Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Ben10]

I have used this RTX video super resolution in the Edge browser with YouTube content and it has amazing AI upscaling technology. VLC has a beta with this technology, I tried it, I can't believe how terrible VLC made my media collection look.  After trying MC30 (from using MC27), I can really appreciate now how much has been done in development.  My minor gripes have been dealt with.

I wonder if AI upscaling is something that may get integrated into future releases of JRiver MC31 in JRVR. Intel has an AI upscaling technology and so does Nvidia. Intel's XeSS can run in software and hardware while Nvidia's tensor cores are hardware yet there do have a software version also.  AMD has their own technology as well, which is entirely software based.  I would welcome these upscaling options incorporated into JRiver.

[Hendrik]

FSRCNNX, available in JRVR, is a machine learning (eg. AI) upscaling algorithm available today.

[eve]

Weirdly enough, compared to what is / has been available or possible with JRiver, RTX Super Resolution is not particularly innovative.
MadVR, and JRVR absolutely do a much better, more faithful job when configured correctly IMO.
I also find that Super Resolution is tuned (and probably trained) with compression in mind. This is a positive for things like YouTube videos which exhibit a wide array of compression 'difficulties' due to the source material, and how they're encoded. This is *less* positive when you're feeding Super Resolution a competently encoded source (like a retail blu-ray remux).

[Ben10]

I am currently using DSR DL (DSR on tensor cores) alongside MadVR.  It renders in 6K then scales back down to 4K for the LG OLED.  I just set everything to desktop settings.  Mainly an MKV library of 1080p content.