Mark Ma is the founder and lead engineer at UnMark, where he leads the development of AI-powered video watermark removal technology. With over a decade of experience in computer vision, deep learning, and video processing, he has built production systems that process millions of frames per month across TikTok, Instagram Reels, and Google Veo content. His technical work focuses on motion-compensated watermark detection, spatial-temporal inpainting, and quality-preserving reconstruction for short-form vertical video. Before UnMark, Mark shipped computer vision infrastructure at scale for content moderation and ad compliance platforms. He writes this blog to document the real engineering, testing, and regulatory reasoning behind UnMark's processing pipeline — every test result, frame analysis, and benchmark published here is reproduced from actual production data, not synthetic examples. His goal is to give creators, brands, and regulators a verifiable technical reference for AI video watermark removal in 2026.
Real test of watermark removal on a Google Veo AI-generated video. Frame-by-frame comparison with actual data: resolution, bitrate, and quality before/after.
Agency video watermark removal is a rights-chain audit: stock previews become licensed masters, inherited proofs are claims, and review marks never ship.
Real estate video watermark removal is mostly compliance work: MLS feeds want the unbranded cut, social wants the branded one, and the proof mark is an invoice.
The generic tool lists miss the point: most podcast watermarks die in the export panel, and the clean master already sits on the recording platform.
The SERP sells watermark removal to course creators. The real issues run the other way: platform anti-piracy marks, your own deterrents, and the archive habit.
A high SSIM score and a clean export are not the same thing. What PSNR, SSIM, LPIPS and FVD measure, where each misleads, and how a remover gates quality.
Mask size decides watermark removal quality: the repair-area economics of bounding boxes, pixel-precise selections, and dilation, with a 30-second check.
The real reason watermark removal flickers — and the per-frame, optical-flow, and DiT mechanics behind stable results, with a 60-second self-check.
The 2026 video inpainting landscape, from ProPainter's optical flow to one-step diffusion — and what each model generation means for your watermark removal.
Loom is the inverted watermark case: clean pixels, a branded page, a download gate. Plan prices, the admin toggles, the re-record fix — checked in 2026.
Hailuo's watermark is the easy case — a corner stamp that holds still. The plan gate that lifts it, the welcome credits that die in three days, and the fix.
Kling's watermark drifts across the frame, and credits cannot lift it. Which plan removes the stamp, and how tracked inpainting cleans stranded files.
