Are next gen consoles feasible?

What if the artist can just make his hi-res piece then just move on to the next thing?
That can only work with on infinite performance hardware. As long as we have finite boxes, we will have artists having to balance assets. Do they pick 10 characters at 20k each, or 15 at 13k each? If the choice is changed, the original asset is going to be reengineered. In a few generations time, the choice will be 40 characters at 200k each, or 60 characters at 133k vertices each. Sort of thing. Until we hit a wall where more detail makes no difference. Certainly next gen though, the same issues of optimiation will still exist. They only wouldn't if devs decide to make the same games as now, only less efficiently to keep it cheap and let the better hardware cope with the lack of opimisation. Instead they'll be trying to squeeze more and more from the hardware, trimming pieces here and there to enable more in other parts of the game.
 
That's true for poly stuff, but what about infinite detail technology?


okay that was a joke, but for scaling down geometry this is true, however it is an interesting approach with megatexture in that while you still optimize for performance you do it in a different way. All the content is created and then the whole thing can be scaled to the platform, with weighting depending on the location - so those mountains in the distance can stay highly compressed, and the stuff near the player is of good quality. Surely this is much quicker than the old method?
 
This is also interesting, albeit slightly tangential, on factors that influence how games are made:

http://www.eurogamer.net/articles/2011-08-24-why-microsoft-wont-publish-psn-firsts

It's not just about PSN firsts, but also about Microsoft's requirement for content and release date parity.

Last time I heard Sony has something similar mind you, but with a slightly different flavor (I like it better) in that if you want to release a title previously released on a different platform, you at least have to add some additional content. This is generally easy to do these days of DLC, as then DLC will be included for free in the later release of the PS3 version.

But these factors too impact the kind of content you find on disc / in games. Of course, Microsoft won't outright refuse titles if they are attractive to have in the first place.

At any rate it is the first time I saw this publicly stated. Maybe warrants another thread (or none at all).

Anyway, http://www.eurogamer.net/articles/2011-08-24-sonys-shuhei-yoshida-interview?page=2 seems to suggest that the launch of the PS4 will be at the earliest two years after the release of Vita, and perhaps even later. The way things are going here, we may be looking at 2014 for Microsoft, and 2015 for Sony. In that respect, Wii U may come at an opportune time and could become the graphics king for 2-3 years, being the best platform for games at least technically (we'll have to wait and see how this holds for the online/services part)
 
Well that's a novel method for getting around having a lot less space on their platforms media.
 
How much artist time does it take to scale stuff down for actual use in game from their higher detailed creations?

Takes quite a lot of time:
- model lowpoly mesh on top of the highres
- unwrap UV coordinates
- bake normal maps from highres to lowres
- paint textures, create shaders
- add skeleton, paint skin weights

The iterative approach can be problematic, ie. significant changes to the lowpoly can force a completely new start on textures.
Naughty Dog and some others solve this by texturing the highres model, and baking color and specular information to the lowres UV layout along the normal maps. It has its disadvantages though.

What if the artist can just make his hi-res piece then just move on to the next thing?

There's still the rigging part which may be highly problematic if the model is made from a million vertices. We in offline animation can of course build a lowres mesh and use various deformers to "wrap" the highres to the lowres, or just copy skin weights, but it's still not practical for games.


Increased hw capabilities will mostly effect the lowres workflow only, though. Highres source assets are already close to the level of detail that we use in our offline cinematics stuff (they're usually quite a lot more messy though).
Eventually I expect displacement maps to become standard, too, adding a little more complexity to the lowres workflow.
 
That's true for poly stuff, but what about infinite detail technology?

Actually, one of the reasons why many people suggest voxels is that you can skip the entire lowres model + UV map + baking part of the workflow, you just convert the highres source asset.
Still, it's probably going to require special tools to be able to edit unique geometry based game levels. Imagine the amount of source data Rage would require in that form! At least one, but probably two orders of magnitude more... Lionhead's Megamesh editor sounds like a good approach, though, but for now it's obviously designed for far, far smaller scale areas.

The other reason is of course that the combination of sparse voxel octrees and raycasting works very fast and render times are decoupled from scene complexity.

All the content is created and then the whole thing can be scaled to the platform, with weighting depending on the location - so those mountains in the distance can stay highly compressed, and the stuff near the player is of good quality. Surely this is much quicker than the old method?

This is not really true, you still have to build the lowres mesh, apply UVs, bake normals from somewhere and so on. Just because the Rage editor will set compression and maybe resolution on its own, the artists will still have to put in the same amount of work.

But again, Lionhead's stuff has a little more automatization, check their presentation to see more.
 
It is. Probably we will see even more use of multiplatform engines.. Fewer developers will adventure and build their own engine from ground up and so on.
I still think home console are a great business. Actually next-generation may be the biggest, but also the last, since when it will be over (2020-2025), computing will be completely different. For example, an improved network infrastructure may allow OnLive kind of services with great image quality.
 
I still would like to see something like a tablet kind of device that can use wireless controllers and networking, that you just put down in your living room and watch the game on the TV and sit down with the controller. Downloading games might not happen on such a platform though if they are to reach 100+ GB in size, at least I don't expect multiple terrabytes of storage in a tablet for a while...

Anyway, the point is to have a single device for most of your computing needs at your home, which is also portable and has a touch screen.
 
I still would like to see something like a tablet kind of device that can use wireless controllers and networking, that you just put down in your living room and watch the game on the TV and sit down with the controller. Downloading games might not happen on such a platform though if they are to reach 100+ GB in size, at least I don't expect multiple terrabytes of storage in a tablet for a while...

Anyway, the point is to have a single device for most of your computing needs at your home, which is also portable and has a touch screen.

Design a dock that for the device that would allow it to be slid in screen-down and that would supply power, additional storage, wired Ethernet, USB ports and A/V out. Then just use Bluetooth for the controllers and headsets.
 
Yeah, that makes sense of course. Although for now Apple seems to prefer their iCloud instead of local storage, we'll see if the general public shares my dislike for that approach.
 
How exactly do we count asset space anyway? Rage has terrabytes of source art that gets compressed to that 20GB, and we know they're using far more aggressive compression methods than most other game devs. It's not the same work as an Uncharted game with 20GB of art assets would be, but it's hard to decide how much more it is, since ND's artists may have to spend a lot of time on being clever because of their tech's limitations compared to Rage.

I wonder how high quality the rage pc versions more detailed megatexture is. Perhaps rage is already close to fidelity next gen consoles can use and there isn't a very big leap for ID in cost to support next gen consoles. Of course additional features like virtualizing models/higher vertex counts and so on would incur a cost to ID but the cost would not be in additional megatexture resolution.
 
Last edited by a moderator:
Yeah, that makes sense of course. Although for now Apple seems to prefer their iCloud instead of local storage, we'll see if the general public shares my dislike for that approach.

Don't think the general public will get a lot of choice in it.
Cloud storage is way to attractive to the platform providers, as I said earlier it's about owning users and if you own peoples data, the cost of a transition to another provider is large enough to retain.

I think you will start to see attempts at new ways to try and monetize games over the next 3/4/5 years.
 
I think I see new attempts at user control and monetizing every year. It's interesting just to look at the big franchies and how they've changed in ways to maximize nickle and dimeing. User control on consoles is minimal anyway considering there is no game modding capability.
 
Development costs will drop once the industry drops the archaic model for game development and move to something more efficient.

There no reason why the vast majority of art content creation has to be done in-house when the vast majority of art content creation isn't and doesn't have to be specially unique.

There is not enough development specialization when it comes to game companies. A company that specializes in animation and services dozens upon dozens of gaming projects would reduce animation costs while able to forward animation as a whole at a much faster pace than a bunch of individual dev houses all mostly operating independently of each other.

The same goes for modelling. Why not polish up a bunch of highly detailed characters collecting dust instead of creating a bunch of generic characters to fill out a scene. I rather have a game scene where you walk into a bar as a hard nose black noir PI only to see Snake as "Sam" working as a bar keep and Drake as "Gigolo Gerry" hitting on floozy #2 who happens to be the female Sheppard while being stop by Marcus Fenix whos playing "Bob the Bouncer" than a bunch of poorly created generic character that serve as scene fodder. That would be a pretty cool easter egg but in reality we don't remember 90% of the human characters created for games, you can drastically reduce their development cost by heavily recycling them across titles. Henchman #2 could show up in dozens of titles every year and most wouldn't recognize him across titles.

I sure recycling happens a lot within dev houses, but if you extend that recycling across pubs and developers, you would reduce art content cost tremendously.
 
Development costs will drop once the industry drops the archaic model for game development and move to something more efficient.

There no reason why the vast majority of art content creation has to be done in-house when the vast majority of art content creation isn't and doesn't have to be specially unique.

There is not enough development specialization when it comes to game companies. A company that specializes in animation and services dozens upon dozens of gaming projects would reduce animation costs while able to forward animation as a whole at a much faster pace than a bunch of individual dev houses all mostly operating independently of each other.

The same goes for modelling. Why not polish up a bunch of highly detailed characters collecting dust instead of creating a bunch of generic characters to fill out a scene. I rather have a game scene where you walk into a bar as a hard nose black noir PI only to see Snake as "Sam" working as a bar keep and Drake as "Gigolo Gerry" hitting on floozy #2 who happens to be the female Sheppard while being stop by Marcus Fenix whos playing "Bob the Bouncer" than a bunch of poorly created generic character that serve as scene fodder. That would be a pretty cool easter egg but in reality we don't remember 90% of the human characters created for games, you can drastically reduce their development cost by heavily recycling them across titles. Henchman #2 could show up in dozens of titles every year and most wouldn't recognize him across titles.

I sure recycling happens a lot within dev houses, but if you extend that recycling across pubs and developers, you would reduce art content cost tremendously.

The consolidation of the industry around a few game engines (most notable the Unreal engine) has already ushered in a era of sameness in visuals. This makes me worry about the effect consolidating around a small number of content creation studios would have.
 
Don't think the general public will get a lot of choice in it.

Yes, I understand your reasoning but I also see a good opportunity for lawyers and class action lawsuits. There's always going to be people who want to manage their own data, and it's likely that these guys will be smart as well, and they will fight for their rights.

Yeah, user agreements, whatever, I can of course see that there's a lot of room here to do all kinds of nasty corporate stuff. I guess it's more about me wanting to believe that there's going to be an alternative...
 
Development costs will drop once the industry drops the archaic model for game development and move to something more efficient.

Studio heads are very clever people, publishers have their own CTOs, and so on. These people are more aware of these issues than most of us are and they've already been looking into the possible solutions for years. They're not sticking to existing practices because of stupidity, laziness or a lack of an open mind. There's a realm of problems from technical through artistic to legal that keep studios from sharing stuff to the level that you'd expect them to do.

There no reason why the vast majority of art content creation has to be done in-house when the vast majority of art content creation isn't and doesn't have to be specially unique.

It has to be unique for a lot of reasons. Each game, even those using the same engine, have very different subsystems for animation, shaders, lighting, and so on. You can't just simply exchange characters between Gears and Mass Effect even though they're supposedly both using UE3.
Moving assets between MGS, UC, KZ, COD, and so on would involve redoing almost everything from scratch, except maybe the high res source art (but even that is completely different between various studios).

Even games like GT5, Blur and Forza are using completely different material and shader systems, physical models, texture and poly budgets and so on. Deferred rendering or forward rendering, 30 fps or 60fps, HDR or LDR, number of cars... these are all very important factors in determining poly and texture budgets and shader complexities. So it's just as impossible to reuse even the car model geometry between these games.

There is not enough development specialization when it comes to game companies. A company that specializes in animation and services dozens upon dozens of gaming projects would reduce animation costs while able to forward animation as a whole at a much faster pace than a bunch of individual dev houses all mostly operating independently of each other.

There are a lot of such companies but it's always very very complicated. Animation is completely tied to the pipeline of not only the given engine but also the entire studio. Gears used standard 3ds max bipeds at a time AFAIK, now they're also using Maya for facial animations; Mass Effect uses 3ds max with custom bones and rigging, even though both have data in UE3 formats in the end.


For example we specialize in CG cinematics, along with Blur, Plastic Wax, Axis, and a few others. The result is that today only Square and Blizzard have their internal teams and everyone else has pretty much stopped maintaining CG departments. It's a lot easier to outsource this kind of work because there's a lot less to sync (nothing we do goes into the game engine), and it's a small contained piece that's easy to lock down early on and requires relatively limited data exchange.
But we've done contract work for Heavenly Sword and The Club characters, and character animation for another title, and it's always been waaaay more complicated, despite a far smaller scale and a lot less people involved in the project.


And besides, outsourcing in itself is an incredibly complex issue, covering stuff like communication, feedback loops, constant technology updates and synchronization, education, art direction... We've outsourced some stuff and on average it takes twice as long to produce the same assets, even though we have very few technical restrictions compared to a game. Takes a lot more time from lead artists who can't focus on their more important tasks, requires a lot of extra legal and HR work (contracts, NDAs, money transfers) and so on.
 
Oh, and let's not even mention how inconsistent it would look to have characters mixed together with their own distinctive art styles. Just put some screenshots of Snake, Drake, Femshep, Fenix and the rest together and see how different they all are...
 
Oh, and let's not even mention how inconsistent it would look to have characters mixed together with their own distinctive art styles. Just put some screenshots of Snake, Drake, Femshep, Fenix and the rest together and see how different they all are...
That's them in costume though. The same natural looking faces in suitable attire for the scene should work. Not that I advocate dodwal's novel scheme, but there is a certain symmetry with Hollywood to be found in the same actors getting recylced. Of course in Hollywood's case it's the polar opposite, with big bucks spent on using the same people over and over, instead of going cheap with some unknown. But still, seeing the same virtual actors over and over wouldn't perhaps be as jarring as one first assumes with the interpretation of these characters linked so closely to their game styles.
 
No, that's also not true because nearly every one of them is stylized and in a different way.
Look at the color variety, smoothness, small skin details, and at the general proportions (Fenix in particular has an incredibly exaggerated face).
Uncharted is far more colored and vibrant than either ME or MGS, but those two games are very different from each other as well. Maybe you can't see it but trust me, if those assets were put into the same scene they'd look like everyone's from a different planet.

I'd not argue with reusing virtual actors if all games had the same art style, but that is not the case. Fortunately, I might add.
 
Back
Top