Why is cropping with videocrop and imxvideoconvert_g2d videocrop-meta-enable=true more efficient?

[vincentz63]

According to this post and the iMX Linux User Guide section 7.3.17 Video Conversion.i.MX Linux User's Guide (nxp.com) cropping is hardware accelerated when using a pipeline like this:

gst-launch-1.0 videotestsrc ! videocrop top=10 bottom=10 right=10 left=10 !
imxvideoconvert_g2d videocrop-meta-enable=true ! queue ! ximagesink display=:0

rather than just:

gst-launch-1.0 videotestsrc ! videocrop top=10 bottom=10 right=10 left=10 ! queue ! ximagesink display=:0

Why this is more efficient? Is the cropping done by the GPU when imxvideoconvert_g2d is used?

If so, what is telling videocrop not to do the cropping? I realise this is a GStreamer question, but I need to do something similar with another element, and it would be super helpful if someone could point me in the right direction.

[brian14]

Hi @vincentz63, 

Thank you for contacting NXP Support.

Yes, in this case the imxvideoconvert_g2d is a proprietary Gstreamer plugin with HW acceleration.
You can check more information about g2d API for i.MX in the following documentation:
i.MX Graphics User's Guide (nxp.com)

I hope this information will be helpful.

Have a great day!