One phone on an overhead rig, one take per product, room sound only.
Turn any video into a
shot-by-shot breakdown.
Paste a link and get a production report: what every shot is, whether it was filmed, generated or built in code — plus a shot list, a material list and the steps to remake it.
Comparison footage from the open-source rea-video project (Apache-2.0). Shown for comparison only.
Five references. Five production routes.
Real videos, already taken apart. Each one is a different answer to the same question: what do I actually need in order to make this?
Digital-human lip sync, generated period scenes, still-image pushes.
Two AI characters, five code-built UI segments, one real screen recording.
Plush characters and cutaway animation, rendered rather than filmed.
75% live action, 25% AI clip generated from the creator's own footage.
Preview clips and case data come from the upstream rea-video project; rights stay with the original creators.
Open a full breakdown →Not a transcript. A production blueprint.
Every claim comes with the signal that produced it, so you can trust it or override it.
Source verdict
Per shot: filmed, AI-generated, 3D, code-rendered, screen recording, digital human, or a mix — with evidence and a confidence flag.
Shot grammar
Framing, camera move, lens feel, 12 lighting setups with gear lists, colour grade, cut rhythm and beat alignment.
Packaging layer
Caption-by-caption timeline with colour, stroke and entrance animation, plus titles, stickers and picture-in-picture positions.
Sound map
Human vs synthetic voice, BPM and key, library vs generated music, and which sound effect lands on which visual event.
Evidence trail
Frame rate, sensor noise, metadata AIGC flags and OCR. Conflicts are marked for a human look instead of being guessed away.
Remake plan
Reverse-engineered prompts, a Remotion route for code shots, a shoot list for live action, and an assembly plan on the original timeline.
A shot list
What to film, how to frame it, how long each shot holds.
What you need
Footage, captions, sound, props and reference images, per shot.
Steps to follow
Assets, prompts and the build order for each route.
One reference video. A plan you can follow.
Bring a reference
Paste a link or drop a file. Once the engine opens it will take a local reference and keep the evidence — not just a description of what is on screen.
Review the breakdown
Go shot by shot: framing, camera move, captions, sound cues, source verdict. Uncertainty stays visible instead of being smoothed over.
Leave with a plan
A shot list, a material list and a route for every shot: film it, generate it, or build it in code — laid out on the original timeline.
A report you can actually work from.
This is a static sample. Real analysis is not enabled yet.
REPORT PREVIEW / 00:00–00:42
42 seconds, 18 shots, 9:16 vertical. Mostly code-rendered UI animation with two AI-generated inserts and one filmed close-up. Grade: high-contrast cool teal, acid accent. Sound: 124 BPM synthetic bed with three impact effects.
Open full sample report →
People who need to know how it was done.
Creators
Study a video you admire and turn it into a repeatable shooting and editing checklist.
Editors & agencies
Reverse brief a competitor spot: captions, pacing, sound cues and transitions in one document.
AI video teams
Recover prompts, model choices and first/last frames from generated shots.
Educators
Teach film grammar with a shot-by-shot, evidence-backed breakdown of real work.
We are building the engine. Come see the shape of it.
Create a free account to register your interest. We will email the small group of early users when analysis opens.