Petrify: scrubbing people out of footage, all on your own rig
Petrify spots and flatlines people out of fixed-camera footage while the live timestamp keeps ticking, running fully on your own rig with no corpo cloud upload.
Petrify is an open-source computer-vision tool that spots people in fixed-camera footage and wipes them, while the live timestamp keeps ticking. The whole pipeline runs on your own rig. Nothing gets jacked up to some corpo cloud service, which is the whole point, choom: the footage you most want to strip people from is usually the footage you least want to send anywhere.
Petrify is not only about privacy. The same person-removal serves a spread of fixed-camera jobs, each keeping what matters and flatlining what does not.
What problem does Petrify solve?
A fixed camera pointed at a space records two things at once: the space, and whatever choom happens to walk through it. For a lot of uses you only want the first. You want to know how a room is used, what the background looks like, or whether a hazard is sitting there, without keeping identifiable footage of every choom who passed by.
The usual gonk answer is to ship your video off to a corpo cloud API for redaction, which trades one privacy problem for another and piles on cost and latency. Petrify keeps everything local on your rig. The footage never leaves the machine, so privacy compliance is the default ICE instead of a feature you bolt on. It is one of the Starside Labs projects built around on-device processing, no megacorp middleman.
How does the pipeline work?
Petrify is a straight-shooting pipeline built on a focused stack, like a netrunner’s clean loadout: Python driving OpenCV and NumPy for the vision work, and FFmpeg for decode and encode. It runs accelerated on Apple Silicon and on NVIDIA chrome, so a laptop or a beefier rig can chew through real footage rather than a toy clip.
The stages are simple to reason about:
- Input frames. FFmpeg decodes the source video into frames.
- Person detection. Each frame gets scanned for people, like a daemon sweeping the feed.
- Background fill. Detected people are flatlined and the region is filled from the static background the fixed camera provides.
- Timestamp overlay. The live timestamp is kept or re-rendered so the footage stays usable as a record, preem for any timeline check.
- Output. Frames are re-encoded back into video.
Because the camera is fixed, the background is stable, which is what makes clean fills possible. The static scene is the asset, the people are the noise the crew wants gone.
Who is this for?
The on-device design opens up a spread of uses that a corpo cloud tool makes awkward. Privacy compliance is the obvious one: you keep a usable record of a space without hanging on to footage of people. Beyond that, wiping the moving foreground makes footage compress far better, since a near-static scene is cheap to encode, saving eddies on storage. The same clean background supports space analysis (how a room or floor is actually used), hazard detection against a known-empty baseline, and training-data prep where you want scenes without incidental chooms in them. Preem for any crew that guards its data.
Try it
Petrify is open source, choom. The project page has the overview, the code is on GitHub under Starside Labs, and the Starside listing sits alongside the other on-device rigs. If you run fixed cameras and have ever wished you could keep the scene and flatline the people without shipping video to some other choom’s server, this is built for exactly that, nova and clean. More work like it is under projects.