The issue is there for long time. Yes there is a profile you can use/try but most time it is not working (anymore).
I guess that arguscamera module inside jetson platform was upgraded many times that old profile no longer works.
As a third party camera developer we have similar issue as well. I am quite surprised that Nvidia is not willing to open the calibration/optimization procedure/definition of arguscamera platform. they ask you to be a partner to get that info. Basically Nvidia is killing the capability of jetson nano camera function.
Checking out the rival platform, raspberry pi, they are willing to encourage all the 3rd party user/developer to use their camera platform. the libcarmera(https://libcamera.org/) platform they used right now is similar to the arguscamera but currently, in my opinion, it is far more advanced. I try our designed camera on it. Without putting much effort, it gives me pretty good result. the color, white balance, AEC/AGC, black level all things are good even without calibration. they even offered the documents with how these works and how to do the calibration as well(
The big issue with raspberry pi is, since boardcom is no longer making cell phone CPUs, their camera/GPU SOC team is dismissed. they can only support video stream up to 1080p and this situation seems won’t have much improvement in the near future.
Jetson nano, on the other hand, it has much better hardware platform. the B01 version can even support 2 CSI cameras. We cannot ask any more for the hardware. However the pain is from the software, why not open the protocol of arguscamera and let other developers help you to perfect it? Until now there still only 1 camera, IMX219 is really support with the platform.
I really hope that nvidia can be more open to the community. Start from open the calibration/optimization info of arguscamera platform. And nvidia is there any chance that you may join the libcamera platform as well?