Data Recovery Case File · Cameras, Drones & Cards · When the Arithmetic Does Work
Half the Card Is Still Untouched
Her enquiry did the sum and reached a conclusion that is, this time, correct. "I unfortunately formatted a 32GB microSD card and then used it to shoot about 17GB of data. I believe this makes it impossible to restore the previous footage as it rewrites, but I'm wondering if there's a chance of restoring the 10–15GB that could still be lingering? I have had no success with" two well-known consumer tools. Her instinct is right and her estimate is close. Roughly half that card has been written over and roughly half has not — and the untouched half is entirely recoverable, which is a considerably better position than the message expects.
| Media | 32GB microSD card — formatted in error and subsequently written with approximately 17GB of new video; earlier footage sought from the unwritten remainder |
| Reported situation | Card formatted accidentally · approximately 17GB of new footage recorded afterwards · earlier footage required · consumer recovery applications attempted without success · card retained |
| Fault class | Partial sequential overwrite following a quick format — earlier region replaced; later region intact and carvable |
| Equipment used | Card imaged write-blocked (DeepSpar USB Stabilizer 10Gb) · written extent measured · format-aware carving for video containers across the untouched region · clips validated by playback and reported by date range |
The decode: why the sum works here, and what the boundary costs
Why a format left everything in place: an in-camera or quick format writes a fresh empty index and marks the data area as available. It does not erase. Every frame of her earlier footage remained physically present, unlisted rather than deleted — which is why the question is about what has since been written over rather than about what the format destroyed.
Why the arithmetic holds this time: after a format the card looks entirely empty, so the camera begins writing at the start of the free space and continues sequentially. Seventeen gigabytes written onto a thirty-two gigabyte card therefore occupies roughly the first half, leaving roughly the back half untouched. That is a genuinely large region and it is exactly where the older footage from later in the card's life sits. Where this reasoning fails is when only a handful of files have been written — a few frames still land at the very beginning and take out the earliest material, so the quantity is small but the position is unfortunate. Her case is the opposite: a substantial, sequential write with a clear boundary.
What the boundary costs: the honest complication. Video files are large, and a clip that straddled the point where the new footage stopped will be partly overwritten — so there is usually a band of damaged material at the boundary rather than a clean line. Cameras also do not always write perfectly contiguously once a card has been used for a while. Expect, therefore: earlier material gone, later material intact, and a handful of clips at the join that are partially recoverable.
Why the consumer tools found nothing: worth explaining, because it is discouraging and misleading. Most of them look for a filesystem to reconstruct, and hers was replaced by the format. What is needed instead is format-aware carving — scanning for the distinctive structure of video containers directly and rebuilding clips from their own headers, without reference to any index. That is a different technique, not a harder-working version of the same one.
The instruction: stop using the card entirely. Every further second of recording extends the written region into the half that still holds her footage.
On the bench
The card was imaged write-blocked behind the DeepSpar USB Stabilizer 10Gb, and the written extent measured first — establishing exactly where the new footage stopped, since that single boundary determines what can be expected and where the damaged band sits. Format-aware carving then ran across the untouched region, rebuilding clips from their own container headers rather than from any index. Every clip was validated by playback, and results were reported by date range so the surviving period was visible rather than expressed as a total.
The outcome
The written boundary measured, the untouched region carved and the surviving footage validated and reported by date. Recovery of deleted or overwritten data from memory cards and USB sticks is charged at a flat figure, payable upfront. The decode, for anyone who shot over a formatted card: the sum works when the write is substantial and sequential — a format leaves everything physically present, and a camera then starts at the beginning of the free space, so 17GB onto a 32GB card takes roughly the front half and leaves the back half intact; expect a band of partly damaged clips at the boundary rather than a clean line. Consumer tools find nothing because they look for a filesystem the format replaced, where what is needed is carving by container signature.
Shot new footage over a formatted card
Stop using the card, then do the sum — because it may be more encouraging than you think. A format doesn't erase anything; it writes an empty index and marks the space free, so your older footage stayed exactly where it was. The camera then starts writing at the beginning of that free space and continues in order, so if you've recorded roughly half the card's capacity since, roughly the back half is untouched and fully recoverable. Expect a band of partly damaged clips where the new footage stopped, since video files are large and one may straddle the boundary. And don't be discouraged that consumer software found nothing — most of it looks for a filesystem, which your format replaced, whereas video clips can be rebuilt from their own container headers without any index at all.
Stop recording now — call Newcastle Data Recovery on 0191 406 1051; imaged write-blocked, written boundary measured, untouched region carved by container signature and every clip played through.
Request a quote online →
Our case files are drawn from genuine enquiries received by our laboratory over the past ten years, anonymised to protect client confidentiality. Each one describes the diagnostic and recovery procedure our engineers apply to that fault, using the equipment listed.