Pretty much everything I said in my last post applied specifically to the Vero 4K and 4K+, in case that wasn’t clear.
The Vero’s CPU has only a tiny fraction of the general computing power that an Intel CPU has, so (for practical purposes) it can’t decode video at above SD resolution without using dedicated hardware to do it; and the dedicated hardware doesn’t support any upscaling algorithms fancier than bilinear.
And even when decoding SD video in software, fancier upscaling algorithms require more CPU power than is available, hence they don’t work correctly even in software mode. Bilinear is as good as it gets. (So, if possible, upscale externally).
I’ve actually not been especially impressed by the hardware deinterlacing on the Vero 4K+. In particular, it doesn’t correctly deinterlace 1080i film material - it deinterlaces it as if it were video, which causes resolution loss, especially when the camera is moving. Frustratingly, it deinterlaces 1080i test patterns correctly in software mode - but of course it lacks the power to decode real (i.e. non-test-pattern) 1080i video in software mode. 1080i video-mode material is handled well in hardware, though. (See Issues with deinterlacing 1080i/50 ).
Also, according to this thread, hardware deinterlacing of SD MPEG2 has issues as well: Why is hardware de-interlacing for SD MPEG2 so bad? So you might be better off decoding interlaced SD material in software. (Remember to select the best deinterlacing algorithm, as it may not be the default).