Mapping with a stereo camera?

[antithing]

Hi, thank you very much for this code!

I have a scenario where I have mapping data from a pre-calibrated stereo camera, and query images from other non-calibrated cameras.
I need to use the sfm / localization to calculate the transform between the stereo camera sensors and the query camera sensor.

Which of these is the best approach?

A. Map with the stereo camera frames, passing the intrinsic and stereo extrinsic as a constraint, then localize the query frames and read the transform.

B. Map with all three camera frames, passing the intrinsic for the stereo pair.

Also, is it possible to pass the stereo baseline / extrinsic data as a constraint?

Thanks!