• State TI Thinks Resolved
  • Locked Locked
  • Replies 3 replies
  • Answers 1 answer
  • Subscribers 98 subscribers
  • Views 186 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

Original question:

TDA4VH-Q1: Verify the LUT decompanding formula

TDA4VH-Q1: CCM innacurate in the Implementation

Dinesh bharathi
Intellectual 751 points
Part Number: TDA4VH-Q1
Other Parts Discussed in Thread: TDA4VH

Hi Team, 

I have created the CCM using DCC tuning tool for a raw image, the matrix generated shows mere results in the dcc tuning but the same works in the pc emulation/tda4vh bench with a green color cast on the output. Can you please help me around this.

ti_ccm.zip

Regards,
Dinesh

  • 0
    TI__Guru 60290 points

    Hi Dinesh,

    In DCC.png, there seems some issue with WDR decompanding.
    Do you input the WDR knee points properly in tuning tool?

  • 0 in reply to Gang Hua
    Intellectual 751 points

    Oh didnt use it initially as it was optional. The default knee points am not sure about, so i have changed the knee points :

    Fullscreen knee_decompand_default_py_imager_based_LUT.txt Download 
    0	0
512	512
837	1024
1162	2048
1487	4096
1812	8192
2137	16384
2462	32768
2787	65536
3112	131072
3437	262144
3762	524288
4087	1048575
    :

    I have used the black level as 168 as given in the python dcc generation code, but still some thing is off. Am verifying the sensor format and everything. Looks everything is matching but the output image is more darker and there is a weird clipping when i turn off the reverse perceptual LUT which i always do.

    This is the updated image with new companding formula as mentioned above: 
    img_00000002.zip
    This data looks a bit like linear data, doesnt seem to have dynamic range as WDR. But am verifying everything possible. Please let me know if am taking any wrong directions here

  • 0 in reply to Dinesh bharathi
    TI__Guru 60290 points

    Hi Dinesh,

    Dinesh bharathi said:
    I have created the CCM using DCC tuning tool for a raw image, the matrix generated shows mere results in the dcc tuning

    To get good CCM, you would need to input the correct knee points so that tuning tool can work properly.

    Dinesh bharathi said:
    but the same works in the pc emulation/tda4vh bench with a green color cast on the output.

    I am not sure what you mean exactly here -- probably the green cast in Output.png.
    That is an AWB issue (not CCM).
    You will have to calibration AWB to get rid of the color cast.

    DCC.png clearly has some decompanding issue and Output.png looks better in terms of decompanding (but not sure if decompanding is 100% correct).