Predict: The Next Generation Console Tech

[scificube]

The PS2's VU0 and VU1 were dark and mysterious until very late in the generation. The GCN's TEV etc simply didn't get used much at all. NV2A and the Celeron in Xbox1 arguable never got a chance to shine because the console died an early death.

Without improvement over the entire generation there would be no meaning to "1st gen, 2nd gen, 3rd gen" games etc. so I struggle to understand why we should be anywhere near the finish line at this point in time in this generation. At year 2-3 of a console's life I have never seen a console mastered and I don't expect I ever will.

I can't say what gains SPUs are making for developers definitively. In all I would have to say something is going right on the CPU side at this point because launch PS3 games and today's PS3 game are not comparable on sheer performance if nothing else...and that's only looking at 3rd parties. At this time last year the state of PS3 games was decidedly 'bad.'

PC software lags behind hardware because not everyone has the top of the line hardware. Secondy, there a fewer development houses competing on the cutting edge in the PC space...and I don't mean that as a dig to the PC platform...just stating where things are trending. Competition drives progress.

A Mid-range GPU from 2011-2012 may suffice but more is better especially from a developer's perspective. Graphic expectations are only going up. When I refer to "handling" HD what I mean is let's say what Crysis is doing with very high settings 1080p at 60FPS - with headroom to do more. Crysis will be old news 2.5+ years from now. Still what's needed to pull that off isn't anything to sneeze at.

Developers aren't going to be burdened if they have more to work with. Developers are going to be burdened if they struggle to extrapolate anything valuable to their games from what they work with.

Giving developers less and telling them to do more is not helping developers.

I feel the focus is totally misplaced here. Developers needs ways to reduce coding complexity, speed up content creation, and reduce the costs of doing both. Lesser hardware solves none of these issues and IMO is likely to exacerbate the problems at hand in the face of ever increasing expectations.

 

[Shifty Geezer]

Without improvement over the entire generation there would be no meaning to "1st gen, 2nd gen, 3rd gen" games etc. so I struggle to understand why we should be anywhere near the finish line at this point in time in this generation. At year 2-3 of a console's life I have never seen a console mastered and I don't expect I ever will.

Although it's nice to have a constant progression of game quality from the consoles over their entire life, I think the Wii points to the possibility that a cap reached 2-3 years before a console ends would be acceptible to the public. If XB360 doesn't improve one jot between now and 2011/2012, that's not going to adversely affect the console. Thus the target for the hardware could be 'what level of hardware performance and complexity will the developers be able to max out in 4 years without having to jump through funny-shaped, moving, flaming hoops to get more out of the hardware?' If PS3 shows marked improvement every year all the way up until 2012, but the devs had to sweat blood to get it, are they and the business better off than if the PS3 was easier to work with and maxxed out earlier? How many people buy a console with the expectation that things will be much better 3 years later (save earliest adopters)? I don't think people will care if KZ3 is no better looking than KZ2, and if KZ2's quality was available to all devs witout too much cost, they'd prefer that. Plus whatever untapped potential PS3 has now, it's really struggled against the easier XB360. Wouldn't most gamers prefer all their games to be better on average? Are you happy having lower resolutions, lower framerates, and less AA for the first half of your console's life as long as in the second half it picks up?

Of course, the geek in me loves crazy high-spec'd hardware! But I'm not sure it makes sense building in a ceiling that's never going to be reached. PS2 is a great example. VU0 never really got used, so why build it in? Going with a simpler, less powerful choice would have been better for the devs. There's no point having potential that can never be used.

 

[scificube]

I disagree. I think the Wii audience is oblivious and I certainly don't think the same holds true for the PS3 or X360 audience. Wii is very interesting but I feel only Nintendo could ever pull off the Wii and I'm not convinced they can do it again. Nintendo is raking in record profits but I don't see that happening for for the majority of Wii ventures out there - those coming from 3rd parties.

The complexity of the PS3 is going exactly to my point. In this thread...in post 994 edit: not post 994 in this thread but another. In any case I mispoke. if memory serves I argue that Cell places the bar too high for the common developer. It isn't that Cell is too powerful to be of use. It is that it is was too difficult to come to grips with. If Cell had the same performance characteristics but came with JUST an instruction cache (an idea I like more and more) things would have turned out quite differently for Cell and even IBM.

The problem is still reducing complexity etc. I have not argued more power at the cost of NOT fixing these problems. Quite the opposite. I also don't feel you can only get usability or power. Both can be had.

I don't give a flip if the tech is cool for that sake alone. It's about whether it helps devs or not - it can be cool after that.

Just for kicks I won't say the GCN should've came without a TEV since so few actually made use of what it was capable of . I will say that majority of dev houses aren't seeking to push a console to the edge. The majority of dev houses in any generation meet the common level of competition and leave a lot of what could be done on the table. There is no way to make a machine where everyone or even a significant majority would utilize it 100% --- well I could think of a way but I'm not trying to sound funny here. Lowering what the hardware is capable of only lowers the average of what is accomplished on the console. Take the Wii for example -- it is by far the simplest console to develop for but at the same time there are not many development houses which have pushed the Wii for what it can handle.

 

[pc999]

For next-gen, if promises from the RAM companies are to be believed, that won't be an issues. We'll have many hundreds of GB/s. I don't know if contention has any impact, but it doesn't look like BW should be a problem to me.

Can you me such promises, I have been out of the loop (can only recall Z-Ram from AMD).

A Mid-range GPU from 2011-2012 may suffice but more is better especially from a developer's perspective. Graphic expectations are only going up. When I refer to "handling" HD what I mean is let's say what Crysis is doing with very high settings 1080p at 60FPS - with headroom to do more. Crysis will be old news 2.5+ years from now. Still what's needed to pull that off isn't anything to sneeze at.

Developers aren't going to be burdened if they have more to work with. Developers are going to be burdened if they struggle to extrapolate anything valuable to their games from what they work with.

Giving developers less and telling them to do more is not helping developers.

The question that I think Titanio is making (at least I am) is: are we still teling them to do more (from a no gameplay POV)? Specially if that came with a cost (HW, less content, less ideas being developed)

I mean some of the main/better/most wanted games have great visuals, but many of the most wanted games dont push that much (eg Wii, EA, and the companys of the like, games and the such). Sure there is evolution but it is slowing down in every sector of gaming.

On the other when it is enought ?
Personally I am playing a game like crysis with almost everything on high, and I most say that for a next game I would prefer if they invested their time in everything else before just gfx related things. Sure I would like better gfx but I can barelly name what I would like to see.

One thing that would be quite benefical is if we saw evolution of the gfx but not just more of the same, ie to give a destinctive look, a litle bit like KZ2 is doing.

 

[scificube]

So long as the gaming industry exists we can never reach the point when we have enough. At that point nothing is left -- that would be a sad day as gaming would essentially have run its course.

I won't enter into a discussion about what gamers are asking for from a game play perspective. This is the wrong forum for that. That said. Tech doesn't inhibit game play or the formation of ideas about game play. Tech is an enabler not something which disables.

As for not being able to name what you want technically in games going forward you would just be a normal person much like normal people that existed 10 years ago, 5 years ago, today and will exist in the future. There are many gamers who demand more and that is enough because many gamers have never been able to say what they wanted in clear technical terms.

No worries though as developers as always will come up with something to amaze you in the end. What needs to happen is that it is substantively easier for developers to do it.

 

[pc999]

Meybe I didnt expressed myself very well.

First, for the first time it is hard to say what one can want besides saying better or more of X, yet many people could say 5 years ago in a PS2 games that they didnt want big dot as shadows, they wanted a real shadowing across the terrrain and cast on himself (self shadowing), the same way that anybody could say that the vegetation should bend and the list goes on...

Second, for everytime a dev spends time and/or money trying to get more details on screen it is time that will not be used to get more content or to explore other ideias (or just making the game cheaper).

And on a personal note, from a gfx POV I have been litle amazed from some time on, finding destinctive looks much more interesting.

It is true that being easier to dev/programe would make wonders, but the question remais should they use the extra time/money to work on the gfx?

I only bring this because while this isnt a tech question, it can be one of the main factors for next gen. It already was been for Nintendo.

Anyway what kind of things are dev wanting to implement (and gamers seems to want) but that are performance limited?

Note that I am did not said that we should stop evolution/revolution in here, just that it may not (only?) need ever more performance and no linear costs/dificults, Wii or Spore, even LBP come to mind.

 

[scificube]

I can think of dozens of things gamers cry about. I can say quite confidently that what will be delivered will not be more of the same.

If you asking me to prove gamers want more you really only need to read more forums I suppose. Complaints/expectation abound. I was in a movie theatre the other day and heard someone say they wanted what they saw in games. What was frightening was that the individual was dead set certain it was going to happen this gen on his X360. Anyhow, I really don't want to diverge further into that discussion here.

I will say that there doesn't need to be contention between designers and engineers. Different people serve in those roles in a studio and so there should be no focus lost on game play. Designers have their own ideas about game design and game play. Typically designers make requests and are only constrained by what can practically be delivered with the budget on hand with the hardware available.

Having hardware doesn't imbue anyone with ideas. Having hardware doesn't take any ideas away. Creativity is mutually exclusive to the hardware available.

There have been many good ideas on expressed on all platforms this generation and most of the successful platforms of any previous generation. The hardware really is not the issue when it comes to ideas - it's people.

 

[eastmen]

First, for the first time it is hard to say what one can want besides saying better or more of X, yet many people could say 5 years ago in a PS2 games that they didnt want big dot as shadows, they wanted a real shadowing across the terrrain and cast on himself (self shadowing), the same way that anybody could say that the vegetation should bend and the list goes on...

Smarter AI , massive amounts of destructive enviroments not just one or two pillars as seen in gears 2 and killzone 2 with real physics where as no two attacks at a pillar is the same. More realistic movement in characters , higher resoultion textures to create more beautifull worlds.

Thats all stuff I want.

Second, for everytime a dev spends time and/or money trying to get more details on screen it is time that will not be used to get more content or to explore other ideias (or just making the game cheaper).

Price of a game has nothing to do with the development costs of a game. There are games on the Wii that are goign for $40 bucks that look worse than $10 games on live and psn .

IN the future perhaps devs should have a more unified aproach to development and instead of making a new engine and art assets each game in a series for the same generation they make better use of recycling all that fun stuff.

I only bring this because while this isnt a tech question, it can be one of the main factors for next gen. It already was been for Nintendo.

Nintendo still increased the specs of the console. They didn't just relaunch the gamecube and i'm sure that the wii is capable of things that the gamecube is not

I find that in every field people are given more powerfull tools and at first don't know what to make of it but in time learn to use those tools for either improving the current work they are doing or opening up into new ideas and ways of doing things that they couldn't before


I don't know what the next gen will bring. I'm sure it will bring many of the things we are seeing ni high end pcs and then some. All I know is that with everything people will grow acustom to it and when they go from next gen back to the 360/ps3 they will see a stark diffrence in quality and wonder how they every played on those systems. You can see this in every generation going all the way back to the first atari . There are still room for improvements that are much needed aside from what i listed at the start of the thread.

Nintendo may have gotten lucky this generation but I still don't know how they are going to sell a wii 2 or wii hd to these people buying wiis

 

[Dregun]

I agree with Eastmen;

The PS4 would be in a really good spot if they simply increased the power of the Cell and GPU (as well as RAM, etc etc). If Devs got into the habit of designing tools that make use of the resources they have available but are scalable they would be in a better spot next generation.

In the age of computers I can't believe we have throw away assets, why are these studios not creating assets today that they can use tomarrow? Spend the time and money on proper texture and modeling work today that once the resources (power) is available you wont have to scale down the assets as much.

Maybe I just don't know what goes on inside a game studio, but I would assume EA Sports would have had character models and textures already made 4+ years ago that have to be scaled down for todays technology. It seems to me that many development houses are not very apt on future thinking and are also usually the ones who complain so much about having to "struggle" with new console technology.

 

[Nano]

To the points above about engine and assett reuse, I agree, but I think we're observing a sea of change already. Middleware licensing is at a new high, and we're finding 'even' Japanese companies producing powerful, proprietary multi-platform middleware solutions (ie MT Framework, Crystal Tools) instead of making new engines for each game. With cost being a huge feature of this generation, I can't see developers going back to old practices, only expanding on the idea of reusing technologies.

Nintendo may have gotten lucky this generation but I still don't know how they are going to sell a wii 2 or wii hd to these people buying wiis

Personally, I think if the time is right, and the price is right Nintendo won't have much problem regaining their current audience. Also, come 2012 or beyond, considering they want to continue with a quiet, affordable console, I don't think it would be difficult to provide a superior package with a price lower than that of the current system (which is indeed a lot higher than it could be).

I believe many lessons learnt this generation will help manufacturers sell units next time round, I can think of quite a few myself

 

[scificube]

I agree with Eastmen;

The PS4 would be in a really good spot if they simply increased the power of the Cell and GPU (as well as RAM, etc etc). If Devs got into the habit of designing tools that make use of the resources they have available but are scalable they would be in a better spot next generation.

In the age of computers I can't believe we have throw away assets, why are these studios not creating assets today that they can use tomarrow? Spend the time and money on proper texture and modeling work today that once the resources (power) is available you wont have to scale down the assets as much.

Maybe I just don't know what goes on inside a game studio, but I would assume EA Sports would have had character models and textures already made 4+ years ago that have to be scaled down for todays technology. It seems to me that many development houses are not very apt on future thinking and are also usually the ones who complain so much about having to "struggle" with new console technology.

This is way off base.

Every game is unique and so assets cannot be reused. If a sequel is being made then old assets can be tweaked but new content must always be generated. This is unavoidable and cost lots of time, sweat, and money.

I don't think you understand what you are asking/suggesting. Do you really want all games to look the same? That is what you would get if assets were frozen and re-used over and over again. Do you want assets to look stale not up to what the hardware can reasonably handle? The same would happen if assets were used indefinitely or even as common practice.

Studios are not being irresponsible by not reusing assets. Developers simply do not have the luxury to do so and never have had it.

Since studios must always produce new assets for their games this process must be simplified, speed up and cost less if in the end you want more produced not less. Efforts like Collada can help but that isn't enough in my opinion. Every vendor offers a package of dev tools ranging from assets formats, to libraries, to offline tools etc. Every single tool needs to be more "aware" of each other. Tools need to eliminate tedium. Tools needs to have easy hook in/plug outs to facilitate easier intergration. Tools need to promote joint work across distinctly different disciplines. Tools need to be quick and responsive. Documentation and examples need to focused, easy to understand and plentiful.

I'll give an example of what needs to change.

I decide hey I'm going to use Collada to ease asset management between all these astranged tools floating around without a common point of entry --- sigh. Whoops - some of the tools provided to me don't understand Collada. I gotta learn what tool A needs from a dae file and write a plugin for it --- whops there's no way to write plug in for the tool --- guess I'll have to write a whole new tool just to handle conversions. Oh well I still need to convert dae files into some binary format my engine understands in the end anyway --- "maybe" I can re-use what I learn from this. So how do I read Collada files - why aren't there samples in some common place in my SKD already? Oh well there's still a site out there somewhere. I can't find the any examples of how to read the materials data from a dae file - guess I'll figure it out. Why wasn't this spec file already in my SDK or updated automatically (at least notify me!) when the spec changed? Good thing I noticed spec 1.4 isn't compatible with spec 1.3 before I spent two weeks writing a collada exporter - well I wish I noticed anyway. What's with this FCollada stuff? Why do I have to pay for it! I'll go with the Collada dom - crap you want me to use std::vectors or daeTArray's , std::strings or daeStrings - so confusing - every two steps I gotta check if I'm still using the right interfaces. WTF??? What is this Collada Refinery thingy - why wasn't I urged hard to use this from the get go! How do write conditioners? Where are the examples? This fox hunt is really getting me down - I've been on this for days!!! Aw crap somebody/something constructed these new dae files in a way I didn't expect. Gotta start over. Sure wish their were some collada to binary examples out there so I could just pic the one closest to what I need and just tweak it. I feel like Sherlock Holmes here...

I'm going over the top with all that but I want to demonstrate that things that should be simple "should" be simple and too many times they are not. When you get a workstation you should get darn near ready to go production environment or something you can make into one without inane hassle. If an artist sits down beside you you should not be at a loss to "show" them something or they you the same. I can't say how to get it done but the focus should be on figuring it out.

As for studios working towards scalable code I think many get the message but everything takes time. It's not as if it's an easy task.

 

[Shifty Geezer]

I don't think you understand what you are asking/suggesting. Do you really want all games to look the same? That is what you would get if assets were frozen and re-used over and over again. Do you want assets to look stale not up to what the hardware can reasonably handle? The same would happen if assets were used indefinitely or even as common practice.

I think the idea is assets should be created at 'perfect' quality and then scaled back automagically to the current hardware level. Thus with your thatched fantasy cottage, it's built at CGI quality and then scaled down to Fable 2. Come Fable 3, the same source model could be used and scaled down less agrressively. In 2,3,4 generations time the CGI asset could be used directly, with no extra creatiopn work beyond the original, a saving of 4x.

This is the same thing an theory as remastering video content. Someone was telling me just the other day about films that were digitally remastered for DVD, but now higher resolution has come out, have to be completely remastered again. So this time, the remastering is aiming for beyond 1080p quality so that whatever comes next, they're ready to just burn it onto whatever medium 4k or SuperHD++ TruVision Pro comes on.

Sadly it just doesn't work like that in game development. It would be great if games could work like movies with a huge bank of 'props' that can just be brought in, instead of absolutely everything being created from scratch per title.

 

[scificube]

I think the idea is assets should be created at 'perfect' quality and then scaled back automagically to the current hardware level. Thus with your thatched fantasy cottage, it's built at CGI quality and then scaled down to Fable 2. Come Fable 3, the same source model could be used and scaled down less agrressively. In 2,3,4 generations time the CGI asset could be used directly, with no extra creatiopn work beyond the original, a saving of 4x.

...

Sadly it just doesn't work like that in game development. It would be great if games could work like movies with a huge bank of 'props' that can just be brought in, instead of absolutely everything being created from scratch per title.

I agree in that I don't think it's possible to predict what assets you'll need in original games years in advance except for in a few corner cases exclusively tied to established franchises. The models used from present day CG in games probably could be reused but there's a high risk of getting locked into a certain look or feel for relatively little savings to the overall amount of work that needs to be done.

If we are not talking about original games then this idea becomes a lot more interesting to me.

The only time I think it could work is in the context of BC. With BC there is no expectation of seeing new locales etc and gamers would probably be both amazed and appreciative of how awesome their old games look running on new consoles.

This would add value to the gamers existing library and would make "classics" sold online much more attractive to consumers who don't own them. Could be a win if anyone was prepared for it.

 

[ban25]

Asset pipeline complaining...

Yeah, my team is dealing with this right now. It's such a pain, especially when 'standards' like Collada really aren't even that, rather just containers to stuff all sorts of proprietary meta-data.

 

[scificube]

Yeah, my team is dealing with this right now. It's such a pain, especially when 'standards' like Collada really aren't even that, rather just containers to stuff all sorts of proprietary meta-data.

Agreed. I understand the need for flexibility not at the expense of usability. <<extra>> stinks.

edit
I should say that Collada has good intentions and can be a help but there are times it really introduces a whole lot more complexity than you bargained for.

 

[Shifty Geezer]

I agree in that I don't think it's possible to predict what assets you'll need in original games years in advance except for in a few corner cases exclusively tied to established franchises.

I dunno. For some stuff, sports, racers, real-world locales, generic content seems perfectly acceptible to me. How many different people have modelled a Porsche 911 for games? Surely one perfect original-CAD-design source model could be applied to every racer out there with this car assuming automagical scaling worked. The difference in look would come from the renderer. The very same model in Forza 2 and GT5 would still look different (again, assuming the assets scaled and were optimized for the engine). Generic European streets and buildings, NY taxi's, geraniums, would all have a place.

There's also the question of how much variety we actually need in some content. To me, pretty much every sci-fi shooter has the same gritty, armour-clad space-marine style, such that it makes little odds to me! I think the assets of Oblivion and Fable 2 are sufficiently different to want unique content, but other games could share those assets quite happily. Gothic 3 isn't that far removed from Oblivion. So take Oblivion's architecture and orcs+goblins from an ADnD library of models, and you have the same look without having to recreate the content.

Throw in the possibility of applied realtime variance next gen with models getting tweaks to the meshes to mix them up a bit, and I think in enough case it'd be workable. I just don't think the scaling systems will work, and until there's a sufficient professional library of assets, there'd be no real start to the notion. Kinda like, if you were to try to piece together a scene using just free models grabbed off of some 3DCG website, the non-uniform styles would clash.

 

[scificube]

Is doing all of this re-working actually saving time and effort? How does this scaling thingy going work? (tessellation,format changes, endianness, inference as to what artist intends to happen,etc.). What if the car game is set in the present and your car models are from 5 years ago? How do we account for new techniques in modeling/texturing/effects/etc? What are the IP implications?

I couldn't say how much variance we need in games. As of now all games have unique assets as the norm. I am skeptical that gamers wouldn't notice the same elf in 5 different games. I also fear artists 1) Getting bored 2) Feeling shackled.

Sounds too risky all around to me. I'll hush now though unless someone else has something to offer on this idea.

 

[Shifty Geezer]

Is doing all of this re-working actually saving time and effort? How does this scaling thingy going work? (tessellation,format changes, endianness, inference as to what artist intends to happen,etc.).

In prinicple, next gen and beyond, using high level representations and efficient rendering. Objects could be, say, subdivision surface representations with tesselation and silhoutte masking. Or something funky. I do hope, though I appreciate it's unlikely, that will move on in terms of efficiency from our current 'lots and lots of polygons' solution as some experimental work is pointing towards. And even if not in engine, if you have a HOS representation you should be able to tesselate down to the required level with a bit of parameter tweaking.

What if the car game is set in the present and your car models are from 5 years ago? How do we account for new techniques in modeling/texturing/effects/etc?

A high-level object should be able to cater to that though. Whatever texturing mode is used, if the highest quality is available offline, it can be 'rendered' to a technique suitable for the game.

What are the IP implications?

Lots of content licensing instead of paying for artists

I couldn't say how much variance we need in games. As of now all games have unique assets as the norm. I am skeptical that gamers wouldn't notice the same elf in 5 different games.

A lot depends on how you mix you things up. We get the same human models in principle, but variation a la Home or EA's customizations show the same basic model can be varied beyond all repetition. But as Home also demonstrates, stick a hundred different models in the same clothes and it looks the same. So the same base elf model with one game adding a +50% to forehead, and another adding +100% to ear length, with different clothing from different libraries, and perhaps one in a more stylised graphics renderer, will be suitably different.

I also fear artists 1) Getting bored 2) Feeling shackled.

Well, the idea is the artists work in creating the libraries. Also, when has ruthless efficiency ever cared to keep the employees amused? IF this universal library concept were attainable and applied, and it did save significant amounts of money when every racer developer grabbed the same off-the-shelf assets, then the workforce would have to adapt, like any other industry. If a machine can make brushheads 50x faster than 10 employees, they get the boot, sadly, and the machine takes over. Of course, the ultimate drive for efficiency would then be machines doing everything and the whole world unemployed and unable to afford to be a partr fo this Utopia

Sounds too risky all around to me. I'll hush now though unless someone else has something to offer on this idea.

I think it's a possibility thwart with issues, but a valid consideration nonetheless. Certainly the logic of 5 different companies all modelling the same car or football player does lead to head-scratching, especially when they're struggling to make money. If they shared the cost to create that model and used the same one, it would be cheaper. The long term practicalities are a real headscratcher though, not least becuase the technology really hasn't moved on much. HOS has been tooted in games every gen, including this one, but it just isn't happening. Automagical scaling of high-end assets seems to be a pipedream as the engines are still to limited and require major optimizations to be efficient. If we had powerful enough hardware, this wouldn't be an issue, but I don't know if we'll ever reach that level of performance with the way things are scaling back. When I look at how far we've come, and then how far we have to go - GI, perfect shadows, complex shaders, realtime variation, all at 60 fps HD with buckets of AA and AF etc - developers will always be hitting a performance ceiling before they have chance to waste processor cycles on making their life easier.

 

[Rangers]

Not strictly on topic, but it looks like the next gen is in no hurry says Robbie Bach:

http://www.siliconvalley.com/latestheadlines/ci_11417526?nclick_check=1

Microsoft will rely on its Xbox 360 game console for longer than the previous version because it's getting harder to persuade consumers to upgrade, the head of the company's entertainment unit said.

 

[scificube]

Real-time assets don't use HOS today. You would have invest in using them and then hope what you've invested in has some practical application tomorrow which I've already cast my aspersions to. Since GPUs work with triangles you're almost certainly limited to working with poly representations of models that were produced offline.

Using Sub-Ds is already a common practice but unless memory is constrained it's better to divide offline->convert to engine format->load and save the cost of doing full on sub-d for every asset real time - I would leave it in the cool effects bucket and use when needed. It would make for easier to maintain code as well.

Sub-D won't make a Ferrari look like a pickup truck. You'll still need artists to plugin and fix things. Artist will still need to paint new UV maps if your Ferrari is altered in any way - say to look like a contemporary Ferrari instead of one from years ago. Artists will still need to assign material properties, apply new contemporary techniques to the assets and properly integrate them it into scenes in the game (lighting,shaders,harmonization,etc.).

In all but the most trivial cases the asset will still need to be sent down a considerable portion of the asset pipeline. I would much rather attack the efficiency of the pipeline itself than attempt to get around it all together. What I mean is that I think that would yield greater dividends.

---

If assets are stored in some proprietary format high level objects won't save you. You would need the secret decoder ring to use them especially if the assets have been obfuscated intentionally -- as any vendor would likely do.

A high level object cannot imbue an asset with information not already saved in some form in it. If the data isn't there it would need to be generated or worse case the asset need be re-created itself. I've provided some basic examples of this already.

---

Content licensing would seems likely to take dollars from artists you pay yourself and give it to whomever you're licensing your assets from. Hopefully you only pay once but I doubt it. For what you get you lose flexibility and creative control. If you request new assets be created for you, you're paying for artists to do it somewhere for you. You have to hope the vendor convinces the artists to work for less - or else you're paying them the normal rate and the vendor on top of that. The vendor would then be pure overhead.

Your game will look like the next game on the shelf who has licensed assets from the same vendor without significant re-work on your end.

I think it's unlikely you won't need your own artists on hand to re-work things. An artist working for you is an artist you pay once to do whatever you request. With this model you're likely to pay twice over for the artist you retain and the services of someone not under your umbrella.

If you integrate a 3rd party (for assets) into to your development team there are major confidentiality concerns as well - as I think you hint at - still there are yet more grave concerns.

Vendors could choose not to sell to you or arbitrarily hike up prices on you between contracts.

I just would feel safer controlling my own content.

---

Model swaps can work within Home but still these assets only have a context within Home. They will not work in WoW, Oblivion, etc. The upfront cost can be justified with Home because it "is intended" to be a long term project. It is difficult to do the same for an exploratory project as all new games that want to becomes franchises are. If you pay for the production of extremely high quality assets but the game does nothing on the market those assets could linger in dissolution forever. Most game projects are one-and-done. This is why IPs with credible "franchise" status are so valuable. You don't have it till you've got it.

--

I can't disagree with much you said in the last paragraph at all. It would be wondrous if everyone shared and shared alike but it would be hard to convince the majority it made business sense.

In short Shifty what you say sounds scarier the more I think about it. Perhaps I'm just too weary but I see tons of risk for marginal gains. Maybe I'm crazy.