Speculation: lack of a next gen media format may be a big problem

[Titanio]

Blu-ray disks will be too expensive to make them economically feasible compared to multiple DVDs

If they're only slightly more expensive to manufacture versus a dual-layer DVD, as per Sony's claim, then that probably isn't true. But in both cases we're talking about cents, not dollars.

and they don't provide a big enough jump. 10GB to 20GB? Big deal, going from 32MB carts to 650MB cds was a big deal, going from 650MB cds to 5GB dvds was a big deal,

Blu-ray offers up to 50GB - the jump from "5GB dvds" to that is larger than from "650MB cds to 5GB dvds"!

But this is all old news.

 

[gokickrocks]

If they're only slightly more expensive to manufacture versus a dual-layer DVD, as per Sony's claim, then that probably isn't true. But in both cases we're talking about cents, not dollars.



Blu-ray offers up to 50GB - the jump from "5GB dvds" to that is larger than from "650MB cds to 5GB dvds"!

But this is all old news.

according to some on the avs forum, blu-ray at the moment is expensive and wont be going down for awhile, yields are not that great and supposedly some have decided to use 20 of the 25gb capacity (still plenty of course) to be able to produce more (less data is written to outer edge which is causing the problems), and hardcoating process is not operating correctly in some pressing plants

 

[Titanio]

according to some on the avs forum, blu-ray at the moment is expensive and wont be going down for awhile, yields are not that great and supposedly some have decided to use 20 of the 25gb capacity (still plenty of course) to be able to produce more (less data is written to outer edge which is causing the problems), and hardcoating process is not operating correctly in some pressing plants

Well, those are varying claims. Even if true, these would be temporary issues. PS3 won't be temporary, it'll be here for the next 5, 6 years and beyond. And it's a moot point anyway, with all PS3 games having to be on Blu-ray.

 

[ShootMyMonkey]

But now you have another problem, because with 4 disks you need another 6GB, bringing your total to 38GB. That needs 5 DVDs to store it. But of course at 5 DVDs the same problem occurs, and now you need yet another DVD. In fact, this problem keeps on going until you will need a total of 7 DVDs, or 56GB of total data storage space before all the data can be made to fit on DVDs. So ultimately, in order to store your 20GB game, you've wasted 36GB, and your 3 disk game balloons into a 7 disk game.

Well, the only reason it works out like this in your case is because you chose to work with a case where the amount of shared data across discs is precariously close to the maximum capacity of a single disc. In reality, the amount of shared data (e.g. character models, textures, etc.) is comparatively small -- compressed on disc, a main character will be less than 50 MB, and the main character gets many times the content density of everything else in the game. On multi-disc titles, it's inherently a bad thing if you actually start splitting whole levels up across discs, so you tend to remanage the flow of things so that you minimize what is duplicated. I can't imagine any studio allowing more than a few hundred MB to be duplicated.

according to some on the avs forum, blu-ray at the moment is expensive and wont be going down for awhile, yields are not that great and supposedly some have decided to use 20 of the 25gb capacity (still plenty of course) to be able to produce more (less data is written to outer edge which is causing the problems), and hardcoating process is not operating correctly in some pressing plants

Wow... that's a load of nonsense. The cost of the disc is insignificantly small compared to other costs. Even if a single bluray disc costs 20x as much as a DVD, that means less than $3 on the cost of a game or movie. And the 20 out of 25 GB sounds like pure misinterpretation. All optical discs have errors, and many ECC layers have to be stored on a data disc since you require exact reconstruction of the data. That's why a CD, which has a physical capacity well over 850 MB can only store 700 MB of data. It's why a 9 GB dual-layer DVD only can store 7.5 GB of data. And it's no different for Bluray.

Why people were relatively honest about the capacity of a CD (~700 REAL MB for data), and not for DVDs or Bluray or for that matter, any hard drive since the gigabyte barrier was broken... well, everybody likes to cheat you.

 

[Pozer]

I still think this will be more of 2008 concern for devs. GC never seemed technically hampered by having a smaller media store.

As I look on my hard drive I see 3 modern games taking up 2.4, 4.0, and 6.2 gigs. All with more detailed textures than these 512mb consoles can display. Also they have many libraries to support thousands of pieces of hardware with scalability and mid-level APIs. I think the 360 is safe for awhile.

Anyway these free roaming world games people seem concerned about reuse alot of art. The first GTA for the ps2 fit on a cd-rom.

 

[zed]

The resulting unique MegaTexture is around 500MB in size. This represents a reasonable tradeoff between ETQW’s visual quality and disk space usage (maintaining a shippable size for the game).

sounds like thats not going to be a problem, and may actually save disc space over using conventional texturing methods

youre forgetting that its 500mb for each megatexture, ie i magaine quakewars will come with more than one map, otherwise itll be very boring eg 20maps x 0.5gb = more than one dvd
personally though i believe blu-ray isnt much of an upgrade comparred to dvds ie it holds only ~6x what dvds hold
something like this would be a worthwhile upgrade
http://www.engadget.com/2006/04/12/tdk-hard-at-work-on-8-layer-200gb-blu-ray-disc
dvds have existed what? one decade, i can see blu-ray being just as shortlived

the cost of blu-ray disks? the expect the early ones to cost under a dollar t make, + within 2 years theyll be giving them away in cornflakes packages.
if history is any indication

 

[gokickrocks]

Wow... that's a load of nonsense. The cost of the disc is insignificantly small compared to other costs. Even if a single bluray disc costs 20x as much as a DVD, that means less than $3 on the cost of a game or movie.

i was never talking about the cost of the media itself...but if you want to go there, at the moment hd-dvd (just assume that blu-ray can be as low as hd-dvd's for the time being) movies are about $10+ more than their respective dvd's

And the 20 out of 25 GB sounds like pure misinterpretation. All optical discs have errors, and many ECC layers have to be stored on a data disc since you require exact reconstruction of the data. That's why a CD, which has a physical capacity well over 850 MB can only store 700 MB of data. It's why a 9 GB dual-layer DVD only can store 7.5 GB of data. And it's no different for Bluray.

it seems like you misinterpreted what i wrote...they supposedly reduced capacity to get better yields, it doesnt really have much, if any, to do with error correction

 

[Mintmaster]

Compression methods aren't really getting better though. The difference between a hard-core compression algorithm and normal compression is only like 15% ish. It's not like games using .zip this gen and fitting on 3 GB can be compressed with UberCompressorX to 1 GB on next-gen hardware. More likely UberCompressorX will get the game down to 2.5 GBs if your lucky. The only way to go with compression is lossy, and hopefully better lossy techniques will work for images (JPEG2000) but that's about as far as it goes I think. There aren't many resources you can go lossy with. You don't really want lossy compression on your models or game code!

You're looking at this in entirely the wrong manner. We're not talking about universal compression methods, we're talking about techniques for games.

Imagine that for geometry, instead of storing every vertex, you store the actions of the artist and recreate the model dynamically on these ridiculously powerful CPUs. Filling up a DVD this way would need tens, maybe hundreds of millions of dollars in content creation. If each net modelling action took 10 bytes (after some simple compression of the raw actions), you could probably store a few million man-hours of modelling per DVD. This won't always be ideal, and you need some tools work, but I'm just giving an outline here.

Audio can consume a lot of space, but I don't see 50+ hours of audio per GB being a limitation. That's a lot of damn sound clips, and even 20 hours of cut-scenes is insane for a game.

Lossy image compression gives far smaller sizes than even DXT compressed textures, and I don't think you can often have more than 200MB of textures in RAM at a time when you consider everything else a game needs. A few GB of JPEG textures (with quality chosen to be appropriate for the final format the GPU reads) could easily be decompressed to over 50 times that 200MB space.

Content creation is really expensive. If you have enough money to honestly run out of room in a DVD, you can spare a thousand hours of programming time to get some creative compression in the tool chain.

Space is not going to be a serious problem, IMO, and won't be even close to something like N64 vs. PS1.

 

[Shifty Geezer]

You're looking at this in entirely the wrong manner. We're not talking about universal compression methods, we're talking about techniques for games.

To date, in the several threads on this matter, I don't think anyone actually presented alternative compression methods other than 'procedural stuffness'. I don't mind people pointing out how compression can save on capacity requirements, as long as they explain them.

As for your specific example, reduplicating the artists actions sounds very implausible to me. You'd basically need the modelling methods of Maya in your game with GUI, and run a massive macro on a subdivision surface or what have you. No matter how quick these processors are, you're talking a real lag there compared with reading the data from a disc. An HDD installation would get round delays during gameplay but then you're adding install times. Boo... And your not saving money either as the artists still need to create the data even if they save a big macro instead of a mesh. The only way to avoid the big costs with asset creation is lots of procedural methods, and I'd guess big libraries of models (Poser people absolutely everywhere!). And that generates space savings too if you can dynamically create trees for example. Something like EA's character modification in Tiger Woods applied to a generic human mesh to produce lots of characters is probably the way thigns will head. I don't know how these savings can be quantified though.

 

[one]

You're looking at this in entirely the wrong manner. We're not talking about universal compression methods, we're talking about techniques for games.

Do you expect great innovation in game compression methods in the next 2-3 years? I think diminishing returns apply to compression.

In Kameo and PGR3, decompression already takes 1 or 1 1/2 hardware threads.
http://download.microsoft.com/downl...ab-df6c7a2580b9/Coding_for_Multiple_Cores.ppt

Content creation is really expensive. If you have enough money to honestly run out of room in a DVD, you can spare a thousand hours of programming time to get some creative compression in the tool chain.

Big projects count on recycling engines and assets for sequels. For such games the cost for content creation is not getting bigger linearly. If UT2007 uses 7GB space, it needs another DVD on Xbox 360 to add new characters and maps for UT2008. Most racing games and sports games will be the same.

 

[Dave Baumann]

For such games the cost for content creation is not getting bigger linearly. If UT2007 uses 7GB space, it needs another DVD on Xbox 360 to add new characters and maps for UT2008.

You're assumine that all the content would still exist...?

 

[Crossbar]

You're assumine that all the content would still exist...?

If we look specifically at the difference between UT2003 and UT2004 AFAIR, basically all content of UT2003 was left intact and they more or less just added the "Onslaught" mode with brand new maps and some other minor things.
I'm not saying it will always be the case, but I think as maps keep growing and the level of detail is increased more storage will probably be welcome.

I imagine flight simulators could take some real advantage of some large storage space.

 

[inefficient]

As I look on my hard drive I see 3 modern games taking up 2.4, 4.0, and 6.2 gigs. All with more detailed textures than these 512mb consoles can display. Also they have many libraries to support thousands of pieces of hardware with scalability and mid-level APIs. I think the 360 is safe for awhile.

I think you just referenced the Windows OS there. ;-)

Anyway, when you are not including the OS + vendor API layer, even on PC gamecode is basically free in the scheme of things.

On the topic of system ram to media capacity ratio. MGS2 and MGS3 were over 4GB. This was a game for a system with only 24MB of system ram + 8MB video ram.

To put this in further perspective, let me evaluate a couple PC games I have installed currently.

Tombraider - Legend.
This easily the shortest Tomb Raider game ever released. It's only about 8 hours of game play even for a casual gamer. And my current speed run clocks the game at just under 4 hours. This game is a whopping 7,671,829,677 bytes on my HD. It's already pushing the limits of the 360's media.

NFS Most Wanted. 3,052,958,257 bytes. This is a game with graphics that a next gen GTA game would have to hit as a minimum! Yet, in scope and world size it is only a fraction of any of the GTA games. But it is a good starting point for a large seamless urban environment sufficently detailed.

GTA:SA. 5,043,591,996 bytes. The graphics in this game when compared to NFS:MW is like comparing a ps1 game to an xbox game.

Oblivion. 4,674,273,280 bytes. This game is just a taste of what is going to be possible next gen. Compared to MW this was a much smaller game. The number of unique environment types were low compared to its predcessor. Imagine a Dungeon Siege2 or a Kotor sized game with art assets of the same quality.

Speaking of Dungeon Siege2. 2,420,848,943 bytes

Fear. 4,933,534,900 bytes. This is large when you consider Quake4 is only 2,606,240,972 bytes. Fear does not offer that much more content than Quake or graphics an order of magnitude better. Still the difference is loudly pronounced in it's storage requirments.

 

[inefficient]

Content creation is really expensive. If you have enough money to honestly run out of room in a DVD, you can spare a thousand hours of programming time to get some creative compression in the tool chain.

Arists are generally not more expensive than programmers. And while increasing the number or artists from 10 to 100 artists at a project will in all likely hood result in the expected 10x increase in content created, going from 10 programmers to 100 will NOT result in 10x smarter, 10x more efficent code. However I would bet that the inverse is probably true.

 

[Arwin]

Regardless of speculation, Mark Rein himself indicates 20Gb plus for Unreal.

 

[Platon]

I don't get this whole thing actually.

On the one hand you have complains that games are getting more and more expensive to make and that a lot of developers and even publishers might not be able to survive these escalating costs.

On the other hand you have people complaining that a DVD will not be enough for next games. I mean what gives. If developers/publishers think that the costs for game creation are skyrocketing shouldn't then the limitation of the DVD disc be god sent, since they know that, unless they want to make a mutlidisc game, they don't need to create more content than what there is room for on a DVD...

 

[Crossbar]

On the one hand you have complains that games are getting more and more expensive to make and that a lot of developers and even publishers might not be able to survive these escalating costs...

Just like there will be good-looking low budget movies there will be good looking low budget games, right now there is a learning curve for the next generation games which probably push the price all over the line.

And you cannot assume that the price for content creation (per x MB of content) will stay constant. I believe there is a greater potential of improving content creation tools in order to improve efficiency than improving the programming tools. In one of the winter issues of edge they reviewed some round-trip model-editors where you could view your model "in-game" with all the right shaders applied without leaving the tool, it obviously was a great improvement for the artists, who had started using these tools.

I also expect there will be better tools for generating environments and objects from real world environments and objects, by using different kind of scanners, there are a lot of possibilities.:smile:
Remember there have already been some movies where the actors movements have been recorded by sensors in order to be used in games later on. If you can recreate the film environment in a similar (automised) way you may speed up things quite a bit (easy if the environment is cg ).

Larger storage space will probably give game creators a larger freedom of adding content of different kinds, will it be prerendered cg or tons of real world textures, etc. and that must be a good thing.

 

[expletive]

XNA to the resuce! (again)...

http://blogs.msdn.com/briankel/archive/2006/01/24/517071.aspx

Theres also a link to this article whch shows some quantitative analysis but clearly (imo) went in with the intention of proving there is no issue with DVD media...

http://gamesfirst.com/?id=1132

I'd be interested in some hard data on how big games were last gen, what was the breakdown by asset types (art, music, fmv, game code,etc), and why we expect the space requirements for those asset tyes to grow over the next 5 years.

I think thats really the only way you can quantify this because in terms of games i think we've reached the limit of 'bigger is better'. In terms of scope and gameplay 'hours' i dont expect that to grow at all this gen, maybe even shrink on average as AAA games become more and more costly to produce.

 

[rounin]

Here is an article arguing that DVD-9 is not enough

http://greggman.com/edit/editheadlines/2005-08-17.htm

 

[Shifty Geezer]

On the other hand you have people complaining that a DVD will not be enough for next games. I mean what gives. If developers/publishers think that the costs for game creation are skyrocketing shouldn't then the limitation of the DVD disc be god sent...

Godsend for devs perhaps who don't want to shell out on more than a DVD's worth of content, but for gamers, don't we want more? eg. What about the next-gen, should that be DVD too just to save on storage and keep costs low? Just as movies now cost inordinate amounts more than 20 years ago, as technology progresses and what you can achieve increases, and so do costs. Either you limit what you produce, to 2 hours of uber-hi-tech gaming instead of 20, or you limit the tech so it's not so expensive to produce 20 hours of game (like Wii apparently), or you provide a big medium and let devs choose how to use it.

Note : I'm not saying games will be exceeding DVD space, only pointing out the argument to Platon's observation.