naushir
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 136
Joined: Mon Apr 25, 2016 10:21 am

Re: Stereoscopic camera capture (for COMPUTE MODULE) - now implemented (2014).

Thu Mar 25, 2021 2:06 pm

I was pointed to this thread by a colleague, sorry it has not been replied to.

A possible cause for this might be replacing the AWB algorithm with a new in-house implementation. I do not believe the new algorithm accounts for (or been tested with) stereo setups. If this is indeed the case, you can switch back to the old algorithm by running the following command command before starting the camera app:

Code: Select all

vcdbg set awb_mode 0

User avatar
Realizator
Posts: 81
Joined: Thu Jul 14, 2016 12:53 pm
Contact: Website Twitter

Re: Stereoscopic camera capture (for COMPUTE MODULE) - now implemented (2014).

Fri Apr 09, 2021 6:36 pm

naushir wrote:
Thu Mar 25, 2021 2:06 pm
I was pointed to this thread by a colleague, sorry it has not been replied to.

A possible cause for this might be replacing the AWB algorithm with a new in-house implementation. I do not believe the new algorithm accounts for (or been tested with) stereo setups. If this is indeed the case, you can switch back to the old algorithm by running the following command command before starting the camera app:

Code: Select all

vcdbg set awb_mode 0
Hi Naushir!
Our tests show no difference between the old and new AWB algorithms.
We also noticed two things:
1. This effect often appears when we have both extra dark and extra light areas on the image. For example, with wide-angle (fisheye) cameras, we always have dark corners on the image.
2. One of our users noticed that this effect also visible on the video. But, if you decrease FPS to 10 or 5, this effect slowly disappears in 20-30 seconds. It looks like the slave camera has not enough time to do appropriate measurements.

In any case, the master camera always works fine. Do you have other suggestions to test?

Return to “Camera board”