thomas_OpenScan
thomas_OpenScan on Hacker News
Re:OpenScan
on Sunday February 22, 2026 @09:16AM (#47111178)
Most recent gaussian and nerf to mesh algorithms are surprisingly good at getting reasonable results for objects that traditional photogrammetry would struggle with.
The main challenge are reflective and uniform surfaces (e.g. lether or coated wood). See this overview what you'd want for perfect photogrammetry: https://openscan-org.github.io/OpenScan-Doc/photogrammetry/b... and also the challenging surfaces lower on that site
the Lidar on the phone is only suited for medium/large objects (>50-100cm). but you are absolutely right about the lowest cost hardware option.
The thought behind the scanner is having a self-contained unit which of course increases the cost.
The OpenScanCloud is just a convenience solution and most photogrammetry pipelines will outperform it. Meshroom is a fully open-source alternative (which needs quite some fiddeling...). And EpicGames just release RealityScan free for individual usage which is a state-of-the-art photogrammetry solution. I just created the cloud solution as an alternative for people getting into the topic, which either do not have the right hardware or maybe do not want to deal with yet another program.
scanning furniture is quite a challenge for photogrammetry. your best option would be NERF or Gaussian splatting and manually guiding the camera.
Re:OpenScan Offers 10-Micron 3D Scanning on a Raspberry Pi and HQ Came...
on Friday February 26, 2021 @05:08PM (#26280438)
Thank you and no worries, I absolutely agree with your valid criticism.
Later this year I will get some more CT reference scans to compare my results too. I think that this will give some better insights.
By the way, even on the professional market there seems to be no „universal“ benchmarking procedure to compare the various mid-range scanners (1-20k€). So this is a very interesting discussion point
Re:OpenScan Offers 10-Micron 3D Scanning on a Raspberry Pi and HQ Came...
on Friday February 26, 2021 @05:04PM (#26280403)
100-200€ plus the printed parts (~50€).
But most people diy some or even all parts for even less
Re:OpenScan Offers 10-Micron 3D Scanning on a Raspberry Pi and HQ Came...
on Friday February 26, 2021 @10:23AM (#26275637)
Cyclododecan has been the industry standard for vanishing 3d scanning sprays, but as Cyclododecan is no longer available from the textile industry, this kind of spray does not exist anymore.
Aesub Blue is a rather new product from a young German company. I have met the founders last year and they are really professional and transparent. Just check out their documentation.
To be fair, as the product is so new, possible interactions with materials are still not perfectly understood. But from what I have seen, the feedback from the professional scanning community is very positive.
Re:OpenScan Offers 10-Micron 3D Scanning on a Raspberry Pi and HQ Came...
on Friday February 26, 2021 @10:16AM (#26275565)
Actually, this is how I started the project. See for instance this one here (OpenScan in 2019 https://www.instagram.com/p/Bt6JtjEoEM7/)
But I have found that smartphones give very inconsistent results (probably due to image"improving" algorithms...), see https://www.instagram.com/p/B1q1syMoHfG/
So I moved on to the raspberry pi + camera, which is much easier and better to control as you need very, very consistent images...
Re:OpenScan Offers 10-Micron 3D Scanning on a Raspberry Pi and HQ Came...
on Friday February 26, 2021 @10:14AM (#26275532)
Yes, this is the current public stage. The device is making a bunch of images, creating a zip file that need to be downloaded and manually processed. The main advantage already is, that the images and the lighting is very consistent, which is major key to successful photogrammetry. The upcoming cloud processing would further simplify this process, but as I am doing all that alone (+ help from the community) this is a rather slow progress...
Re:OpenScan Offers 10-Micron 3D Scanning on a Raspberry Pi and HQ Came...
on Friday February 26, 2021 @10:11AM (#26275499)
Hi, this is Thomas, the guy behind OpenScan. I really would like to clarify several things.
- I have never made such a bold claim of 10 micron accuracy for the device. Yes, I made several scans, where the accuracy is quite good, but I would never say, that this can be reached under all circumstances.
- To be fair, I have been kinda click-baity in the past on a handful occasions, but everyone who is doing at least 5mins of research will find for instance this one here: https://en.openscan.eu/quality
- Comparing this 100-200€ device with an industrial 3d scanner is like comparing a hobbyist 3d printer with a highend injection molding machine...
Anyway, feel free to discuss and ask me anything about it :)