Nvidia DLSS 3: Fake Frames or Big Gains?

For me even in dlss 2 only quality mode is acceptable, in the other modes I can easily spot the difference in quality.

Great info on the latency angle of dlss 3, getting more frames without considerably lowering the latency, really not sth you would call amazing...
 
However, we're uncertain if the main issue of latency can ever be solved given DLSS 3 fundamentally needs access to a future frame to slot in its generated frame. If Nvidia can somehow fix latency that would be amazing and elevate DLSS 3 to killer feature status, but we wouldn't hold our breath for that to happen.
This is, by far, the biggest hurdle to overcome and probably not one that can be - after all, two frames need to be buffered before all three (#1, #2, DLSS frame) appear. The game engine has only cycled twice for three frame presents.
 
In short, DLSS 3 will be of benefit to you provided you meet a few criteria:

1. Before enabling frame generation, you need to already be rendering the game at a decent frame rate, in the 100-120 FPS range.
2. You'll need to have a high refresh rate display, ideally 240 Hz or higher.
3. You'll need to be playing a game that is slower paced and not latency sensitive, like a single-player title.

That's maybe up to 0.1% of all owners.
 
Great analysys, thanks for it.
In summary this may be a classic "When you can't deliver the performance which you praised, at least pretend it by faking it"?
Better should have called it FFDLSS 0.1, aka FakeFrameDLSS 1 beta version.
I think that media and consumers have to be more careful nowadays and take a stand against this trend.
 
Last edited:
Just thinking logically about the technology, I had already come to the conclusion that most of negative aspects of what was said in this article would be the case.
1. You already need to have a better than playable framerate
2. It's really only for high resolution, high refresh monitors
3. It's really just a motion fidelity feature, not a performance feature
4. More frames and increased lag will feel weird, especially in lower rendered frames scenarios
5. You will not want to exceed the refresh rate of your monitor
6. Higher input lag will make it useless for the # 1 users of really high refresh monitors, e-Sports gamers

I just don't see this as a must-have technology. in some ways, you have to ask yourself who its even for. Obviously, Nvidia is going to push it to be implemented with every new DLSS title, the question becomes, if it's harder to implement than say DLSS 2.0, FSR 2.0 or XeSS could DLSS 3.0 actually come back to bite Nvidia now that there are other viable upscaling options?
 
That's maybe up to 0.1% of all owners.
Exactly, who the heck is this actually for? I have seen other articles saying around 80 is the sweet spot, but still, that's a more than playable framerate that will already look great in motion. Adding latency and other artifacts for a little bit smoother animation seems ridiculous. Maybe in something like a third person action game, but I always thought the primary reason for high frames was input lag. Most people are just fine with a solid 60fps in just about any title except competitive ones.
 
This is, by far, the biggest hurdle to overcome and probably not one that can be - after all, two frames need to be buffered before all three (#1, #2, DLSS frame) appear. The game engine has only cycled twice for three frame presents.
It's a visual fidelity feature. Whatever you think of Nvidia and RTX 40, marketing this as increase performance was intentionally deceptive, around my parts, they call that a lie.
 
This is, by far, the biggest hurdle to overcome and probably not one that can be - after all, two frames need to be buffered before all three (#1, #2, DLSS frame) appear. The game engine has only cycled twice for three frame presents.
It's first version and you're dismissing it as a whole already? Interesting....
 
It's first version and you're dismissing it as a whole already? Interesting....
No, I'm not - merely pointed out that the issue of added latency isn't likely to be addressed. That latency itself isn't necessarily a problem; in some games, it won't even be noticeable. The frame generation algorithm will certainly improve and the number of the visual issues should decrease in due course.
 
Very interesting info here.

This just reinforces my thoughts about how unnecessary this new version of DLSS is and that Nvidia is using it to help pad the numbers of this new generation of GPUs to help try to push sales.

Maybe in the future Nvidia can figure out how to fix things so it works better, but for now it just seems....well, rather pointless and misleading.
 
Maybe I am slow, but I don't see any value in those really high frame rates if the input latency is not there. Reviewers use FPS to benchmark graphics cards and it sounds like they found a way to exploit that without really delivering value.
The only value is motion fidelity. Looking like you are running the game at a higher refresh rate so that motion looks smoother. There is no other perceivable benefit and as far as I'm concerned, at least right now, the negatives outweigh the positives.
 
Back