way to find later fragments once the recording's data stops being contiguous. The resulting output can have a completely valid-looking beginning and still fail partway through playback, because the carver simply had no path to the remaining fragments once contiguity broke.

Filesystem metadata should be tried before falling back to carving

The safer recovery order runs from recovering the directory or file record, to recovering allocation information and extents, to extracting the exact video file those extents describe, to repairing the container afterward if needed — and only then, as a last resort, attempting raw fragment reconstruction directly from unallocated space. This is the same principle covered in File-System Recovery vs. File Carving, applied specifically to the large, easily fragmented files that video recordings produce.

Deleted directory records can still describe the true extent layout

Even after deletion, a surviving directory record or its equivalent allocation metadata can describe exactly which clusters or extents made up the original recording, in the correct order — information no signature carver has any way to reconstruct on its own. Recovering that metadata first, before falling back to carving, is what turns a fragmented recording into a correctly reassembled file rather than a plausible-looking fragment.

Codec and frame markers are secondary, corroborating evidence

Internal codec or frame markers — recognizable NAL unit start codes, frame boundaries, or similar structural cues within the compressed stream — can help validate or locate fragment boundaries once filesystem metadata has narrowed down where to look, but they are secondary evidence: they help confirm a boundary that allocation metadata already suggested, not a substitute for that metadata when it is available.

Camera-specific allocation behavior matters

Different camera and drone manufacturers allocate storage for ongoing recordings differently — some preallocate a large contiguous region up front, others grow the file incrementally, and some split a single long recording into multiple linked files entirely. Knowing the specific camera model and its typical recording behavior can meaningfully change which recovery approach is likely to succeed and roughly how fragmented a given recording is likely to actually be.

A reference recording documents expected behavior, on a different card

A same-camera, same-settings reference clip can help establish the expected container structure and allocation pattern for later analysis, exactly as in general MP4/MOV container repair — but that reference clip should always be recorded onto a different card, never onto the card that still needs to be recovered.

Combine filesystem reconstruction with container repair

A fragmented recording frequently needs both stages in sequence: filesystem-level reconstruction to correctly reassemble the file's extents in the right order, followed by container-level repair if the recording was also interrupted before finalization. Treating either stage as sufficient on its own, without the other, is a common way to end up with an incompletely recovered clip that still will not play cleanly.

Stop writing to the card immediately

Every additional write to the card — a new recording, the camera's own re-indexing, or an attempted repair run directly on the card — is a chance to overwrite exactly the fragments a proper reconstruction would still need.