Use of Nvvidconv to resize a video stream and display it with nveglglesink

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1

Hi,

I am processing a MPEG4 352 * 576 stream and am trying to resize it to 352 * 288 for display. Using the deepstream sdk 3.0, I have tried a few pipelines with nvvidconv but the results differ from expectations.

This first pipeline displays the full image in a 480*540 window with the original ratio of 352:576.

gst-launch-1.0  -v udpsrc address=239.27.36.104  port=50010 
! application/x-rtp, media=video, clock-rate=90000,encoding-name=MP4V-ES, payload=101 
! rtpjitterbuffer ! rtpmp4vdepay  
! queue ! nvdec_mpeg4  
! nvvidconv 
! 'video/x-raw(memory:NVMM), format=(string)NV12, width=(int)352, height=(int)576' 
! nveglglessink sync=false window-x=1 window-y=1 window-width=480 window-height=540 &

This pipeline displays a cropped image (only the upper half of the video is visible) in a 480*540 window.

gst-launch-1.0  -v udpsrc address=239.27.36.104  port=50010 
! application/x-rtp, media=video, clock-rate=90000,encoding-name=MP4V-ES, payload=101 
! rtpjitterbuffer 
! rtpmp4vdepay  
! queue ! nvdec_mpeg4  
!  nvvidconv 
! 'video/x-raw(memory:NVMM), format=(string)NV12, width=(int)352, height=(int)288' 
! nveglglessink sync=false window-x=1 window-y=1 window-width=480 window-height=540 &

This pipeline shows a green window

gst-launch-1.0  -v udpsrc address=239.27.36.104  port=50010 
! application/x-rtp, media=video, clock-rate=90000,encoding-name=MP4V-ES, payload=101 
! rtpjitterbuffer 
! rtpmp4vdepay  
! queue 
! nvdec_mpeg4  
! nvvidconv 
! 'video/x-raw(memory:NVMM), format=(string)NV12, width=(int)480, height=(int)540' 
! nveglglessink sync=false window-x=1 window-y=1 window-width=480 window-height=540 &

How should nvvidconv be used in order to obtain the full video displayed with a 352*288 size ?

Thanks in advance

2

Tesla nvvidconv does only conversion, no scaling. Can this answer your question?

3

Thank you for your answer. Yes it answers my question.