I've never personally owned an OLED tv. Is it that much better than a LED?
-
Ever wanted an RSS feed of all your favorite gaming news sites? Go check out our new Gaming Headlines feed! Read more about it here.
Next-gen PS5 and next Xbox speculation launch thread |OT5| - It's in RDNA
- Thread starter Mecha Meister
- Start date
- Status
That's what they say every gen...
EDIT: Here are some examples of what faces already look like this gen -
It won't let me post the images directly here for some reason.
Last edited:
Ya, I wouldn't be surprised if the first ND game next gen surpassed this!
Look into the tests that are available at rtings.com or on TV-centric channels on youtube and you will learn where OLED is better than LED. It depends what you wanna do with your TV and the environment the TV is in.
For example this review highlights pro and cons of an OLED quite good:
Last edited:
OLEDs are stunning. Perfect black levels and contrast, great viewing angles, and new ones have very low input lag. However they aren't nearly as bright as LEDs so they're not great for every setting
Even way better, nothing particular is happening here, no interaction or IA.
The animation on this video are not at the level of motion matching we saw on TLoU II.
Thanks! I was always cautious of OLED because don't they slow "die" over time?
If we're purely fantasizing here I'd say that CBR won't be widely used in next gen if at all. Resolution reconstruction will be handled by a mix of temporal accumulation (already used in many current gen games) and wide usage of variable rate shading in both texture and screen spaces. Current implementations of VRS are rather "lite" and naive since they are essentially tackled onto console renderers without going through the whole pipeline looking into what may in fact be shaded at a lower rate. A renderer which will be built with VRS being used throughout the whole pipeline will provide a lot more performance benefits than what we see right now in Wolf2 and CivVI, for example.
Sharpening always was and always will be a part of your regular TAA, whether its coupled with resolution upscaling or not. "RIS" doesn't bring anything new into console space at all since it's nothing more than a "hack" provided by PC GPU drivers to be used in games which have sub-par (i.e. not sharp enough) TAA/post-AA implementations.
And as for DLSS and other AI powered image enhancement techniques I doubt that next gen consoles will have enough h/w power to handle such AI approaches at speeds necessary even for 33.3ms frames.
Sharpening always was and always will be a part of your regular TAA, whether its coupled with resolution upscaling or not. "RIS" doesn't bring anything new into console space at all since it's nothing more than a "hack" provided by PC GPU drivers to be used in games which have sub-par (i.e. not sharp enough) TAA/post-AA implementations.
And as for DLSS and other AI powered image enhancement techniques I doubt that next gen consoles will have enough h/w power to handle such AI approaches at speeds necessary even for 33.3ms frames.
lol 5 posts in a row. you do know, you can add multiple quotes in one post right? you can also edit your post to insert more quotes.
dont want to be a forum nazi, but typically double posts are frowned upon. never even seen 5 posts in a row before.
Exactly. Unity's tech demos within the past few years haven't impressed me that much. God Of War 2 or HZD 2 will likely match or surpass it, unless they choose to favor physics, asset streaming, AI, etc. In that case I would prefer it to better visuals.
Sorry, I'm still getting used to this whole Resetera thing. I usually use Reddit.
One thing I don't see mentioned a lot that could be called an "optimization" over the PC components is that the consoles use a monolithic chip and won't need the "Infinity Fabric" or PCIe lanes tying all the parts together. That's one of the reasons the 6TF X1X punches above it's weight class.
I can't tell if you are agreeing with me or disagreeing cause both could apply.
But I'll elaborate. And feel free to correct m if I am wrong.
All are techniques designed to achieve the same end product. And that is taking a natively lower rez image and making it appear higher rez. All at a minimal performance hit to the GPU.
- RIS sharpens an upscaled lower rez native image.
- Nvidia uses "pixel training" to generate "additional pixels" to give an image the "impression" of being higher rez than its native resolution. Its targeted for games already struggling to hit higher frame rates (either cause the game just demands that much or cause you are using RT) because it has a fixed GPU load so gains are only seen when a game is struggling already.
- Sony's CBR renders out half of the targeted native rez per frame and uses pixel ID tagging and tracking to syn new and old pixels towards generating a completed image.
So yes, they all are trying to do the same thing albeit going about it differently.
Now why I say CBR is better, its because CBR is the only one of those methods that can be built on top the "targeted" native resolution. So unlike the other methods that start at a lower rez's framebuffer before doing their thing, CBR starts at the frame buffer of the intended final rez. Which means that things like on-screen furniture (text and HUD...etc) are also rendered at the native resolution.
over to you....
Last edited:
The two examples you mentioned are using hardware supported VRS - it's not just a different approach in software. AMD have a patent for VRS, but it's not a feature on the 5xxx cards. Here's hoping it turns for the consoles and the next generation of RDNA hardware.
I'm not sure what you mean by "always was and always will". It should be used in conjunction with AA (especially TAA) to preserve finger details, but many games have TAA and no sharpening.
Weird that RIS being referred to a "hack" as a bad thing tbh. Console games are made in spite of the tech, not because of it. Developers are trying to make the best looking games they can for the console budget, that means using any "hack" you can get your hands on. RIS is a perfect fit for console games, same as checkerboard, because it lets the developers push the game resolution as high as possible for the lowest performance cost, with a close enough final result.
EDIT:
I will add that for the same reason i predict most developers wont utilize ray tracing as much as some here hope.
Ray Tracing is too big of a cost for its payoff. Most developers will still opt for faked GI in games, SSR etc. We might see ray tracing in certain scenes but it will rarely become a main feature in the games
EDIT:
I will add that for the same reason i predict most developers wont utilize ray tracing as much as some here hope.
Ray Tracing is too big of a cost for its payoff. Most developers will still opt for faked GI in games, SSR etc. We might see ray tracing in certain scenes but it will rarely become a main feature in the games
Last edited:
I wouldn't be surprised if it made it into consoles. They had a similar quote about VRS to their raytracing responses. Something along the lines of they know it's a highly requested feature but they're only talking about current 5700 GPUs right now.
Not really. My guess would be early 2020's for consumer level pricing.
The draw distance, lighting effects, reflexion, smoke and fire. It was a slapp in the face for a console game. Just rewatched it and the big leap since this reveal is animation. Even pre rendered ones are really not great.
This is incorrect.
RIS is just a post process sharpening filter with an option of running said filter after a rather basic upscale routine which just gives you a better end result than upscaling an already sharpened image. The upscaling in it is rather straight forward from what I can tell and isn't remarkable in any way compared to other such upscalers, including those which are used by modern console h/w on display output (when you're outputting a 1440p game to a 4K TV for example).
CBR is a resolution reconstruction technique which can be used instead of a naive upscaler mentioned above but it requires quite a lot of renderer work and breaks a lot of things alongside it. You can just as well run a sharpening filter on top of CBR resolve if you feel that its too soft.
DLSS is a combination of image reconstruction with antialiasing as it tries to reconstruct a supersampled image from an aliased one. It's closer to CBR in a sense that it also tries to reconstruct a higher resolution image from a lower one but the approach is completely different, and it also provides AA while doing such reconstruction. You can run DLSS in final output resolution and in this case it will try to reconstruct supersampled AA only, without upscaling the image. And again, you can just as well run sharpening filter on top of any DLSSed image if you feel that it's necessary from IQ perspective.
So technically nothing stops you from, say, rendering in 4K CBR (1/2 4K native), running DLSS on that resolved image for some AI anti-aliasing and then applying RIS on top of resulting frame to make it a bit sharper. All three can be used simultaneously as they essentially handle different parts of final frame rendering.
Not 100% sure what you mean but all of these can be used before rendering HUD elements without affecting their rendering in display final resolution. CBR isn't unique in this.
VRS is h/w, it can't be a different approach in s/w. CBR is h/w too btw as it's using MSAA h/w. TAA upscaling is pure s/w though.
I'm about 99% sure that it will.
All implementations of TAA which I've seen are using some level of post-resolve sharpening as TAA generally produce a very blurry image without it. It's possible of course that there are games which use TAA without any sharpening passes but I personally haven't ran into such implementations being presented anywhere.
It's not a "bad thing", it's just not relevant for consoles where you don't have a user driver option of running RIS on top of any game you'd like. And most console games already use sharpening as a part of their TAA / postprocessing routines - maybe these give worse or better sharpening than RIS does but they are there already nevertheless. Considering that CAS is free to use I'd imaging that some games will use it instead of their own sharpening filters down the line, similarly to how FXAA essentially killed off most in-house PPAA solutions at the end of previous console generation.
I wonder if we'll get another Killzone 2/Uncharted 4 moment. Those two trailers truly showed the capability of the next gen machine. Unfortunately they were both a little bit too good to be true (UC4 with 60fps and Killzone 2 with... everything).
Then what do you mean by "naive implementations" and "VRS being used throughout the whole pipeline will provide a lot more performance benefits than what we see right now in Wolf2 and CivVI". What is a mature implementation, how does it different from Wolf2 and CivVI and how does that provide performance? I haven't used VRS so I don't know if shaders require extra data to determine how they're treated by VRS, but I can't imagine that a whole pipeline needs to change if it's already implemented.
Most? Sorry, that sounds like a guess to me rather than a confirmation after reading multiple papers on various renderers an their AA solutions. Although I could be being pessimistic because I hate that UE4 doesn't feature a post-TAA sharpening solution out of the box (or any attempts to remove ghosting) and the latest paper I read was on DS: Remaster's CB renderer and that sharpening is something they didn't get a chance to implement :(
Show me a techical demo in a TPS game where the player can do portals in differents higly complex worlds and jump in in less than 2 seconds.
Basically, don't show me what I am already playing but prettier / crispier / with higher frame rate, show me what was never done before.
Make me dream (pun intended, imagine instant loading in a game like Dreams)
Can I get a link? I assume you're talking about Dark Souls here but I could swear that ran at native 1800p.
That'll finally be possible with an SSD! Imagine a game like Skyrim that has Portal physics. You just walk through it and your in a different part of the map. That could literally change open world design.
Mecha Meister
Idea for a poll for the next OT:
Poll: What would be your red line for loading times between game levels on#nextgen #consoles (the time you start bitching about it)?
a) >= 1 seconds
b) >= 2 seconds
c) >= 5 seconds
d) >= 10 seconds
e) >= 30 seconds
What do you think?
Idea for a poll for the next OT:
Poll: What would be your red line for loading times between game levels on
a) >= 1 seconds
b) >= 2 seconds
c) >= 5 seconds
d) >= 10 seconds
e) >= 30 seconds
What do you think?
This is about our expectations and is less about/independent to what they may achieve!
I really am fine with this level of fidelity for linear games next gen:
I do think FP studios like ND and SSM would go beyound that in their late gen games!
I do think FP studios like ND and SSM would go beyound that in their late gen games!
I think it will be possible on ps5I really am fine with this level of fidelity for linear games next gen:
I do think FP studios like ND and SSM would go beyound that in their late gen games!
Now imagine if you could interact with every building!I really am fine with this level of fidelity for linear games next gen:
I do think FP studios like ND and SSM would go beyound that in their late gen games!
Hope your mom is well, Dictator?
Your phrasing kinda alluded to you being there to support her?
What about a bunny meat burger eaten on a spaceship?... perfection!
Afaik, inference based techniques are implemented using compute and with AMD hw generally stronger in compute than NVidia, and with RPM providing double the throughput at lower precision (which inference-based techniques can often cope with pretty well), i don't see DLSS or similar AI-based AA solutions being beyond next gen consoles at all.
Thanks for the info. Though I was aware of most of what you said with exception to a few (the part about other methods and HUD).
But what I am saying is "technically" not wrong. If you want me to be extremely "basic", all those methods allow you to get a final image that appears sharper and better than the native rez the image is rendered in.
RS is just sharpening after upscaling, which is something they must have felt is necessary since standard upscaling always makes the final image softer. CBR is designed to circumvent the pitfalls of basic upscaling and generally yields results better than just basic upscaling since its a reconstruction method. But it too ultimately allows you to get a final image that is "better" than ts native render. And DLSS also does something similar.
Their implementations are all different, but I don't see how you can dispute that they all contribute towards allowing you basically get "more for less".
How does this contradict anything stated here? MicroLED is 5+ years away from being viable for the consumer.
Well I'll be danged. There's another checkerboard game that fooled DF into thinking it was native, notably also 60fps. It's a shame it only speeds things up by 25% though.
Unfortunately they only reported frame time measurements of the PS4 Pro. It would have been interesting to see how big the improvements were on a Xbox One X because of the better GPU as an advantage and to do some of the things with manual shaders as a disadvantage.
More importantly, they are going to start producing 49" OLED models! That will also bring the price down a bit.
Referring to the bold text at the end of your comment: That is not just a risk, that is a given!
Maybe I am in the minority, but I prefer a loading screen over any climbing / heavy doors / pipe with ducking / ...
- Status