Hi,
In a gstreamer pipeline, I’m using nvvidconv to create NV12 block-linear buffers from 1280x720 UYVY input images. I then have a plugin that imports these NV12 buffers into VPI using vpiImageCreateWrapper with VPI_IMAGE_BUFFER_NVBUFFER.
The problem is that the resulting VPI image has BLOCK16_LINEAR format, and all VPI algorithms seem to exclusively accept BLOCK2_LINEAR, which is also what VPI_IMAGE_FORMAT_NV12_BL and others reference.
Can I somehow make nvvidconv produce BLOCK2_LINEAR? Can none of the engines actually deal with BLOCK16_LINEAR or is VPI just narrowing the compatible image formats too much?
Thanks,
Stefan