CCTV Nightmares: Chain of Custody Secrets from Scene to Courtroom

Deepfake Forensics: Beyond the Deepfake Detection Button

In this episode of the Amped Podcast, Marco Fontani, Forensics Director at Amped Software, and Martino Jerian, CEO and Founder of Amped Software, discuss why deepfake analysis cannot be reduced to a simple “real or fake” button. They explain the difference between deepfake detection and deepfake forensics, why confidence scores are often misunderstood, and why investigators, experts, and courts need a more rigorous and explainable methodology. The conversation covers how forensic analysis can combine detector outputs with metadata, compression traces, pixel statistics, geometry, shadows, perspective, reflections, and other visual clues to reach stronger conclusions. They also discuss the future of synthetic media, including better generators, the “liar’s dividend”, public distrust, and the risk of fake forensic claims being used as propaganda. Listen to this episode and drop a comment if you liked the conversation. Read the blog post about Deepfake Forensics [https://blog.ampedsoftware.com/2025/08/05/deepfake-forensics]. Chapters 00:00 Integrity over shortcuts 00:21 Why detection is oversimplified 02:45 Detection vs forensics 05:08 Beyond real or fake 07:33 Confidence scores 10:02 Don’t oversell tools 12:25 Where detectors help 14:52 Explainability in court 17:12 The liar’s dividend 19:34 Fake forensics as propaganda 21:38 A real forensic workflow 24:05 Source files and laundering 26:28 Creating credible AI images 26:55 Generators and misuse 28:59 Metadata and visual clues 31:23 Video frames as images 33:49 Compression analysis 36:15 Explaining confidence 38:42 Making findings clear 41:05 Final advice for analysts

Courtroom Presentation: Explaining Technical Topics Clearly and Building Demonstratives

In “Courtroom Presentation: Explaining Technical Topics Clearly and Building Demonstratives”, you’ll join Emi Polito and Sam Abbott from Amped Software for a practical conversation on one of the most important - and often underestimated - parts of forensic video analysis: presenting video evidence clearly in court. From early casework mistakes to powerful courtroom demonstratives, Emi and Sam explore how the presentation of video evidence can shape understanding, influence decision-making, and help judges and juries avoid dangerous misinterpretations. They discuss real-world challenges, including low frame rates, compression, lighting, motion blur, timing, 2D perspective, slowed-down footage, still image sequences, annotations, and the limits of what a video analyst should - and should not - say in court. The episode closes with a mock cross-examination, giving listeners a front-row seat to the kinds of questions video experts may face when their work is challenged on the stand. Episode outline: 01:00 - Why evidence presentation matters in court 01:25 - Sam’s first presentation case: 48 hours of footage and multiple DVRs 02:12 - Early lessons learned from building timelines without the right tools 03:05 - Emi’s route from television into forensic video analysis 04:41 - Using broadcast skills in early courtroom presentations 06:46 - The impact of 3D reconstruction and persuasive demonstratives 08:41 - Explaining technical limitations so evidence is not misinterpreted 11:11 - Video playback speed, low frame rates, and the risk of misleading the jury 13:14 - Slowed footage, frozen frames, and maintaining an unbiased presentation 15:17 - Delivery formats: DVDs, compression, and courtroom limitations 17:24 - Using image sequences, PDFs, annotations, and chronologies of events 19:41 - What to do when key contact is not captured between frames 21:40 - Motion blur, frame-by-frame analysis, and simple analytical reporting 23:35 - Staying within your expertise as a video analyst 25:32 - Courtroom challenges: answering clearly without over-explaining 27:21 - Preparing for court when the case is years old 29:10 - Notes, audit trails, and documenting every step of the workflow 31:37 - Jury bundles, illustration packs, and making sure demonstratives are seen 33:49 - Mock cross-examination begins 36:13 - Compression, artifacts, and what the jury is really seeing 38:15 - Lighting, clarity, and evaluating footage quality 40:10 - Contact, obscuration, and limits of interpretation 42:11 - Perspective, timing data, frame rates, and validation

CCTV Nightmares: Chain of Custody Secrets from Scene to Courtroom

In “CCTV Nightmares: Chain of Custody Secrets from Scene to Courtroom”, you’ll join Lucy Carey-Shields (Forensic Analyst at Amped Software) and Blake Sawyer (US Operations Director at Amped Software) as they discuss real chain-of-custody pitfalls they’ve seen in the field: mismatched camera times across multiple locations, systems accidentally reformatted, footage lost after a “simple” time correction, and exports that quietly change the evidence. You’ll also get practical guidance on: * Capturing the most forensically sound version (why proprietary exports matter) * Starting continuity early with hashing and disciplined handling of USB-based collections * Avoiding “quick conversions” that can drop frames or distort timing * The frontline checklist: stop and think, document everything, and do a time check If you collect, process, or present video evidence, this episode gives you a clear, field-tested mindset to protect integrity from scene to courtroom. Episode outline: 00:12 - Official welcome + host intros (Lucy Carey-Shields and Blake Sawyer) 02:41 - How they each got into video forensics 04:45 - Blake’s “trial by fire” case: missing person investigation, large-scale CCTV collection, building a video workflow 06:11 - Real-world chain-of-custody pitfalls from the field 08:35 - Starting continuity at the scene: disciplined USB collections, hashing as a “digital fingerprint”, and clean handoff to storage/server 10:52 - Handling digital evidence like physical evidence: don’t change data, document everything, be competent if changes are unavoidable (UK ACPO principles) 13:16 - The pressure problem: modern casework data volumes and time constraints; why asking the right questions early matters 13:57 - Cloud video (Ring/doorbells): consent vs legal process, exigent circumstances, and getting the “best” version (plus confirmation from the provider) 17:04 - Why proprietary exports matter: least-altered version, stronger integrity, and downstream needs (court playback, public release without “video-of-video”) 19:33 - The danger of “quick conversions” (e.g., FFmpeg): dropped frames, skipped data segments, lost timing; why forensic tools must preserve frame/time metadata 21:58 - Testifying without “nerd talk”: translating technical video issues for humans/juries 23:44 - Training & certification: LEVA vs IAI workflows, what gets tested, and why it changes how you approach analysis/reporting 26:01 - Evidence retention and storage: CDs/tape/cloud tradeoffs, vendor lock-in risk, and long retention timelines 29:09 - Frontline checklist (the “3 simple rules”) 35:54 - Common agency mistakes: detective-managed files, cloud upload authenticity concerns, and the “one flash drive for every case” problem