- Ask a related questionWhat is a related question?A related question is a question created from another question. When the related question is created, it will be automatically linked to the original question.
Hello, Mayank,
Now our 0820 sensor works well on TDA4 after patching the AE and pre-AWB codes.
But now I meet another problem when the sensor capturing objects under the sunlight. For example, Fig 1 is the raw image of cars under the sun, and the bright spots looks quite normal; Fig 2 is the yuv image with TI's default ISP parameters, obvious purple edges are observed around the bright spots.
Fig. 1
Fig. 2
To vertify the cause of this problem, we bypass the CCM step to get yuv image (Fig. 3), (to bypass CCM, we set the color corection matrix to [[256, 0, 0],[0, 256, 0],[0, 0, 256]]), but the purple edges still exist.
Fig. 3
Therefore, I guss this problem is caused after raw capture and before CCM. Could it be caused in data bits companding step?
looking forward to your reply.
Jianxiong Hua
The sensor compands the raw data from 20 bits to 12 bits before its outpus, the knee points are listed below (also see the atached file).
In DCC tool, we set the knee points as below, is it correct?
I'm also confused about the setting. In my opinion, two curves are needed to firstly decompanding raw data from 12 bits to 20 bits, and then companding from 20 bits to 16 bits for VISS process. But how can I know the two curves from the DCC knee points setting?
