It watches your footage. It keeps what matters. And it tells you exactly why.
REAL OUTPUT · 1080×1920 · CAPTIONED · −14 LUFS
scrollEvery amber mark is a decision. Touch one.
| # | In | Len | Phase | What the model saw |
|---|---|---|---|---|
| 01 | 40:58.00 | 3.0s | Hook — A person is plating cooked shrimp and broccoli | A person is plating cooked shrimp and broccoli |
| 02 | 00:24.54 | 5.2s | A person is washing broccoli under running water | A person is washing broccoli under running water |
| 03 | 02:52.09 | 3.9s | A person is mincing garlic on a cutting board | A person is mincing garlic on a cutting board |
| 04 | 06:20.37 | 4.7s | A person is peeling shrimp in a kitchen sink | A person is peeling shrimp in a kitchen sink |
| 05 | 13:18.80 | 6.0s | A person is chopping broccoli on a cutting board | A person is chopping broccoli on a cutting board |
| 06 | 14:58.39 | 4.8s | A person prepares shrimp for cooking | A person prepares shrimp for cooking |
| 07 | 19:46.80 | 6.0s | A person is chopping herbs on a cutting board | A person is chopping herbs on a cutting board |
| 08 | 24:02.33 | 4.6s | A person is squeezing lemon juice into a glass container | A person is squeezing lemon juice into a glass container |
| 09 | 25:27.80 | 6.0s | Pouring oil into a glass on a digital scale | Pouring oil into a glass on a digital scale |
| 10 | 29:06.46 | 5.0s | Shrimp are being cooked in a pan on a stovetop | Shrimp are being cooked in a pan on a stovetop |
| 11 | 32:00.63 | 5.5s | Shrimp are being cooked in a pan on a stovetop | Shrimp are being cooked in a pan on a stovetop |
| 12 | 36:00.68 | 5.6s | Shrimp are being cooked in a pan on a stovetop | Shrimp are being cooked in a pan on a stovetop |
| 13 | 37:36.80 | 2.0s | Shrimp dish being garnished with herbs in a pan on a stovetop | Shrimp dish being garnished with herbs in a pan on a stovetop |
| 14 | 40:58.80 | 6.0s | A person is plating cooked shrimp and broccoli | A person is plating cooked shrimp and broccoli |
We label the footage every couple of seconds — so the clip ends after the reaction instead of on the word, and the crop follows whoever is actually talking.
A vision model labels each sample for action, ingredients, tools and shot type — on hardware we own, so your footage never leaves it.
Each frame is scored on its own — what is in focus, how cleanly it shows one step, how much of the phase agrees with it. The moments that survive are the ones the evidence supports.
Every kept moment carries its reason. And when the footage isn't something we handle well, it says so instead of guessing.
Cut from the plan above. Same 14 shots, same order, no hand edits.
We tune each vertical by hand before it ships. Tell us what you shoot.
We'll come back to you personally — not with a drip sequence.
We store your email, your answer, and — with the request itself — your IP address and browser string. Nothing more: no list sharing, no tracking pixels, no analytics. Delete any time: 112@nectr.video. What we keep and why.
Sign in to see your footage analysed, and tell us what you want made.
Sign inYour video is analysed on our own hardware in the EU and is never sent to a third-party model. We keep it while we are working on it, and delete it on request — how long, and why.