Meshroom for Stereo Photogrammetry? [question]

[Fratorhe]

Hello,

we would like to perform close range stereo photogrammetry. Our objective would be to re-construct (part of) the surface of a sample while it undergoes recession.

I think your software Meshroom could be a good option for us, but I see most of your applications are based on re-building a 3D object from (a lot of) pictures with one camera, while in our case we need 2 fixed cameras to re-build an object and its time evolution.

A similar idea to what we would like to do can be seen in the attached paper.

If this is feasible, we would be very interested in this project.

AIAA2014-2248.pdf

[natowi]

Meshroom can reconstruct objects from only two images, but you need to increase the reconstruction quality to the best settings. In some cases you might only get a sparse point cloud.

As for time evolution... there is no such feature for Meshroom at the moment.

You could generate pointclouds or meshes and then use CloudCompare to compare the differences.

-- https://github.com/AndrewHazelden/CloudCompare4D https://github.com/ArcticSnow/photo4D (micmac implementation, maybe a starting point)

[Fratorhe]

Thanks for your reply. Could you give me some more guidelines on the workflow I should follow? I do not find much information related to this topic... For the moment, let's simplify and not consider the time evolution, this will come in second step. I guess, I should: 1) calibrate both cameras (using a pattern?) 2) calibrate the stereopair (?) 3) introduce my images and recontruct the point cloud (?).

Thanks in advance for your help

[fabiencastan]

@Fratorhe We could setup a confcall to discuss about your use case. You can contact me at fabien.castan[at]mikrosimage.com.

[skinkie]

A small tutorial for this would still be appreciated.

[natowi]

@skinkie A tutorial for what exactly? Camera calibration?

[skinkie]

Single and Stereo Calibration, getting the files with the correct filenames in the image source folders.