mirror-based photogrammetry – Thomas Megel’s OpenScan Mini experiment uses a mirror as a turntable to photograph an object’s top and underside in one go, producing a single scan that includes both the object and its mirror image—an approach aimed at reducing the gaps that usually come with
On a normal day, a 3D scan fails in the same predictable way: the camera can only see what it can see. If the underside is blocked, shadowed, or simply too hard to reach, photogrammetry leaves holes.
Thomas Megel is trying to cut out that frustration with a setup that feels almost like a magic trick. His experiment uses a mirror as the turntable platform, letting a camera capture the underside of an object at the same time it images the top.
The proof-of-concept centers on a small tabletop gaming mini perched on a mirror. The mirror serves as the turntable platform inside Megel’s self-designed OpenScan Mini machine, which is built to take highly structured photos of small objects for scanning purposes.
In the end, Megel isn’t just collecting two separate scans and stitching them later. The process produces a single scan file containing two views at once: the original mini and its mirror image.
That matters because aligning separate models and combining them is a common workaround for partial or incomplete scans. The new idea is to get two scans at once—without repositioning the object in between—then rely on what the system already knows to do the separation and alignment. With the mirror plane known exactly. Megel says it should be possible for the software to automatically separate. align. and combine the two components.
As a proof of concept, the results are described as encouraging. Megel is still “playing with the idea” and looking for suggestions from others, including feedback on what could make the approach work even better.
Photogrammetry, of course, is nothing new in the broad sense. It involves taking lots of photographs and using software to turn them into a 3D model. and it can be done with little more than a mobile phone’s camera. But the core problem Megel is targeting—getting around the limits of camera visibility—remains familiar. and the mirror-based workaround is a direct attempt to change what’s possible in the next scan.
3D scanning photogrammetry mirror turntable OpenScan Mini Thomas Megel computer vision hobby scanning
So it’s like a mirror selfie but for 3D printing? Honestly sounds kinda pointless unless it fixes the holes.
I don’t get it… does the mirror make it “see” the underside or is it just guessing with AI? Like it can’t physically see through stuff, right?
reply_to: 1 I think the trick is the camera angle, not the mirror. Mirrors always mess with reflections though, so how do they know what’s real vs the reflection copy? Also “single scan file” sounds like marketing, I’ve had stitching fail way too many times.
This reminds me of those magic trick things where it shows both sides at once. If it works automatically separating the mirror image, that’s cool I guess. But if the mirror is “known exactly” then what about cheap mirrors with weird coatings? Like my bathroom mirror is not exactly calibrated lol.