• State Resolved
  • Locked Locked
  • Replies 3 replies
  • Subscribers 97 subscribers
  • Views 482 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:

Linux/TDA2E-17: target DDR3 memory size of 3D SV DEMO in Vision SDK

RTOS/TDA3MV: 3D performance and feasibility of TDA3 for 3D SRV

Sung-IL
Mastermind 30850 points
Part Number: TDA3MV
Other Parts Discussed in Thread: TDA2E

Tool/software: TI-RTOS

Dear Champs,

My customer is revisiting on our TDA3 device for 3D SRV as they decided to use internal ISP of SoC, and 4ch camera will be attached to TDA3.

But, my customer is still worrying about the 3D performance and feasibility of TDA3 running on DSP, and wants to know 3D SRV features can be implemented with similar quality with GPU of TDA2E.

Could you please check if below target requirement can be implemented on TDA3 as well?

Is there any limitation in the 3D SRV on TDA3 compared with GPU based solution like TDA2E?

Target requirements are

- 4ch. 1MP camera(LVDS)

- 2MP display after merging 4ch input image

- transparent mode

- Auto SRV Calibration 

Thanks and Best Regards,

SI.

  • 0
    TI__Genius 12455 points

    Hi SI,

    The 3D SRV on TDA3x relies on SIMCOP and DSP to stitch 3D image vs GPU on TDA2Ex.

    On Resolution : Yes, we can support 4 1.2 MP camera with 2MP display. 

    On Transparent mode : The car image is applied by the DSP and in our reference implementation the CAR image is not blended but just copied. With simple blending transparency could be achieved.

    On auto calibration : Yes, in our reference implementation, we do support auto-calibration. We have a separate usecase for the same. Calibration is a chart based calibration, where the CAR is placed on a fixed pattern chart and calibration routine would look for this pattern to generate calibration information.

    Couple of questions though

    1. When you output is 2MP, is it the size of display screen or the stitched output.

    2. Auto calibration : Would this be required at run-time or would this be a done in the factory

    Regards, Sujith

  • 0 in reply to Sujith
    TI__Mastermind 30850 points

    Hi Sujith,

    Thanks for your immediate response.

    My answers are

    1. I mean the 2MP output is the size of display screen.

    2. Auto calibration should be required at run-time.

    Thanks and Best Regards,

    SI.

  • 0 in reply to Sung-IL
    TI__Genius 12455 points

    Hi SI,

    Okay 2 MP display should be okay. Our demo application demonstrate it.

    Out demo calibration is for static calibration (done once and not rerun while SRV is running). Customer could modify to suite there needs.

    Regards, Sujith