TDA2EG: Surround View Display Size Full Screen Issue

Hi Sujith

    I have changed the SRV_OUTPUT_WIDTH_TDA2X_2D from 880 to 1280, and got the error as:

    "[DSP1  ]     29.770481 s: DMA block size doesnot match Photo Alginmnet block size"

    VSDK just provides Surround View's algorithm's lib file, not the source code file. So it is hard to debug this error's cause.

Thanks

xuanbo

Hi Xuanbo,

On the algorithm assert, let me check with my team members and get back to you.

Please refer usecase "apps\src\rtos\usecases\lvds_vip_sv_standalone\chains_lvdsVipSurroundViewStandalone.c" for Vpe example.

Regards, Sujith

Hi Sujith

    For the video1 pipeline, i use the follow params to realize a "full screen" display:

    inWidth/Height: 880*1080;

    tarWidth/Height: 1280*720,posX/Y: 0*0; (so the scalar of video1 pipeline will be triggered)

    displayWidth/Height: 1280*720

    The display result is as follows:

   But this kind of "full screen" is just like put a 880*1080 image to fill in a 1280*720 display container. The final surround view after synthesis will be scale again in order to realize this kind of "full screen". So I think this method may be not so good. The real "full screen" should be that we don't scale any size, the SV output width/height just match the display width/height very well.

  Does the VPE scalar's effect just like above? 

  If it is, i think the only way to realize "real" full screen is to modify the SV outWidth/Height...

Thanks

xuanbo

Hi Sujith

     Thank you for your suggestion.

     My usecase/need is to run the SRV on our custom display LCD used in vehicle "perfectly". Here, the meaning of "perfectly" is that SRV

can adapt its size to our custom display LCD, which means it should be full screen without the current black area on either hand of the SRV view. It seems that the only way to realize it is to change the SRV output resolution.

    According to your explanation, my understanding is that:

   1) For the VSDK version <= 3.2, the 2D SRV do not support to change the output resolution; but 3D SRV supports;

   2) For the VSDK version > 3.2 (released in the furture), both 2D/3D SRV will support the ability to change the output resolution.

   Does my understanding is right or not?

Thanks

xuanbo

Hi Xuanbo,

2D SRV can be achieved by 2 different usecases / Algo's. The 2D SRV algo that you are referring to dosen't support change of resolution.

If you want to change the resolution, I would recommend you to use "iss_mult_capture_isp_dewarp_3dsv_tda3xx"

In this usecase, VSDK versioned 3.2 has a defect, which prohibits change in resolution. This defect should be fixed in VSDK versioned 3.3.

Regards, Sujith