I’ve been trying to use OpenCV3 to read and write some video frames on my TX1 using the cuda module, but I wasn’t able to get the camera working (I can see in the CMake output that it had trouble finding gstreamer and Video I/O was disabled, not really sure how to turn it on.)
Before I fiddle too much with that I thought I’d switch back to OpenCV2 and try the L4t OpenCV libraries. I downloaded and installed them, and some basic code ran fine, but when I tried to create a window with OpenGL support I got an error saying the library was built without OpenCV support.
I suppose that would be the answer to my question, but is this really the case? And if so, why? When I compiled OpenCV3 with OpenGL support on the device I had to make the change referenced at the bottom of this post:
Which I thought was a bit strange. Is this still an issue? Commenting out that line in the cuda header did work, and I was able to create and use OpenGL windows fine (although I didn’t have the cudacodec::Video* classes I wanted when I built from source, which is why I decided to try the prebuilt libraries…)