DM8148 HDVPSS: video compositor and video blender modules

After composing video and graphics do you just want to display the image or do you want to encode it ?

If you only want to display then composition of video and graphics is already happening when you use fbdev to display grpx plane over video.

If you want to encode it, DVRRDK supports a features called captureWriteback which enabled writing the video+graphics composited image back to memory which can then be encoded.

I am not sure if the release you are having from UDworks supports this. You can migrate to latest DVRRDK release 4.1 .

CaptureWriteBack is supported only on onchip HDMI display.

After composing video and graphics do you just want to display the image or do you want to encode it ?

Both, one project for make HW overlay only for dispay, second project - for encode with overlay, and maybe third project - both options together: HW overlay for display used for system menu, user and system info, etc. and overlay on video for encoding used for mark targets, mark motions, etc..

If you want to encode it, DVRRDK supports a features called captureWriteback

my version of dvr-rdk 3.5 havn't this example.

. You can migrate to latest DVRRDK release 4.1

Let me know, please, where I can download DVRRDK-4.1 This issue is very important for our project.

Where I can recieve detail description of HDVPSS module - registers, memory, etc...?

You can get latest DVRRDK 4.1 from TI FAE/Sales contact.HDVPSS internal documentation is also available via TI FAE/Sales Contact. It requires NDA to be signed

is already happening when you use fbdev to display grpx plane over video

dvr-rdk version 3.5 have example or documentation for FBDEV?

and what about use video plane, not only graphics plane, for overlay?

there are sample application which uses fbdev

You could use video plane either from linux using v4l2 driver or from m3 driver. You could get example for this in dvr-rdk3.5 as well.

Blender uses two layers and blends them using alpha value, it needs two layers.

Regards,

Brijesh

Hi Brijesh,

please let me know - what you mean in "using v4l2 driver or from m3 driver"? Are these 2 different ways for perform video overlay? 

Hi Badri,

thanX alot for your answer. But unfortunatelly, ud-works not relase yet mcfw with latest dvr-rdk. Latest version available on ud-works site is dvr-rdk 3.5

Hi,

Where can I find captureWriteback sample ? Is is the same as SWOSD in dsp ?

Thanks,

Ran

No, it is not same SWOSD.

You could find an example of using write back in one of the usecase of TI8107.

Regards,

Briujesh

The following usecases show captureWb usage. it is not same as SWOSD. CaptureWb will write the image displayed back to memory.

/dvr_rdk/mcfw/src_linux/mcfw_api/usecases/ti816x/multich_hd_sdi_dvr.c

/dvr_rdk/mcfw/src_linux/mcfw_api/usecases/ti810x/multich_progressive_8ch_vcap_venc_vdec_vdis.c

So just to summarize, if we need to encode image+video, we can use one of the following capabilities in RDK:

1. SWOSD - transfer the video to DSP, and DSP is adding images and return to A8.
2. CaptureWb - writing the video back to memory

As to option 2 , I think it is less used in the samples for video+graphic, and using 1, is more used, Is that right ?

Thanks,

Ran

Hi,

> "CaptureWriteBack is supported only on onchip HDMI display."

When using DVO2 tied to HDMI, does it mean the CaptureWriteBack is supported ?

Thank you,

Ran

Hi Ran,

The path is only connected to HDMI, so it captures images displayed on HDMI. If you tie HDMI and DVO2 and keep the contents same, ie Video and GRPX both contents, then DVO2 output is captured via HDMI writeback.

Regards,

Brijesh