• State Resolved
  • Locked Locked
  • Replies 4 replies
  • Subscribers 99 subscribers
  • Views 429 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.

Previewer on the fly

MattLipsey
Genius 3575 points

I am working on using the previewer with input taken directly from the ccdc on a dm6435.  I am getting an output image, but I am not sure if the resulting image width is correct.  I am acquiring an image that is 1280 pixels wide in raw bayer format.  I expect to lose 4 pixels from cfa interpolation in the previewer.  All of the other previewer features have been disabled.  But instead of getting 1276 active pixels out, I am getting 1272 good pixels out (the output line offset is 1280 pixels to meet 32 byte alignment requirements.)  The documentation (spru977d, section 5.5.6) says that I have to set the previewer starting pixel to a minimum of 2 and the ending pixel to at least 2 from the end when the input is taken from the ccdc.  Am I supposed to lose those 4 pixels in addition to the 4 pixels lost to cfa interpolation by the previewer?

  • 0
    Genius 3575 points

    As an additional related question, I noticed that my yuv4:2:2 image output by the previewer looked like it had been divided by 4 (255 input values produced 63 Y component values).  I am assuming that this is due to using the video port data formatter block as part of the input chain to the previewer.  It appears that the data formatter is expecting 10 bit inputs, and I only have 8 bit data coming from the ccdc.  I am guessing that my 8 bit input is getting stuck into the 10 bit path at the least significant bit level, which would explain the divide by 4 output result.  Is there a proper way to set my ccdc -> previewer pipeline to work with 8 bit data, or do I need to find a workaround for this?

  • 0
    TI__Intellectual 1870 points

    You lose 4pixels due to the limitation of SPH and EPH as you said.

    All other reductions (of good data) are depending on the functions (filters) used.

     

    regards,

    Sang-Yong

  • 0 in reply to MattLipsey
    TI__Intellectual 1870 points

    1. If you can use FMTCFG.VPIN, then it is the best. To use this, your connection should not be bottom 8bit.

    2. Otherwise, you can use any Gain in Preview Engine.

        My recommendation is using DGain (2x) and WBGain (2x on top of your WBGain applied).

     

    regards,

    Sang-Yong

  • 0 in reply to SANG-YONG
    Genius 3575 points

    Thanks, that was what I was afraid of.  My connections are on the bottom 8 bits, but some digital gain is an acceptable work around.