• State
  • Locked Locked
  • Replies 3 replies
  • Subscribers 98 subscribers
  • Views 203 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.

TDA4AL-Q1: Highlight around the blue edge

wei.pan
Prodigy 190 points
Part Number: TDA4AL-Q1

Hi TI experts,

  We have a problem with highlight around the blue edge,which is based on TI SDK 8.06、J721S2 and DCC tool v2.5. Our image sensor is OX08B40.

As shown in the following figure, the blue edge around the outdoor scene highlighting is very serious. Viewing the raw image, it is found that there is a blue edge in the high light of the raw image,and the related yuv and raw images have been uploaded to the compressed file

  

The blue edge problem seems to be due to chromatic aberration,can the following two modules optimize this problem?

1.LTM

2.CAC

8306.yuv&raw.zip

Please help us to sovle these cases
Best Regards

  • 0
    TI__Guru 59250 points

    Hi wei.pan,

    Sorry about the delay in my response.
    I missed this thread in the past week.

    Are the companding knee points below correct for the share raw image?
    What is the black level after decompanding?

    0 0
2047 2047
2175 3071
2303 4095
2431 8191
2559 16383
2687 32767
2815 65535
2943 131071
3071 262143
3199 524287
3327 1048575
3455 2097151
3583 4194303
3839 8388607
4095 16777215

    These extreme blue artifacts are not expected in general and I doubt if CA or LTM can cause such bad artifacts on their own.
    In the raw image, I did see those pixels have very strong blue response, e.g., R:G:B ~= 1:1:4.

    Do they happen in both still and moving vehicles?
    i.e., for the shared example, do you see exactly the same blue artifacts when the car moves and stops?

  • 0 in reply to Gang Hua
    Prodigy 190 points

    Hi Gang,

    Gang Hua said:
    Are the companding knee points below correct for the share raw image?

    Yes, it's correct

    Gang Hua said:
    What is the black level after decompanding?

    The black level is 64

    Gang Hua said:
    Do they happen in both still and moving vehicles?
    i.e., for the shared example, do you see exactly the same blue artifacts when the car moves and stops?

    Both still and moving occur, and the phenomenon is the same

  • 0 in reply to wei.pan
    TI__Guru 59250 points

    Thanks for the information!

    I am not sure about the real cause.

    Looking at the linearized 8x8 raw pixels around one blue area starting at 1050x310, I can see that blue channel is consistently 4x~5x as large as other color channels.
    In that case, VISS output would be blue for sure.

     11519  4351 13183  4543 15871  4671  17791  4895 
  5119  4767  5247  4703  5471  4991   5823  5279 
 28287  6143 32511  7071 37375  7711  40447  7487 
  7039  6783  7839  6495  8639  7775   8319  8159 
 45567  7615 46591  9407 50943 10687  56831 11455 
 11135 10943 13823 11071 12351 12799  14527 13951 
 66047 20735 74239 23935 80895 24575 104447 28287 
 34815 30079 32383 24319 27775 28543  42239 47615