• State TI Thinks Resolved
  • Locked Locked
  • Replies 3 replies
  • Answers 1 answer
  • Subscribers 98 subscribers
  • Views 381 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: How to using DSS Overlay Manager composites two display frame buffer

TDA4VM: Enable the composition of 2 frames from different video streams at DSS Overlay Manager

Carlos Delgado
Prodigy 60 points
Part Number: TDA4VM

Hi,

We are wondering how can we make a composition of 2 images to display one side by side using the different layers avaliable at the DSS Overlay Manager. Something similiar as the 'mosaic' tivx node does but at DSS.

Our approach is to use the same pipeliine where the frames of the 2 different video streams are sent alternatively. Between the sent of every frame, we want to change the configuration of the layer at the Overlay Manager connected to the video pipeline. This way we want to achieve the merge of the 2 different frames at different layers. To achieve this, we want to extend the API for the TIOVX Display Node that now only supports chages in the display position and channel, related to the display driver. We want also to extend it tu support the display controller driver, responsible of the configuration of the layer parameters for the Overlay Manager.

While we do our test (in progress), we would like to know if such composition is feasible, as I see the configuration of the overlay manager is done at start and not once the video chain is running. Is there anyhting special we should consider for our implementation?

Regards,

Carlos Delgado.

  • 0 in reply to Nikhil Dasan
    Prodigy 60 points

    Hi,

     I have gone thought it and it's not exactly the samme case, as we want to use the same pipeline and switch the layer-z order to do the blending in alternate frames at runtime, expecting the previous layer to be maintaned to allow the merge, instead of flushing it with the new input. Also we consider the normal case where output is video port, not M2M configuration. We have to deal with the limitation as we will only have one video pipeline avaliable in the final SoC we will finally use. 
    I have seen that at runtime only some crop values (x, y position) are avaliable. as depicted in file 'vx_display_target.c'. I still have to keep digging to explore the possiblity, as this configuration options are not avaliable in the current software. I would like to know in advance if there is any limitation concerning the swith of configuration of z channel in runtime.

    Regards,

    Carlos Delgado.

  • 0 in reply to Carlos Delgado
    TI__Guru* 85196 points

    Hi,

    Carlos Delgado said:
    as we want to use the same pipeline and switch the layer-z order to do the blending in alternate frames at runtime, expecting the previous layer to be maintaned to allow the merge, instead of flushing it with the new input

    Do you mean that you would want to do blending with just one pipeline?

    If yes, The IP does not store the image internally and also the z-order is with respect to the other pipes and not within 1 pipe.

    Carlos Delgado said:
    We have to deal with the limitation as we will only have one video pipeline avaliable in the final SoC we will finally use. 

    Is it because the other pipes are used for something else?

    The z-order could be changed, but this means that the other pipe would come on top, rather that the data within the same pipe. 

    Hence, this is not supported by the IP.

    Regards,

    Nikhil