AM62A3: About the demosaic algorithm used by TDA4/AM62A

Hi Ding,

Do you have some sample images to share?

You may find CFAI information in TDA4 TRM in the FCP section.
It is gradient based interpolation.

pic.zip

Hi Gang,

Here is the picture, thanks.

Hi Ding,

What's the image resolution?

Thanks,

Jianzhong

Hi Jianzhong,

The image resolution is 1920 x 1080, and the bit depth of the raw image is 12.

Best,

Hi Ding,

Could you please also share the sensor companding knee points?

0    0
1023    1023
1279    2047
1535    4095
2303    16383
2559    24575
2943    49151
3071    65535
3327    131071
3455    196607
3519    262143
3647    524287
3775    1048575
3903    2097151
3999    4194303
4095    16777215
65535    16777215

Hi Gang,

Here is the sensor companding knee points.

In addition, please tell me what other data you need before 18:00 on February 7, Beijing time. I will take my Spring Festival holiday later than this time.

Best

Hi Ding,

Ding Y said:

What is the algorithm used by AM62A/TDA4 to do demosaic? Could the cause of the purple circle be related to this?

It looks like the issue is much more related to GLBCE than CFAI.

The purple edge is present after GLBCE and before CFAI.

Hi Gang,

Since this only happens when the bilinear interpolation de-mosaicing algorithm is used at very high contrast. I wonder if it's possible that it's the contrast enhancement by GLBCE that's causing CFAI to cause the purple edge problem. In other words, the disappearance of the purple edge problem when GLBCE is turned off is the result of a low local-contrast factor.

Best