Hi everyone,
Having troubles with the Camera export again. Adding gcps for scale etc… seems to break the camera export function for maya files:
Without GCPs:
Screen Shot 2016-05-31 at 17.12.52.jpg
Same project with GCPs:
Screen Shot 2016-05-31 at 17.13.25.jpg
Some of the cameras appear to be in the correct position, but the wrong orientation, but the majority seem to be no where in space. See below camera position values:
Screen Shot 2016-05-31 at 17.20.09.jpg
Screen Shot 2016-05-31 at 17.20.14.jpg
The values for those cameras are normal without using GCPs.
It looks like GCPs does not break the export via bundler .out files. However using this format seems to break the connection for scale and position between the model and the cameras.
Am I doing it wrong? Is there a solution for this?
Hi Joseph,
thanks for posting. We will check it. Looks like a regression caused by the latest update.
Any solution for this - I’m seeing the same -infinity rotation etc. within maya. (and NaN values)
CSV export shows values of -1IND…
It would be great if the Maya export also had a scale factor as an option.
I was able to get much more consistent results by changing the scale of my constraints. Maya historically doesn’t like extremely small values so scaling everything by 100 gives me a result that works out of the box. I also defined 3 GCPs with coordinates that push the result into Y-up so I have a clean set of cameras in Maya.
Question for Milos is will this affect any noise filtering etc? Are you considering world scale and distances when determining your level of error correction or given it’s float internally is the scale arbitrary with respect to calculation?
As an example:
I have a a calibration distance that is 47.73 mm
I previously set this to 0.04773 in RC to correspond to it’s internal units being meters.
I NOW set this to 47.73 and no NaN or undefined values in Maya.
My results, while not invalid, are not consistent with RC. See the attached screenshot… my camera orientations are all over the place. It looks like it might be some kind of gimbal lock issue or rotation order?
wrong_orient1.JPG
wrong_orient2.JPG
Hi Robert Moggach
Thanx for information, we could reproduce it and we will fix it…
Happy to test a beta when you have it - you have my email.