cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
abdullah.tukl
Observer
Observer
294 Views
Registered: ‎11-29-2018

Two layered video output comparison between AXI VDMA and Video Mixer IP on Implementation and Functional level

Jump to solution

Hello Community Members,

We have a project where we are able to recieve an incoming xga or full HD video with 10-bit 2 pixels per clock YUV 4:2:2 properties via ethernet, decode it via H264 VCU and show it on SDI display with sdi-tx-subsystem and UHD-SDI GT in PL.

We have a gstreamer running on our petalinux 2019.1 build.

I need to implement the video in video concept in this project. My video should look like below

 enframing.png

The green part will be our xga incoming video and black full HD picture should be constant.

we could already transmit one video (either with xga or with full hd resolution) with our petalinux build on zu mpsoc with gstreamer but now the requirements have changed and we need to frame the incoming xga resolution and place it in a FULL HD black constant frame like above picture.

We have incoming camera resolution of xga with yuv4:2:2 10 bit per pixel and 2 pixels per clock and that we need to ouput in a full hd yuv 4:2:2 10-bit per pixel and 2 pixels per clock format with SDI-TX Subsytem on a zynq mpsoc custom board.

which solution will be more suitable interms of resource and design time efficiency according to your point of view? using an AXI VDMA or a Video Mixer IP from Xilinx? can we exploit the xilinx video framebuffer reader to do that as we already have that in our design?

your comments in this regard will be highly appreciated.

Best regards,
Abdullah

0 Kudos
1 Solution

Accepted Solutions
abdullah.tukl
Observer
Observer
111 Views
Registered: ‎11-29-2018

Hi @florentw

Yes we are using petalinux. We have solved the problem using video mixer.
We have configured one overlay layer with memory mapped interface of video mixer with required video format for sdi out.
The background or primary video stream in layer is left unconnected. So by default it appears blue in background, which we turn into black by changing the particular ycbcr registers of videomixer.
We actually removed the frambuffer reader and used the video mixer as the master out is also AXIS and the video mixer is also doing dma in this configuration.
The overall resource utilization difference is around +2% between framebuf read and video mixer which is perfectly suitable for our end project.

View solution in original post

0 Kudos
2 Replies
florentw
Moderator
Moderator
160 Views
Registered: ‎11-09-2015

Hi @abdullah.tukl 

As you are using linux, I would say that this might be doable with the Video frame buffer as the background is a constant. There might be a plugin in Gstreamer to control this. You might need to look around.

The Video Mixer might work as well but it my be a bit a waste of resources if this does not have to change over time.

Regards


Florent
Product Application Engineer - Xilinx Technical Support EMEA
**~ Don't forget to reply, give kudos, and accept as solution.~**
0 Kudos
abdullah.tukl
Observer
Observer
112 Views
Registered: ‎11-29-2018

Hi @florentw

Yes we are using petalinux. We have solved the problem using video mixer.
We have configured one overlay layer with memory mapped interface of video mixer with required video format for sdi out.
The background or primary video stream in layer is left unconnected. So by default it appears blue in background, which we turn into black by changing the particular ycbcr registers of videomixer.
We actually removed the frambuffer reader and used the video mixer as the master out is also AXIS and the video mixer is also doing dma in this configuration.
The overall resource utilization difference is around +2% between framebuf read and video mixer which is perfectly suitable for our end project.

View solution in original post

0 Kudos