-
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.
Ratchet and Clank: Rift Apart PS5 - The Digital Foundry Tech Review
- Thread starter zombiejames
- Start date
Haha, you beat me to it! On mobile they literally look identical, that's some crazy impressive camera work to line up a floating camera and a free camera to what appears to be the exact pixel.
Literally, lol.
s y what part of the game is that and how did you get the camera lined up pixel perfect (or is it cropped?) I might try and recreate when I'm at home.
Devs are getting smarter with photo mode now. Makes sense to just push super high fidelity everytime you're snapping pictures.
Here's hoping
Lol, I always see you pushing for better audio
Here's hoping
The assets used in cinematics are the same ones used in gameplay. Every asset is chains of levels-of-detail: models have LoDs, textures have mips, and even many materials get more and less complex as you get closer to or further from them. If you construct the LoDs optimally, the transitions happen at just the right camera distance to be invisible yet gain rendering efficiency. You wouldn't use the highest LoD when the main character is at your standard gameplay distance, because that would be a huge waste—you'd be spending frame time on tris and texels that you don't have screen pixels to display. So yes, if the camera is up close, you will see something "better" than if the character is far away, but that's because of the number of screen pixels you have to render them onto, not the fact that we're making some choice about how detailed to be in cinematics, gameplay, or photo mode.
If you need some other evidence for this, you can note that we rarely have hard cuts out of cinematics and into gameplay -- we try to blend between them and blur that line as much as we can. If we were swapping out assets, that would fall apart. (That's also why the frequent suggestion to run cinematics at 30fps and gameplay at 60fps would never work for our games.)
If anything, the primary quality differences in cinematics come from per-shot lighting rigs and complex facial animation, which don't apply to gameplay.
The logic behind forcing Fidelity settings in photo mode is mostly about encouraging people to enjoy the Performance modes. We anticipated a lot of "I don't want to play in 60fps because I want to have 4k photos with ray-traced reflections" (especially before the Performance RT mode came out), so it was nice to free people up from needing to make that choice. 60fps is not really worth it in photo mode, anyway. (There's some nuance here -- we can increase resolution and turn on ray-tracing when you enter photo mode, but we can't re-introduce VFX/lighting/scene density without reloading. So it's kinda somewhere between Fidelity and Performance.)
I think in the PS4 Spider-Man games we do a slight distance bias on the suit LoD in photo mode, but on PS5 it doesn't make a difference.
If you need some other evidence for this, you can note that we rarely have hard cuts out of cinematics and into gameplay -- we try to blend between them and blur that line as much as we can. If we were swapping out assets, that would fall apart. (That's also why the frequent suggestion to run cinematics at 30fps and gameplay at 60fps would never work for our games.)
If anything, the primary quality differences in cinematics come from per-shot lighting rigs and complex facial animation, which don't apply to gameplay.
The logic behind forcing Fidelity settings in photo mode is mostly about encouraging people to enjoy the Performance modes. We anticipated a lot of "I don't want to play in 60fps because I want to have 4k photos with ray-traced reflections" (especially before the Performance RT mode came out), so it was nice to free people up from needing to make that choice. 60fps is not really worth it in photo mode, anyway. (There's some nuance here -- we can increase resolution and turn on ray-tracing when you enter photo mode, but we can't re-introduce VFX/lighting/scene density without reloading. So it's kinda somewhere between Fidelity and Performance.)
I think in the PS4 Spider-Man games we do a slight distance bias on the suit LoD in photo mode, but on PS5 it doesn't make a difference.
once you get on a pc or something look at my initial post with the screens......one of the first main missions(La Nochebuena). Not that hard to position the camera super close in any game, just stand next some geometry and spin the camera:
and there you have it.
I mean why wouldn't you increase the base res or the precision of certain effects, it's not as if the performance is so important in there. That would be the least I would expect to kick in in photo mode.
Thanks for clarifying. Hopefully that'll put an end to some people's concerns. And thanks for such a beautiful game.
This just reminds how damn good the character models were in Spider-Man MM. I mean R&C is a BIG step up graphically over all, but would be happy enough with those main character models in the next Spider-Man...Overall though, I expect Spider-man 2 to look crazy.
Nice how you see confirmation in a post by a random user confirming your bias and disregard hard evidence (screens from in-game and photo-mode) by a different user.
I'm not sure you understand what the word "allegedly" means. All bit of a moot point now really but thanks for having a go at me without justification.
The only thing I notice here is that lamp could use a higher res texture. lolI can't think of any besides racing games. Last of us 2, tsushima, assassin's creed origins/oddysey were all the same in and outside of cutscenes.
The same LODs are available and work the same way in game as in photomode. they aren't switched to LOD0 when photomode is activated.
Just took some Miles screencaps.
And if you want to look at super up close detail:
Ratchet & Clank Rift Apart to have Performance and Performance Ray Tracing modes with the day one patch (short video of of performance RT mode in OP) News
This is a good way to think about it, even if you flip it around. You have twice as much CPU time to use in a 30 fps mode, so you've got to use it for something that isn't happening in the 60 fps mode, or it just goes to waste (and we don't waste CPU time in this house!). That something can't...www.resetera.com
Insomniac devs are wizards! :)
These guys make it look so easy to come up with high quality games so quickly. I wonder if the PS5 is just that good a machine to develop games for, and the praise it received pre launch is now being justified. I know they have the talent, it's plain to see, but obviously a kit that makes their job easier will produce what they envisage sooner.
I believe they've lowered density of objects and particles.
Do we have a specific number on what the drop in resolution is? Because if it never goes below 1440p, I will be a happy camper as that's the monitor I play it on anyway and the other effects I can live with.
Edit: wait, is this about Ratchet and Clank or Miles Morales? Well, either way, can I get a number on both?
Thank you for killing off some of the misguidance seen in here!
Just want to say that I appreciate all the discussion from everyone, and mostly it's just exciting that people care! I wasn't trying to dub anyone right or wrong (or misguided); I just thought it was a good opportunity to fill in some gaps, even though I can't keep up with _all_ the back-and-forth right now. Hope everyone gets a chance to enjoy the game this weekend (and to take some awesome photo mode shots)!
I'll be spamming the console screenshot thread from Friday onward:)
But correcting a misguidance isn't particularly a bad thing. :)
In this case, it was informative and corrective.
100%
I'd go so far as to say this feels like the biggest leap since PS1->PS2.
I assume you're talking about me? It's ok, you can tag me, I won't bite! 😁
This is a technical discussion thread, if we can't ask questions like "that's Photo mode, other games add enhancements/change the LoD in that mode, can we really say it's a gameplay picture?" without it being labelled as being misguided then that's a poor place to be in my opinion. These threads are where we should be able to ask/debate those questions and more.
I'm incredibly grateful for fitzymj taking the time to come along and explain the details, most of the time these things go round and round without us ever reaching a satisfactory conclusion. We now know more than we did, and that should be celebrated! 😁
It's not the models that change in photomode these days but rather draw distance of objects/shadows, resolution of post processing effects/shadows etc.
Cutscenes use the same models but use better/softer lighting that are carefully placed, and that's what make the models look better not the actual model quality. It's been this way for years now. And ofcourse the superior post processing/shadows and draw distance etc that photo mode benefits from.
Not even Uncharted 2/3 on PS3 was swapping models in cutscenes. But rather using better shaders and lighting in their cutscenes which they ended up pre rendering anyway.
Cutscenes use the same models but use better/softer lighting that are carefully placed, and that's what make the models look better not the actual model quality. It's been this way for years now. And ofcourse the superior post processing/shadows and draw distance etc that photo mode benefits from.
Not even Uncharted 2/3 on PS3 was swapping models in cutscenes. But rather using better shaders and lighting in their cutscenes which they ended up pre rendering anyway.
Fidelity mode is photomode in this. It seems like when you are up close to objects in fidelity mode you get the same quality. The impressive thing here is the quality of assets across the board. Even the thousands of npcs in the crowd in the opening level look good.
Sounds like RT perf is the way to go, about to try the game now and see if it's obvious or not.
wow the performance RT mode still looks awesome and dat 60. Definitely the way to go
Great video! The drop in resolution is quite noticeable and I don't think I'll stomach going all the way down to the RT mode. I'll probably play in standard Performance mode.
I don't think it's been picked up in the video but there's a clear downgrade in the DOF quality in some of the scenes too (which seems like Fidelity > Perf > Perf RT), and it's unfortunately a signature visual component of the game. It can be seen in this Draw Distance segment (timestamped).
I really hope this 60fps trend will accompany us throughout the whole generation. People then can finally choose how to play their games. What makes me a little bit more optimistic is, that the first current-gen only game allows us to choose. I will always go max fidelity all the way. But a lot of people just want their smoother frames and we might finally get it this gen.
Yeah I have no clue. I would say performance mode but it almost looks jarring to not have RT with how this game is designed. Performance RT mode looks great but 1080p at the lowest is quite low and and obviously noticable but most likely rare. I'll also assume it's the mode with the lowest fps drops also which are probably also rare. Feels great to have these choices either way. I'll have to wait for DF video and be able to slap it on my c9 to see what's what.
Yeah, again. Makes me wonder if for their next game they'll just optimise around this mode and maybe not bother with the other 2 settings
I'd love to see that. I'll be on RT Performance mode myself but I recognise that having 3 modes to choose from can lead to a minor case of choice paralysis for some folk.
With RT Perf looking so good it doesn't feel like we really need a discrete performance mode (though I appreciate some would prefer that mode to RT Perf)
It's for sure one that you'll have to toy around with.
Perf RT looks unbelievably good. Temporal injection is certainly doing its job here. Can't say that I've noticed many differences during my playthrough so far
Fidelity mode does look very, very sharp though.
Also, 3D Audio doesn't sound muffled on this game! It's very clean
Perf RT looks unbelievably good. Temporal injection is certainly doing its job here. Can't say that I've noticed many differences during my playthrough so far
Fidelity mode does look very, very sharp though.
Also, 3D Audio doesn't sound muffled on this game! It's very clean
Just start with performance RT and dont even look at the other modes. You wont miss them :D
Changing back and fourth is the worst
You were educated on the subject and all is fine. I say you because my profession already afforded me the ability to know what you were told. I was simply happy to see it explained.
I apologise, I didn't realise you work at Insomniac and knew the specific details of the games photo mode, you should have said earlier 👍🏻
Perhaps you'd care to share some more information, I'd love to know more about the RT on the eyeballs: Surely that can't be running all the time? If not, when does it kick in? Is it when the eyeball is in camera view? Only at a certain LoD etc?
Last edited:
I think what that poster was saying is that when you work in this field (not necessarily at that studio) you understand how something like level of detail scaling works. If the only time the fidelity in photo mode increases is when the camera gets closer to the assets (just like in gameplay), then there's no reason to assume photo mode is handling LODs any differently than gameplay; any differences in the ability to manipulate the camera in photo mode compared to gameplay will be responsible for any perceived difference in quality. This would be more obvious to someone who understands how the technology works compared to someone who doesn't. And that's ok, clearing up misunderstandings based on experience is a good thing.
There's nothing special about this from a technical standpoint; it needn't be treated any differently than any other reflective surface, unless the devs simply want to squeeze out extra performance by only counting intersections against the cornea at a certain distance from the camera. Only the devs (or an analysis of the code) would be to tell us that.
Bottom line: applying reflections to materials is not as expensive as knowing what objects/materials that other materials should reflect.
I guess I was mainly interested from a performance perspective. Most of the time during gameplay the player won't see the eyes, so was interested if this was something that was always happening (and it had any significant cost associated with it) or as you suggested, has a cut-off of camera distance and/or being visible to the player camera.
We've seen so few games use RT, and I appreciate it's early days yet, so to see it on eyeballs as well as general use is somewhat unusual (on console at least) and very interesting! 😁
Yeah the surfaces showing the reflections isn't the hard part, even if those surfaces are tiny. It's getting the detail for reflections that's the hard part. Even cubemapped reflections can be applied to corneas or even water droplets.
When I mentioned corneas not being including in the BVH, I misspoke, btw. I meant the materials that the cornea would otherwise be reflecting.