Sunday, August 30th 2026
Warhorse Developer Defends DLSS 5, Says It's Not an AI Slop Filter
Kingdom Come: Deliverance 2 director Daniel Vávra has praised DLSS 5 again, saying the tech now looks closer to how his team originally envisioned the game's lighting and character faces. Speaking on X, he said character models that once looked flat in-engine finally read the way artists intended, closing the gap between how faces looked on artists' workstations and what actually shipped in the game.
Vávra singled out shadow rendering as a big win. He said Warhorse struggled for years to get shadows from hats, hoods, helmets, and small props like buckles and pouches to render on characters, calling it a persistent frustration during development. DLSS 5 apparently solves this, and Vávra called it an excellent upgrade to the character lighting model, though he noted it roughly halves frame rate in its current form. He also sees potential in DLSS 5 when applied to older games:
Sources:
Daniel Vavra on X, Daniel Vavra on X
Vávra singled out shadow rendering as a big win. He said Warhorse struggled for years to get shadows from hats, hoods, helmets, and small props like buckles and pouches to render on characters, calling it a persistent frustration during development. DLSS 5 apparently solves this, and Vávra called it an excellent upgrade to the character lighting model, though he noted it roughly halves frame rate in its current form. He also sees potential in DLSS 5 when applied to older games:
"It's not the infamous AI slop we saw couple months ago. I see huge potential here for the future, and it's absolutely great that even older games lighting could be improved this way."This is a calmer take than most of the industry's reaction to DLSS 5. A TechPowerUp reader poll found 58% of gamers don't want AI altering their games at all, and an early DLSS 5 mod for Skyrim shows how dramatically the tech can reshape faces and lighting when pushed hard. Vávra's own examples look more restrained than those comparisons. He has defended DLSS 5 since its reveal, so this is a continuation of that stance now that he sees it applied to his own game. Warhorse is currently working on two new projects, including a Lord of the Rings RPG.



69 Comments on Warhorse Developer Defends DLSS 5, Says It's Not an AI Slop Filter
It looks like physx which was nvidia only also as far as i know in the past.
Anyone who wants to play a game and want those visual will not have any real argumetns not getting a nvidia graphic card as those game features are all locked to nvidia.
@DrUsmanSaleem You are doing a real good job in finding those topics. Thank you
It’s highly adjustable, so you can tweak it to your liking, at least with the modded one, and in some games it looks bloody impressive. It works well on the 4000 series as well, thanks to the community, but with the same massive performance hit. It’s like ray tracing: it was a massive performance hit on the 2000 series; now not so much. Give it a few more generations and this will have a negligible performance impact.
There is a big difference between a developer implementing DLSS5 and tune their game for it and just slapping it on via modding to an existing game.Well just look at GTA5, there are mods that make cars hyperreal, but they DO take a lot of performance. DLSS5 is doing essentially the same thing, it just adds faces and lighting to it.
Can't wait to play Saboteur again at 4K.
Here is Cyberpunk 2077 on a RTX 4080. Stunning:
more:
Joking aside, the extra detail could be something, but it has to be implemented by the developer, imo.
I can't recall seeing a graphics upgrade mod of any game that managed to preserve the artistic intention of the original. Extra graphical fidelity and realism don't always help, on the contrary sometimes.
Up to DLSS 4 everything was perfectly fine. Soon, if this whole thing doesn't stop, we're gonna run out of water cooling these AI slopcenters. Like this if you agree.
If the claim is that DLSS 5 remains useful despite the severe performance drop that skeptics highlight, the missing experiment is straightforward: show that reaching the same quality with conventional methods costs even more. Until that measurement exists, a halved frame rate is only a penalty, not a trade-off.
About 17 years ago, AMD’s Radeon group (now RTG) made exactly that case when it demonstrated hardware tessellation. Normal mapping could not match the quality. Parallax occlusion mapping could get close, but only at several times lower frame rates.
Normal maps, POM, and hardware tessellation each occupied a different quality–performance sweet spot. For users who were no longer satisfied with the first two, tessellation became a new Ultra option. That is why it survived the early pessimism about its cost.
Without equivalent data, praise for DLSS 5 is a developer’s personal preference. That objection is already easy to find, including in TechPowerUp comment threads.
Players are also living through an unprecedented HW price spike, and they are unusually unwilling to donate frame rate. Reaching them requires precise wording and complete evidence, not a partial observation dropped on the table as if it were the whole case.
I doubt anyone at Warhorse will read this. But if a developer or NVIDIA employee who wants to defend DLSS 5 does: start with the actual problem. Why do most players not share your view of the technology?
I dunno about that, I'd consider them among my primary considerations for a video game. I guess I don't speak for most of us though.
All the more power to you if graphics are your primary consideration. But to be a video game - a thing to be played - requires more than visual fidelity. It's in the name; "game."
That implies some form of "play" and preferably a "hook" that draws us in. The video part simply means a game rendered through a screen. Pac-man is a video game. So is Tetris.
To me, and this is from closely observing these threads, it seems some people are pre-occupied with hyping something that adds an extra layer of sprinkles. And as I've said before, sprinkles don't make a badly baked cake taste any better.
But I'm not here to argue with anybody's subjective opinion. Each to their own, until it treads on my toes.
Because when applied to real video clips, it changes the light and shadows and, of course, makes the real to look "more real” - right? :roll::banghead:
DLSS 5 is Slop, nevermind do you like it or not :)