Enhancement an examiner can re-run.
Most forensic image tools are black boxes that output a confident result you cannot reproduce. Obsign does the opposite: a deterministic, reversible, signed operation and a report another examiner re-runs bit-for-bit -- the reproducibility forensic science is supposed to have.
Bit-for-bit, on any machine
The enhancement runs in a fixed-point integer mode; re-run the receipt and you get the identical pixels and the original back. No "comparably trained examiner, comparable result" hand-waving -- the same result, exactly.
Facts vs. calibrated evidence
With a receipt, Obsign knows exactly what changed. Without one, it localizes traces and reports calibrated evidence with explicit limits -- never a verdict. Deterministic findings are facts; statistical ones are evidence; only a receipt is proof.
Signed, re-runnable report
A chain-of-custody report shaped for Daubert / FRE 702: file hashes, tool version, pipeline, findings, limits, an Ed25519 signature, and the exact re-run command. Works on images and video, frame by frame.
See it on your own file
Run the reproducible forensic examiner, verify a signed receipt, or read a self-verifying certificate -- all on your device, nothing uploaded.
Honest scope: the court rules on admissibility -- Obsign confirms the technical criteria (testability, reproducibility), it cannot promise the verdict. Receipt-bound analysis is proof; bare-file statistical detectors are calibrated evidence that can false-positive and are not yet calibrated against a large labeled corpus (validation roadmap on request). Absence of a finding is never authenticity.
Validate it in your lab
We take a small number of forensic design partners. A 30-minute call, a use-case review, SDK access, and a pilot with a validation protocol and examiner-report templates.