60 Frames on Past Consoles.

I haven't played CoD: AW. Mostly extrapolating from my experience with the few CoDs I did play, and they were all tightly scripted corridor crawls where the only way was straight ahead, and with basically nothing deserving of the term A.I. either. If the game has something akin to KZ's forest level and enemies don't just pop out of cover and shoot randomly this time, then just forget what I said.
 
Well, I think that forest level was the only one in Shadow Fall that really allowed for slightly more open ended play. Many of them were very much corridor affairs, take item x to point y, with not much in the way of enemies inbetween. All very linear.

I'm not suggesting for a second that COD is anything but linear, although it does have some missions in very large environments where you can pick your route. I haven't finished it yet, I'm thinking of a very specific level where you're on the surface of a huge warship type of boat - the thing it is really massive - and you're just bounding around this ship with dozens and dozens of enemies and teammates battling it out around you. None of my experiences playing SF had anything like that amount of activity or space to play in. I felt like a small part of it, rather than the sole responsible person having any effect.

Also, this mission gave me a greater sense of immersion and "appeared" much prettier to my untrained eye. So COD had way more action, a similarly large space, many more NPCs, and running at twice the frame rate! All while looking more pleasing.
 
Well the big question is, is ND able to pull of U4 at 1080p/60fps while maintaining the quality as promised in the trailer?
If yes, it can show that it is very possible to have both 60fps and very detailed realistic visuals. It may create motivation for other devs to reach that quality that was previously thought impossible. I wonder if the CU's or some new rendering techniques can make that target feasible.
 
You seem to be misunderstand the situation. The market wants realistic visuals in a lot of games. Therefore, the have a product that'll sell to their audience, some developers don't have the option for cutesy games. Can you make a cartoony shooter? Yes. Can Infinity Ward make a cartoony COD? No, because then it won't be COD. Can you make a cartoony racer? Yes. Can Polyphony Digital make a cartoony GT? No, because then it won't be GT.

Nintendo (at the moment) prioritises games to be family friendly and visually light-hearted, so whatever game they make, they have the option of going cutesy and light-weight on the visuals, but they play to a niche. The whole market can't follow suit and serve that same niche. If every game in 2015 was Nintendo-style cartoony, gamers would be up in arms and developers would struggle to stand out and make sales.
Actually, Borderlands is pretty demanding and performance is lousy (:love:0fps on PS3 with poor IQ). Cel-shaded isn't really what we're talking about here, but simpler visuals it's easier to hit 60 fps with. Borderlands running at 60 fps on PS3 with clean lines would need a far simpler look - simpler geometry and simpler shaders.
Yes, on paper, but in reality it's not true when you're trying to make a competitive product that sells and your audience likes realistic visuals.

No, I understand the situation just fine, its hard to not understand the situation when the biggest selling games year after year are shooting for realistic looking visuals. I understand that the larger market wants that, and that's ok. The point I am making is that developers do ultimately have the same choices (even if financially the choice becomes obvious), we aren't talking about sales here, we are simply talking about the context of making a good looking game that runs at 60fps, and ultimately every developer has the same options on the table that Nintendo does. Im not saying that all games should go for cartoon, not at all, but Nintendo is in no way privileged as the only developer who can go that route with good results that are critically praised. De Blob comes to mind as an excellent looking game that is very Nintendo in their approach to the visuals.

I think the idea that Nintendo's games are so painfully simple that 60fps is a walk in the park is disrespecting the skill and talent of Nintendo's teams. I would argue that Mario Kart 8 looks just as impressive as Little Big Planet Karting on PS3, and does it at a rock solid 60fps (or 59fps LOL). Smash Bros is going to have 8 characters on screen, 1080p and 60fps. So even within Nintendo's "simpler" cartoon visuals, they are still pushing their hardware pretty hard, and are technically impressive in their own right.
 
Nobody's saying 60Hz is easy. People are saying it's EASIER to do it they way Nintendo does it, rather than the way Sledgehammer did it (if we ignore the dropped frames in CODAW for the moment).
 
I think the idea that Nintendo's games are so painfully simple that 60fps is a walk in the park is disrespecting the skill and talent of Nintendo's teams.
I agree, and I haven't done that. When it comes to picking and implementing an art style to achieve a good 60fps on a platform, Nintendo probably has the best devs going. The rim lighting in Mario Galaxies was artistic genius.
I would argue that Mario Kart 8 looks just as impressive as Little Big Planet Karting on PS3, and does it at a rock solid 60fps (or 59fps LOL).
It looks better, but MK8 is without the completely user designable levels. There's no option for things like precomputed lightmaps in LBP karting that MK8 could use. In order to get the realistic material look in LBPK and the user generated content, getting all that at 60 fps was a tall order. Nintendo would rather prioritise 60 fps and pick the game and visuals to match.
So even within Nintendo's "simpler" cartoon visuals, they are still pushing their hardware pretty hard, and are technically impressive in their own right.
This isn't about who's pushing the hardware more. It's about reaching a 60 fps target, so 16.666 ms per frame instead of 33.333. You have to make visual concessions to reach the higher framerate thanks to the laws of physics. Those concessions would be more noticeable in a game trying to reach photorealism than a game that's going for a non-photorealistic renderer. The choice of 'cartoon' or 'simple' makes it easier or more aesthetically comfortable to hit 60fps and still look good. eg. Mario Galaxies' N64-quality dot shadows still work in that game because of its cartoony style. It was an aesthetic choice that enabled a super-simple shadowing technique to still work within the visual structure of the game and not look completely out of place (significantly because the antagonists were broadly spherical). And by using super simple shadowing techniques, it was easier to hit 60fps. The end result is a great looking game running at 60 fps and not looking like it was taking 'shortcuts'. It was a smart move and any similar cutesy platformer from any other dev could have made the same choices (does anyone know what framerate Ilomilo is on XB360 for comparison?). It was likely also using Wii's hardware to its maximal capacity. COD can't stick dot shadows under all the player characters though. ;)
 
There's nothing hard about reaching 60fps with realistic graphics. You just have to target a lower graphical level. That's it.
 
There's nothing hard about reaching 60fps with realistic graphics. You just have to target a lower graphical level. That's it.

That's true, but the discussion was really about making a aesthetically pleasing game at 60fps, and that cartoon graphics make it easier to obtain that 60fps, and still end up with a good looking game. Honestly, they have actually done an excellent job over the years making COD look better and better on the 360/PS3, and actually improving the framerate stability in the process. So you can do it, but like Shifty pointed out, there would be certain things that would make a game look severely dated with realistic graphics, where they are more likely to go unnoticed with cartoon visuals.
 
Art always wins, that's true. RE5 still looks great today thanks to the art design.

EDIT: take a look at this:

(19:58 - 21:41)

Old OoT is certainly far more grounded in realism than the 3DS remake yet its atmosphere still looks great today while the 3DS version, despite the higher-res textures, more polygons and the fact that it is far more cartoony, looks like vomit.

More advanced rendering features will never supersede art.
 
Last edited:
That's true, but the discussion was really about making a aesthetically pleasing game at 60fps.
I think we managed to make an aesthetically pleasing Xbox 360 game at locked 60 fps. Here's some footage from Trials Fusion (Rider's of Rustlands DLC):

I have to say that this DLC looks much better than the original game. It took our artists some time to get used to our new physically based material model. Obviously 30 fps would have made the game look much better, but this kind of reaction based physics engine driven game really needs 60 fps to play well.
 
I think we managed to make an aesthetically pleasing Xbox 360 game at locked 60 fps. Here's some footage from Trials Fusion (Rider's of Rustlands DLC):

I have to say that this DLC looks much better than the original game. It took our artists some time to get used to our new physically based material model. Obviously 30 fps would have made the game look much better, but this kind of reaction based physics engine driven game really needs 60 fps to play well.

Looks nice to me. Was it a big challenge for you guys to obtain the solid 60fps? If there were challenges, can you elaborate on them, and explain how you were able to get the desired performance you were aiming for. For example, were you finding it more diffiuclt on the CPU side of things because of the heavy physics engine, or was it still primary a fillrate issue with holding that 60fps mark? I would love to have a better understanding of the challenges developers face, and how they find solutions to these problems.
 
Was it a big challenge for you guys to obtain the solid 60fps? If there were challenges, can you elaborate on them, and explain how you were able to get the desired performance you were aiming for. For example, were you finding it more diffiuclt on the CPU side of things because of the heavy physics engine, or was it still primary a fillrate issue with holding that 60fps mark? I would love to have a better understanding of the challenges developers face, and how they find solutions to these problems.
Locked 60 fps (16.6 millisecond frame budget) is hard. You need to have a different mindset compared to a 30 fps unlocked game development. 30 fps games often tend to dip to twenties when stressed, while a 60 fps locked game NEEDS an average fps of around 80 (you need to ensure a minimum fps of 60, and there's always fluctuation). Basically you need to start your graphics feature set design by time boxing your frame. You need to think very hard what you can afford and how you can combine everything you can afford in a way that minimizes the data movement and repeat work. Usually this means that you have to combine as many passes together as possible and think very hard how to combine multiple algorithms in an efficient way (benefiting all of them). Many times this means that you need to compromise. For example our Xbox 360 blur kernels handle bloom, DOF and fake god rays at the same time (none of them are particularly perfect, but at least we can have them all at 60 fps).

Virtual texturing has been a godsend. It allows us to save HUGE amount of work in decal rendering (our terrain is mostly textured with decals, it has over 100k decals in total). Virtual texture basically acts as a surface cache. We only need to render decals to surface pixels when the pixel first appears in the screen. The decals will stay "burned" on the textures as long as they go out of the virtual texture cache. This makes decals pretty much free compared to traditional rendering pipelines. We couldn't have afforded any decals (= a nice looking terrain) without virtual texturing.

Virtual texturing also means that you can theoretically have "unlimited" texture detail, since the increased texture quality doesn't consume any memory (memory is always very tight on last gen consoles). However in our case we were actually limited by the downloadable package size (since digital distribution is very important for us). 2 GB was the limit, and that obviously limited how much texture resolution we could use.
 
Locked 60 fps (16.6 millisecond frame budget) is hard. You need to have a different mindset compared to a 30 fps unlocked game development. 30 fps games often tend to dip to twenties when stressed, while a 60 fps locked game NEEDS an average fps of around 80 (you need to ensure a minimum fps of 60, and there's always fluctuation). Basically you need to start your graphics feature set design by time boxing your frame. You need to think very hard what you can afford and how you can combine everything you can afford in a way that minimizes the data movement and repeat work. Usually this means that you have to combine as many passes together as possible and think very hard how to combine multiple algorithms in an efficient way (benefiting all of them). Many times this means that you need to compromise. For example our Xbox 360 blur kernels handle bloom, DOF and fake god rays at the same time (none of them are particularly perfect, but at least we can have them all at 60 fps).

Virtual texturing has been a godsend. It allows us to save HUGE amount of work in decal rendering (our terrain is mostly textured with decals, it has over 100k decals in total). Virtual texture basically acts as a surface cache. We only need to render decals to surface pixels when the pixel first appears in the screen. The decals will stay "burned" on the textures as long as they go out of the virtual texture cache. This makes decals pretty much free compared to traditional rendering pipelines. We couldn't have afforded any decals (= a nice looking terrain) without virtual texturing.

Virtual texturing also means that you can theoretically have "unlimited" texture detail, since the increased texture quality doesn't consume any memory (memory is always very tight on last gen consoles). However in our case we were actually limited by the downloadable package size (since digital distribution is very important for us). 2 GB was the limit, and that obviously limited how much texture resolution we could use.

Why aren't more developers using decals? Is it something that doesn't work for 3D games?
 
Decals are expensive unless you can bake them onto something like a virtual texture (bake once, eat for free in subsequent frames).
 
Why aren't more developers using decals? Is it something that doesn't work for 3D games?
Many do use decals, it's just that with a traditional texturing model (where surfaces share textures in memory instead of each surface getting its own texture memory), you have to blend the decals onto the surface every frame. The bigger the decal is in terms of on-screen pixels, the bigger this cost gets, so in snazzy situations with a free camera and lots of unpredictable creation of large decals, performance can be all over the map. Hence developers often restrict themselves to small decals that vanish quickly.

Games that use large decals liberally tend to have unstable performance as a result. Halo 1 allows you to obliterate groups of enemies with grenade chain reactions, leaving large black scorch marks all over the place and painting the walls purple in blood. But walking past the mess and looking at the decals can cause spikes even when you're in a restrictive simple environment and "nothing is happening."
 
Guys I have a question: do higher frame-rates improve IQ? in my own experience I'd say that it positively affects IQ, Bioshock on 360 with it's frame-rate option is a good and quick way to test this...going from locked frame-rate to unlocked shows a sharper and cleaner image IMO. I also think that it's harder to accurately represent the IQ of a 60fps game via screengrabs versus a 30fps title (that runs at the same res and with comparable AA methods of course).

So...am I imagining things or it's something that it's true? also is there a technical explanation for this? I didn't know where to ask about this but I think that in this thread it may not be too off-topic. :oops:

Thanks in advance.
 
Well, you're seeing more data/second. I can't see how else it makes IQ appear better though.
 
[nostalgia] The cost of 60fps reminds me of the Namco System 21 (Polygonizer) back in the late 80s. At the time nothing could touch its computing power in the arcades until the Sega Model 1 in 92. It was a massive beast, no less than 4 PCBs (1 CPU and the other 3 responsible for the graphics) 2x 68000s, 2x 68020, tons of sound chips, 5x TI TMS320C20 DSPs + custom stuff, 3.2 mb ram all to play a racing game called Winning Run at 496x480@60fps. The Coin-Op companies were doing cutting edge stuff back then in a never ending quest for peoples attention. [/nostalgia]


I guess my point is that delivering cutting edge real time 3d graphics at 60 fps was difficult and expensive back then and it still is. I think if this generation of VR prototypes catches on (I am personally not convinced), the consumer demand for better visual fidelity at high frame rates will help renew demand for powerful computing hardware.
 
I have been playing a ton of Bayonetta and Bayonetta 2 lately, and one of the things that I didn't pay much attention to until recently, is that Bayonetta uses mostly pre-baked lighting and shadows. Its not something that most gamers are really going to think about or notice, but when really paying attention, you will notice how frugal Platinum Games was with their resources, and was more than willing to use static shadows and lighting throughout, and this likely helped them hold tight to the 60fps target. There is a flatter look to the games because of these static lights and shadows, but honestly, static shadows are often less obtrusive than dynamic shadows that are jaggy as hell. Dynamic shadows are preferable, but if they are low res and come at a high cost of performance, then the developer is likely wise to choose static when they want to hold that 60fps. Platinum Games does not skimp when it comes to post processing effects. They compromised on the shadows and lighting, but post processing effects are used extensively in battles. Bayonetta 2 saw Platinum Games being a little less frugal. I see more dynamic lights and shadows, and far larger set pieces. They certainly did strive for a more impressive visual presentation than they did with the first game, but it comes at a cost. Bayonetta 2 generally keeps a higher framerate, but their is a slight judder effect that is somewhat annoying when the framerate gets a bit lower. Vsync is great to eliminate tearing, but it does come with some cost, and that's a slight judder effect. So overall, I would like to see more developers choose static shadows and lights in order to maintain framerate.
 
You seem to be misunderstand the situation. The market wants realistic visuals in a lot of games. Therefore, the have a product that'll sell to their audience, some developers don't have the option for cutesy games. Can you make a cartoony shooter? Yes. Can Infinity Ward make a cartoony COD? No, because then it won't be COD. Can you make a cartoony racer? Yes. Can Polyphony Digital make a cartoony GT? No, because then it won't be GT.

Nintendo (at the moment) prioritises games to be family friendly and visually light-hearted, so whatever game they make, they have the option of going cutesy and light-weight on the visuals, but they play to a niche. The whole market can't follow suit and serve that same niche. If every game in 2015 was Nintendo-style cartoony, gamers would be up in arms and developers would struggle to stand out and make sales.
Actually, Borderlands is pretty demanding and performance is lousy (:love:0fps on PS3 with poor IQ). Cel-shaded isn't really what we're talking about here, but simpler visuals it's easier to hit 60 fps with. Borderlands running at 60 fps on PS3 with clean lines would need a far simpler look - simpler geometry and simpler shaders.
Yes, on paper, but in reality it's not true when you're trying to make a competitive product that sells and your audience likes realistic visuals.
OT: You just reminded me of MotortoonGP or whatever it being called. I love the game and especially the stretchy car.
 
Back
Top