• State Resolved
  • Locked Locked
  • Replies 4 replies
  • Subscribers 98 subscribers
  • Views 495 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:

TDA4VM: ISP

TDA4VM: Shifting RAW10 using the decompanding block does not work.

Spencer Hill
Prodigy 110 points

Part Number: TDA4VM

Hello,
We are attempting to integrate a RAW10 Bayer camera into the TDA4VM PSDK (8.02). The camera is up and streaming, however, the video is being interpreted as 12 bit despite being configured as {TIVX_RAW_IMAGE_16_BIT, 9}

We have followed the settings to create a decompand block for masking the 12 bits back down to 10 as described in the e2e here, however this seems to not cause any change to the visual (or YUV saved file) output on the TDA4.

The decompand block is set up as disabled (with the patch applied so that mask and shift are loaded even if the decompand is not enabled), a mask of 0x3ff, and a shift of 0.

If you have any advice for getting a 10bit camera up and running it would be greatly appreciated.

Thanks,
Spencer Hill

  • 0
    TI__Guru**** 491605 points

    Hi,

    Did not get it, what do you mean "The camera is up and streaming, however, the video is being interpreted as 12 bit despite being configured as {TIVX_RAW_IMAGE_16_BIT, 9}" ?

    In SDK8.2 release, i see the OpenVX node supports RAW10 format, so above settings looks correct. What do you get in the output?

    Regards,

    Brijesh 

  • 0 in reply to Brijesh Jadav
    Prodigy 110 points

    Hi,
    Sorry I was not very clear about what the issue is.
    The TDA4 software appears to be interpreting the data as 12bit, this is causing the image that gets displayed on the screen to be incredibly dim. Shining a flashlight into the camera produces a grey spot with a Y value of 39 out of 128 (viewed on vooya), despite the fact that the RAW capture shows a white spot composed of maxed out (1023) 10 bit values.

    Thanks,
    Spencer

  • +1 in reply to Spencer Hill
    TI__Guru 63870 points

    Hi Spencer,

    You may refer to "imaging/sensor_drv/src/imx219/dcc_xmls/linear_10b/" in PSDK for the settings of IMX219 (a 10-bit linear sensor).
    I typically move the 10-bit input from sensor to the 10MSBs of the 16-bit pipeline by the decompanding block.

    If your sensor input is already on the 10MSBs, you may skip the decompanding step.

  • 0 in reply to Gang Hua
    Prodigy 110 points

    Using the lut to do the conversion instead of the mask within the decompand block seems to have solved the problem. Thank you.