I assume DLSS 5 will be more useful for post-processing on 3D stills and artwork, than gaming. At least for a while, and of course while the card market is fucked.
I know this is the AI BAD circle jerk, but all the screenshots in that article look amazing.
The problem isn’t that they don’t look good, it’s that they don’t look right. Modders extracted the DLSS 5 library from this game and tried using it with a bunch of other games, and it completely changes how characters look (down to altering facial structure).
Combine that with DLSS 5 decimating performance even on the highest end machines, while also coming out when nobody can afford to upgrade their GPU, and it’s just a bizarre choice of technology to push right now. It feels like it’s being pushed to keep the AI hype train moving despite there being virtually no market for this tech.
Modders extracted the DLSS 5 library from this game and tried using it with a bunch of other games
But it looks good in stills in this game. It should be obvious why it looks bad in games which assets were never optimised for it and that the highest adjustment level will never be used in a game like the FFVII remake were the model tries to contort the anime faces into photographies.
The last I saw, the output of DLSS 5 could barely be tweaked at all by devs (only two variables exposed IIRC?). So it’s not so much optimizing assets as needing to use ones that already look realistic in the first place and hoping the algorithm doesn’t mangle them anyway.
The problem isn’t that they don’t look good, it’s that they don’t look right. Modders extracted the DLSS 5 library from this game and tried using it with a bunch of other games, and it completely changes how characters look (down to altering facial structure).
Says who? If I paid for the game and the tech, can’t I be the judge of what looks right? You mention modders, who themselves often completely change the look and feel of a game, and characters.
I assume DLSS 5 will be more useful for post-processing on 3D stills and artwork, than gaming. At least for a while, and of course while the card market is fucked.
I know this is the AI BAD circle jerk, but all the screenshots in that article look amazing.
The problem isn’t that they don’t look good, it’s that they don’t look right. Modders extracted the DLSS 5 library from this game and tried using it with a bunch of other games, and it completely changes how characters look (down to altering facial structure).
Combine that with DLSS 5 decimating performance even on the highest end machines, while also coming out when nobody can afford to upgrade their GPU, and it’s just a bizarre choice of technology to push right now. It feels like it’s being pushed to keep the AI hype train moving despite there being virtually no market for this tech.
But it looks good in stills in this game. It should be obvious why it looks bad in games which assets were never optimised for it and that the highest adjustment level will never be used in a game like the FFVII remake were the model tries to contort the anime faces into photographies.
The last I saw, the output of DLSS 5 could barely be tweaked at all by devs (only two variables exposed IIRC?). So it’s not so much optimizing assets as needing to use ones that already look realistic in the first place and hoping the algorithm doesn’t mangle them anyway.
Says who? If I paid for the game and the tech, can’t I be the judge of what looks right? You mention modders, who themselves often completely change the look and feel of a game, and characters.