Face Restoration for Legal & eDiscovery
What face restoration does for legal & ediscovery footage, when it helps, and how BetterVideo's API applies it to the real capture.
🔒 Footage deleted on your schedule — with a certificate to prove it
- Enhancement can't invent detail — anything that fabricates content is worthless as evidence.
- Privileged and case footage has to stay confidential and never feed someone's AI model.
- Retention windows matter: material often must be deleted once a matter closes.
Where face restoration helps
Faces in exported surveillance and body-cam footage are often small, blurred and heavily compressed. GFPGAN reconstructs facial detail so a face is more legible for review. It's important to be exact about what this is: the model reconstructs plausible detail from the degraded capture — it does not recognise or identify anyone. The restored frame supplements the original for review; it is never identification-grade proof.
How BetterVideo does it
BetterVideo applies face restoration through one REST call — GFPGAN v1.4 reconstructs degraded faces so features are more legible — and for legal & ediscovery, enhances real footage without inventing detail, never trains on it, and deletes it on your schedule with a certificate to prove it.
Common questions
Can face restoration identify someone in footage?
No. GFPGAN reconstructs plausible facial detail to make a face more legible — it is not facial recognition or identification, and a reconstructed face must never be relied on as proof of identity. Treat it as a review aid alongside the unmodified original.
Does it change the person's face?
It reconstructs plausible detail from the real capture; because the model infers detail it can't see, keep the unmodified original as the source of truth and use the restored version only to review.
Face Restoration for your footage
One REST call to enhance — deleted on your schedule, with a certificate to prove it.
Get your API key →