Will Dev Costs Flatten Next Gen (PS4, Xbox3)?

[MfA]

Texture reuse would be done at the object level ... cut parts out of existing objects and stick it where you want it. If you later draw over part of the texture it would automatically instantiate storage for the new data. This is simply Megatexture taken to it's logical conclusion IMO. I didn't say they have let go of vertices, I said they need to let go.

The problem with stereoscopic is that even if high end TVs can do it, without a home cinema setup it won't look that good without sitting close. If it only fills up a small part of your field of view it will be less than convincing. I think it will work better for PC gaming than for consoles, especially Nintendo consoles which I don't see being targeted at home cinema owners ... hell maybe I'm wrong though, maybe they will be able to convince people to wear VR goggles

 

[Fafalada]

Another 3 years of refinement and the prospect of it being both low cost and perfectly refined are quite appetising.

Hopefully they work out the kinks in polarized emitting displays in those few years. I still find 120hz 3d painful to look at, and while I haven't seen 240hz yet I am skeptical of IQ LCD can get with that kind of refresh.

 

[Acert93]

As I wear glasses I am dubious of any technology like this until hoards of glasses wearing folks swear by it. Remember, a lot of us nerds who buy this stuff first are nearly blind from decades of over exposure

 

[Fafalada]

Polarized stereo works fine as simple clipons (homemade even) or contact lenses.

Also I've seen several mentions of glasses-free stereo displays, though I have no idea what principle they work on and how good the quality can actually be.

 

[Laa-Yosh]

Personally I don't see why everything couldn't be done with 3D painting.

Stuff like Bodypaint and Deep Paint are already used for texture painting... the tools ARE there. But what you need to understand is that a very large part of the painting process is to use maps generated from the high res models used for the normal maps. Apart from normal maps the texture artist also gets ambient occlusion maps, cavity maps, color-coded maps for the various materials, and even some other stuff. All these are rendered into the same 2D UV space that the normal maps use.

All this means that a lot of time can be saved compared to starting from scratch. And also, most texture artists prefer to paint in 2D anyway.

 

[MfA]

If you are just going to reiterate the problem I'll reiterate the solution ... I like the appeal to authority argument better, at least it has some finality 2D mappings can be provided to the artist temporarily and then stamped on the 3D model.

Regardless of how the texture creation will be done "normal maps ... ambient occlusion maps, cavity maps, color-coded maps for the various materials" will still all be in one 2D UV space and their generation from the high resolution model is just as "easy" regardless how they are painted on subsequently. None of that changes with 3D painting, what changes is the parametrization of the surfaces, which becomes easy to automate and more efficient. If the texture has to be highly segmented to avoid inefficient allocation of texels so be it, the 3D paint program doesn't care and it's hidden from the artist. Seams in "illogical" places? Again the 3D paint program doesn't care and the artist can't see them (not to say that it's easy to handle the problems it causes with mipmapping, but let that be a developer problem rather than an artist one).

Pipe dream for now ... pretty sure it's where things are going though.

 

[Laa-Yosh]

MfA, please, don't try to tell artists how to work. They are more then content with the current method.

Also, there are tools like what you're talking about and it's NOT fit for the games industry.
http://www.disneyanimation.com/library/ptex/ptex.ppt
Doesn't give enough control and doesn't work with existing hardware rendering systems.

For example every seam breaks triangle strips and thus too many seams lead to very inefficient models...

 

[MfA]

Next generation is not existing hardware (and even on existing hardware there's the existing software renderers in CUDA which are so fashionable now).

Thanks for the power point ... but why are you making my point for me? All the reasons they stopped trying to parametrize textures for models in the classical ways are just as valid for games.

For example every seam breaks triangle strips and thus too many seams lead to very inefficient models...

The age of tessellation is coming, the efficiency on low end hardware of low poly modelled objects is neither here nor there. As high as the number of seams would be with automated unwrapping it's still going to be orders of magnitudes less than the geometry density next generation ... the relative magnitude is all that matters for efficiency.

 

[Arwin]

MfA, please, don't try to tell artists how to work. They are more then content with the current method.

No offense, but following that argument we'd never have had Pixar.

After reading some of the Siggraph 2009 stuff I'm firmly siding with MfA on this one. Although I can imagine there may be even better solutions than tesselation eventually, I'm fairly sure that textures will eventually be replaced with 'proper' 3D stuff.

 

[Nano]

I'm not sure. I don't think architectures will be as exotic as before, and I feel a lot of the knowledge from this generation will transfer really well to next. In addition, I feel multi-platform middleware and powerful tools to help make development easier (tech and art) will see a considerable boom (and bust). Costs tend to rise, but there are a lot of factors to consider. Even assuming there will be the customary rise in expenditure for next gen development, I don't think it will be such that it will strictly flatten games development.

Also consider, the industry is getting stronger and richer, and with a change in economy or market, we've also seen a change in the way developers big, small, conglomerates and independents operate to accomodate these.

 

[V3]

Is this 3D technology viable for handheld ? I assume handheld would need another gimmick after touch screen or dual screens. 3D screens might be a good gimmick for handheld.

 

[MfA]

If you hold it up really close to your face.

 

[fearsomepirate]

Dev costs have to flatten out for economic reasons. There isn't enough ceiling for them to increase by a factor of ten again. It doesn't really matter if a technological answer can be found for making even better models and environments. Publishers simply won't be able to sell games that cost $500m to make at a price that very many consumers are willing to pay, not if they're going to consume games at the rate they do now.

3D stereoscopic vision won't be the next big thing, at least not from a consumer perspective. If HD graphics and surround sound haven't resulted in a product as popular as last-gen consoles, I doubt even more "immersion" is going to light any fires.

 

[TheWretched]

What are you talking about though? I really doubt, that the most expensives games last generation did only cost 5mil to make and now are at 50mil.

 

[fearsomepirate]

Last gen, games generally cost on the order of $1m to make (Shenmue cost $70m, but it was the most expensive game ever made for quite a long time). This gen, they cost on the order of $10m (GTA IV cost $100m, but we're more frequently hearing of numbers in the $10-$30m range--Gears of War 2 was a $12m game). The typical cost of development is more important than the absolute maximum.

Did that clarify things?

 

[V3]

If you hold it up really close to your face.

How close is close ? Or you mean be eye level to the screen ?

Or do you need something like Virtual Boy ?

Virtual Boy was 3D right ?

 

[ERP]

To a certain extent the cost of games is no longer being driven by hardware complexity and features. But by the expectations of content, and game complexity.
I was discussing Genesis games with a friend recently and it occured to me that many Genesis/SNES games although advertising 10+ hours of gameplay could actually be finished by someone familiar with the game in well under an hour.
In contrast modern games don't rely on difficulty to pad out the experience, and there is an expectation that a majority of purchasers can complete the title. This leads to a dramatic increase in the quantity of content even without the cost/asset growing that's a dramatic increase in the cost of development.
I do think cost/asset is levelling off and I think we're starting to see shorter higher quality experiences in games, so yes I think the costs in general will start to level off.

And FWIW there are plenty of PS3/XBox360 titles made for a lot less that $200K, there just not the "tripple A" titles people crow about.

 

[Fafalada]

MfA, please, don't try to tell artists how to work. They are more then content with the current method.

I remember artists that were more then content doing things on graph-paper too