Hi!

I am trying to export the camSchnappr data so it can be used in another other 3d software.

There are a some older topics on this but it looks like no one has figured this out yet.

It seems to be bit more complicated than just extracting the values from the extrinsic and intrinsic data tables, especially since the intrinsics seem to represent a projection matrix and not just camera paramaters. I was hoping that anyone with some openCV knowledge or maybe the creator of camSchnappr could help with that.

Thanks

Felix

@malcolm, I saw your reply on another post regarding this topic. Do you have any idea? :]

I finally figured it out…

I wrote a python script that extracts translation and rotation from the extrinsics.

The other parameters can be found in the raw intrinsics dat.

FOV looks incorrect at first, but this is because it’s beeing calculated from the render size (e.g. 1920). Using this value as aperture width compensates the high FOV. Alternatively the width can be divided by a more suitable aperture value and the FOV can be divided by this quotient. For example: FOV/ (1920 / 35)

And then there is the principal point which has to be normalized on X/Y by dividing it by the render size (e.g. 1920x1080)

Depending on the software it might be necessary to add or subtract 0.5 or invert the sign…