Hi,
I'm working with a stereo camera (two cameras with fixed relative positions, registering synchronized images). This system also runs SLAM (simultaneous localization and mapping) module, so I have an initial estimation of the poses for each stereo pair.
How can I load this information to ContextCapture? Ideally, I'd hope to use the known relative poses between the views in each stereo pair as fixed parameter and poses from SLAM as initial guess for optimisation. The latter should speed up the processing but the former should allow for building model with accurate scale. In the GUI I can see an option to use multi-camera rig, but I cannot find any place to load the setup parameters (relative poses of the cameras in the rig).
I'd appreciate any help with this.
Kind regards,
Tomasz
Contextcapture doesn't support stereophotogrammetry as minimum number of photos are 3. Didn't receive any low-level settings which would enable that. There was Agisoft StereoScan but it is also no longer available so the niche is free. Ability to use stereo cameras are missing feature and I would expect less noisy models from CC as from Agisoft. I totally agree that the more parameters we can input the more accurate models we get in output the automatic calibration not always work as expected.
Hi Oto,Thanks for the answer. I don't really need sterophotogrammetry - it's fine to use as many images as it wants to, but if the fixed orientation of the stereo cameras is taken into account, the model will be reconstructed up to scale (i.e. you would be able to take accurate meaurements without scaling the model first). Agisoft Metashape has this option but I'm looking for an alternative to their software.
Can you tell how it is named in agisoft?
agisoft.freshdesk.com/.../31000160468-how-to-process-an-image-group-captured-with-leica-blk3d
you have import both camera image, into two different photogroup (cc should automaticaly divide). Once selected one photogroup (do it for each photogroup) on the photo tab on the right set multicamera rig "main" (and more probably you have to use camera model type fisheye).
for the position you should use the import wizard.