Defining coordinate system / origin for camera stations

[Marburg]

Quick question.   We have designed a hemispherical multi-camera rig.   We're using station groups to bundle the photos from each "snapshot" with the rig.   This works great and we are able to generate panos and from each group.

However, one of the cameras/image is mechanically fixed at center of the array and points "forward".   I'd like to be able to define this camera as having 0 pitch/roll/yaw in the coordinate frame of the group.   So far, the origin/orientation for the camera group seems pretty arbitray.   

As a side effect I'd want this center image be at the center of any panos (and be the least distorted).

Is it possible to do this?

[Alexey Pasumansky]

Hello Marburg,

I think that it can be solved in two ways: inputting the coordinate information for some cameras to the Reference pane and then using the same values for the further projects, using Python that will do that automatically.

The first way seems rather simple: using Rotate Object tool orient that model as you wish, then type in the following command to the Console pane:
PhotoScan.app.document.chunk.transform.matrix = PhotoScan.app.document.chunk.transform.matrix
The [R] sign should appear next to the chunk's label in the Workspace pane and in the Estimated tab of the Reference pane you'll be able to see the coordinates for the cameras that can be also exported to the text file and used in the other projects.

[Marburg]

Thanks, I was able to get that to work well enough.   

Is there a way to do it numerically --- that is, explicitly define one of the cameras/images as being 0 pitch / 0 roll / 0 yaw?  Tried manually importing a camera reference file with just one image in it, but that didn't seem to have effect.

[Alexey Pasumansky]

Hello Marburg,

You need to have at least three reference points (camera centers and/or markers) to set up the coordinate system from the Reference pane.