NextGen (ps4/xb720) Game Development Issues

[TheChefO]

I'd like to hear the opinions of our esteemed game developers of the board what they feel will be the difference between the previous gen, this gen, and nextgen game development.

Target Hardware
If we assume that nextgen hardware will be based on 32nm/28nm process node and that the hardware will be a derivative of known hardware which will fit inside a 4billion transistor budget, we have a pretty good idea of what the capabilities of the machines will be.

• GPU - Radeon 6970 would be the upper limits given the transistor budget and TDP
• CPU - 6-9 core PPE or a 16-32 spu enabled Cell
• RAM - A bit of a question mark but if we assume 8x our current market that puts it at 4GB
• Media - BRD disks will likely be standard nextgen given the need for a physical medium and larger storage space

Increasing Development Costs
One thing we saw at the outset of this generation was a huge increase in development costs to meet the increased demands for higher game detail and fidelity. This lead MS and Sony to try and recoup some of the costs by raising the price of retail games for the first time since ps1 hit the market in 1995. This still did not prevent many of what most would consider to be "good" development houses going under in this gen after releasing top rated games that sold in the multi-six-figure range.

One thing that has happened with this huge upswing in development cost over prior generations is that many development houses have shifted over the years to a development model which outsources many components of game development.

Another is that the end result is significantly higher fidelity in game assets over prior generations.

So onto the questions for the developers:

1) Are there costs to be saved by further changing the methodology of game development?
Perhaps have certain development houses which compete in certain niches for enriching specific elements of game design. Similar to what has taken place over the years with UE3 for a game engine, or Euphoria for animation, or Physx for ... yeah, Speedtree for vegetation, outsourcing for 3d modeling, etc.

2) Do you expect a similar increase for development costs to what we saw between ps2 and ps3?

3) Do you expect an increase in development time?

4) What would you like to see change this gen in software?

5) What do you expect to see change this gen in software?



I have a few theories of my own, but I'd like to see what the experts think!

 

[TheChefO]

My theory:

Game development costs will remain the same or lower!
Game development costs rose double or triple what they were in the transition from ps2 to ps3.

Why the huge upswing?

Characters
Detailed 3D models with millions of polygons are created which are then broken down to more usable & efficient levels for this gen consoles.

This, in contrast to the few thousand that were used to create characters in the ps2 generation.

Game World
Applying nearly the same detail increase as the characters to in-game assets and the "game world" as well.

Textures
Along with the increase in geometry comes higher resolution textures. Higher detail makes it more difficult to assume a given look and more work is needed to create the same level of immersion.

Animation
More detail in the game world requires more attention to be paid for how objects move in the game world to keep the same suspension of disbelief.

Lighting
Pre-baked or Real-Time computed, the lighting (and shadows) this gen has seen a dramatic improvement. In the ps2 gen, flat lighting was the norm and most devs never had to pay attention to it. This gen, the lighting is ever closer to CGI and replicating the real world.

Cut-scenes/Cinematics
The quality/time/money invested here varies a great deal from developer to developer, but as always, the more attention/time/money put into it, the better it blends in with the game world being created.



So, how could Nextgen possibly be cheaper than this gen?

Better development tools

As we saw at the outset of this gen, utilizing technologies like Speedtree and UE3 enabled many developers to save time and money by not having to invest in "recreating the wheel". New versions of these same tools and more will enable developers to focus more on the gameplay & story, rather than fighting the clock and the hardware.

Specialization

Further, I expect more companies to begin focusing on specific aspects of game development to allow for greater advancements in these specific fields. This gen we saw UE3 dominate. More competition in that regard seems to be coming out of Crytech, Frostbite2, ID tech5, and a few other notables for their time like Capcom's framework engine and Avalanche Engine 2.0 by Avalanche studios.

If a few of these guys spun off (or focus internally) to concentrate on fleshing out these engines for other developers to utilize as Epic has done, this will lead to greater competition, variety, better technology, and cheaper licensing fees.

The same could be done for other technologies like Speedtree for vegetation. Euphoria for animation. Havok and Physx for physics, etc.

The one thing we haven't seen be spun off yet is AI. Obviously due to the complexity of the implementation, but this needs to be thought of in future iterations of game engines. AI should be design to be modular and engines designed to allow for this modular approach as it will allow for much better ai in game characters and at a cheaper rate down the road.

Libraries

In much the same way many studios use sound libraries for game soundfx, other libraries need to be built which are more accurate, more diverse, higher performing/more efficient, and cheaper/faster to implement.

Again, using the same ideology of not "reinventing the wheel", much of the advancement that went into making this gen stand head and shoulders above last gen (ps2) will be able to be leveraged for ps4.

Same, but Better!

The lighting will continue to get better, the textures will get better, the animation will get better, and so on and so forth, but it will mostly get better because the hardware will allow techniques and work that already exists, to be used at higher fidelity.

For others, I see custom studios setting up shop which focus primarily on one specialty: model creation, texture creation, story, level/game design, game mechanics, sound effects/soundtrack, etc.


In summary:

Games will be cheaper/faster to make by taking advantage of a market of specialty houses which do one thing and one thing only but they do it very well and by competing with others in that same field, they will push each other to keep getting better and offering products/services at a reasonable rate.

Game Studios as we know them today will be smaller, leaner, more efficient, and more creative.

They will be more along the lines of producers/directors rather than the one stop shop that produces everything a-z in-house as many are now.

The only weakness to this approach is by spreading the game around amongst many different companies/offices, it will be difficult to keep a lid on development for those that choose to do so. But if a subcontractor/individual is caught doing this enough times, they will likely lose business and find themselves out of work so there is always that balance.


I think in the end, this is the most efficient way forward to bring about lower development costs and keep the industry sustainable for generations to come. It will allow for certain specialists in style and form to develop and flourish, it will allow for certain Japanese developers to see their style of game see the light of day without bringing down the development/publishing house.



And most importantly, it will allow sustainable creativity to flourish.

This industry is too big to fail, but too inefficient to stay the same.

 

[TheChefO]

No responses? No thoughts on the matter?

 

[ERP]

Games will be cheaper/faster to make by taking advantage of a market of specialty houses which do one thing and one thing only but they do it very well and by competing with others in that same field, they will push each other to keep getting better and offering products/services at a reasonable rate.

Game Studios as we know them today will be smaller, leaner, more efficient, and more creative.

They will be more along the lines of producers/directors rather than the one stop shop that produces everything a-z in-house as many are now.

The problem with outsourcing is that in real terms it can be more expensive than employing people.
There are plenty of case studies that demonstrate this for various branches of software development, but the same thing applies to any of the disciplines.

Honestly I think we'll see more of the same next gen, low end titles won't balloon in cost the very high end stuff will, studios won't restructure.

The biggest cost difference was and is people, the mistake being made by the big companies is to keep very large numbers of people as permanent staff even when there is no work for them. It makes it impossible to control headcount and as a result cost. I've said for going on 10 years the model the industry should follow is the one used by the special FX business.

You employ your core staff (perhaps 10% of the total) full time and you hire contractors to do the rest, they'd still be in house, and you'd repeatedly hire the same ones, you would just pay for what you need, not pay a cast of thousands for something 10 people could be doing for the next 6 months.

And if we're concentrating on cutting costs does it really make sense to develop games in silicon valley, you're paying people 30-60% more to employ them there than 90% of the country.

 

[Rangers]

You employ your core staff (perhaps 10% of the total) full time and you hire contractors to do the rest, they'd still be in house, and you'd repeatedly hire the same ones, you would just pay for what you need, not pay a cast of thousands for something 10 people could be doing for the next 6 months.

Seems like Microsoft may follow a model to some extent like this in the way they deal with Epic. Epic makes the games, MS (I assume) isn't paying Epic's people beyond that. I understand that isn't exactly what you're talking about.

And if we're concentrating on cutting costs does it really make sense to develop games in silicon valley, you're paying people 30-60% more to employ them there than 90% of the country.

Again getting back to Epic, obviously North Carolina is a pretty big dev base. And of course Canada.

 

[TheChefO]

...I've said for going on 10 years the model the industry should follow is the one used by the special FX business.

You employ your core staff (perhaps 10% of the total) full time and you hire contractors to do the rest, they'd still be in house, and you'd repeatedly hire the same ones, you would just pay for what you need, not pay a cast of thousands for something 10 people could be doing for the next 6 months.

This is very much the train of thought I had. However, if these key people are doing nothing while waiting for the one studio to need them for their next project, how will they pay their bills?

My suggestion is have them work for a specialty studio that focuses on a specialty which can then sub these specialists out. It keeps the best of the best, full time, doing what they do best, refining their skill amongst others that are the best of the best.

The main development studio gets to do as you suggest (not pay them for time not worked) and the specialist gets to pay their bills on time by being under the umbrella of a specialist studio that schedules their work accordingly and ensures a steady paycheck.

Everyone wins.

Granted, the studio will likely pay MORE per hour than they would in-house, but they don't have to worry about "floating" them until their services are needed for the next project.

And if we're concentrating on cutting costs does it really make sense to develop games in silicon valley, you're paying people 30-60% more to employ them there than 90% of the country.

Another good suggestion and one which I see more developers paying attention to by dispersing their development offices outside the confines of California.

 

[TheChefO]

The problem with outsourcing is that in real terms it can be more expensive than employing people.
There are plenty of case studies that demonstrate this for various branches of software development, but the same thing applies to any of the disciplines.

Honestly I think we'll see more of the same next gen, low end titles won't balloon in cost the very high end stuff will, studios won't restructure.

The biggest cost difference was and is people, the mistake being made by the big companies is to keep very large numbers of people as permanent staff even when there is no work for them. It makes it impossible to control headcount and as a result cost. I've said for going on 10 years the model the industry should follow is the one used by the special FX business.

You employ your core staff (perhaps 10% of the total) full time and you hire contractors to do the rest, they'd still be in house, and you'd repeatedly hire the same ones, you would just pay for what you need, not pay a cast of thousands for something 10 people could be doing for the next 6 months.

And if we're concentrating on cutting costs does it really make sense to develop games in silicon valley, you're paying people 30-60% more to employ them there than 90% of the country.

I'm glad to have you contributing in this thread ERP!

Would you mind giving your thoughts on these?

1) Are there costs to be saved by further changing the methodology of game development?
*Direct subcontract specialists* -ERP
*Move development outside silicone valley* -ERP

2) Do you expect a similar increase for development costs to what we saw between ps2 and ps3?
*Majority, no. AAA, yes* -ERP

3) Do you expect an increase in development time?

4) What would you like to see change this gen in software?

5) What do you expect to see change this gen in software?

 

[DeadlyNinja]

Development cost is one of my major concerns for the next generation. It's crippled creativity for the most part throughout this generation.

 

[Laa-Yosh]

Asset production costs will increase, there's no escape from that.
Today's game character assets lack
- hair in general, and the necessary dynamics
- proper cloth and its dynamics
- proper geometry/silhouette detail
- proper deformations
- realistic shaders
- geometric detail

Datasets in the movie VFX industry are still at least 10-100 times higher. That's right, a Gears 3 character uses like a single 2K texture in the game, with about 10K vertices. An everyday movie character nowadays would use like 20-100 2K texture layouts, and 100-500k vertices, without the hair.
Sure, Epic builds 20-40 million polygon source models, but that part of the workflow is far from the most complicated in VFX production - building the model you render, creating the deformation systems, and painting those detailed textures is the really time consuming part.


I've got a little less insight to environments, but there is another very important issue - movies cheat a LOT. Most of the environments can be handled through 3D matte paintings, where an image is projected onto simple goemetry so the scene allows for a level of camera movement. This is not possible in a game environment where you have an interactive camera and you also want interactive lighting and possibly destruction as well.
Game environment artists are actually the unsung heroes of the game art industry, they have to deliver a LOT more than their movie VFX brothers, and their work isn't appriciated as much as it should be.


All in all, more dynamic environments and more geometry/texture detail are still going to be very, very expensive to produce. And there's little room left for improvements of any other kind in graphics IMHO - with proper linear HDR lighting implemented, the looks are lacking only in the level of detail, object density, and of course the dynamic aspect.


Regarding graphics programming, image quality will become the other big issue - right now the two main offenders are the unrealistic handling of geometry (very little detail, dynamics, and bad skinning/deformations) and the lots of various kinds of aliasing and texture filtering related issues. Coincidentally these two require the most resources.
Okay, shadows could still use some improvement, too

 

[Laa-Yosh]

Oh, and particles and volumetrics could use serious improvements too - movies are true 3D stuff using fluid simulations, rendered at full res with AA, whereas games use 2D texture mapped polygon quads at 1/4 resolution. Quite a lot of room left there, too.

 

[ERP]

This is very much the train of thought I had. However, if these key people are doing nothing while waiting for the one studio to need them for their next project, how will they pay their bills?

Ignoring a few weeks holiday, at large companies at least, there is never downtime between projects, you roll from one to another, the issue is that you don't want to roll to another project with the 100 people it took to finish the last one. There just isn't the work for them, so they end up doing busy work.

My suggestion is have them work for a specialty studio that focuses on a specialty which can then sub these specialists out. It keeps the best of the best, full time, doing what they do best, refining their skill amongst others that are the best of the best.

The problem with outsourcing design and pre-pro which is what your talking about here, is the team building the game doesn't own it, I've never seen a good game built where the core team didn't feel like they owned it.

Seems like Microsoft may follow a model to some extent like this in the way they deal with Epic. Epic makes the games, MS (I assume) isn't paying Epic's people beyond that. I understand that isn't exactly what you're talking about.

Not really this is more a typical publisher developer relationship.
I'm talking about an EA making 80+% of it's development workforce contractors, they end up paying them more per hour, but they regain control of head count and as a result cost.

Again getting back to Epic, obviously North Carolina is a pretty big dev base. And of course Canada.

Yes plenty of developers exist outside California, Texas used to be a big development hub as well. But many large publishers still have a lot of development in or around SF or LA.

Do you expect an increase in development time?

No, development time is already largely artificially constrained.

What would you like to see change this gen in software?

I don't really understand the question, game development is about so much more than software and programming, it's already very much about software engineering and logistics.

 

[TheChefO]

There just isn't the work for them, so they end up doing busy work.

So the alternatives are as you and I outlined. (are there others?)

Continuing to pay top dollar for busy work isn't sustainable.

I've never seen a good game built where the core team didn't feel like they owned it.

Interesting insight. Perhaps the relationship could be more along the lines of a temp agency where the employee is onsite working on the game, but is formerly employed by an agency/specialist studio.

No, development time is already largely artificially constrained.

Well, if detail is to be improved, either head count would have to go up, or development time would have to be extended.

Or, better tools and libraries are made available.

I don't really understand the question, game development is about so much more than software and programming, it's already very much about software engineering and logistics.

Logistics?

Actually, if you wouldn't mind, could you break down a typical budget for game development:
time x employees for x activity
or just monetary bottom line.

Textures
3D Modelers
Game Design
Story
Concept art
Voice acting
Animation
Music
Sound FX
Programming/tweaking (I'm sure this is more involved than just programers as assets might have to be adjusted to accommodate performance issues).

 

[TheChefO]

Asset production costs will increase, there's no escape from that.
Today's game character assets lack
- hair in general, and the necessary dynamics
- proper cloth and its dynamics
- proper geometry/silhouette detail
- proper deformations
- realistic shaders
- geometric detail

Datasets in the movie VFX industry are still at least 10-100 times higher. That's right, a Gears 3 character uses like a single 2K texture in the game, with about 10K vertices. An everyday movie character nowadays would use like 20-100 2K texture layouts, and 100-500k vertices, without the hair.
Sure, Epic builds 20-40 million polygon source models, but that part of the workflow is far from the most complicated in VFX production - building the model you render, creating the deformation systems, and painting those detailed textures is the really time consuming part.


I've got a little less insight to environments, but there is another very important issue - movies cheat a LOT. Most of the environments can be handled through 3D matte paintings, where an image is projected onto simple goemetry so the scene allows for a level of camera movement. This is not possible in a game environment where you have an interactive camera and you also want interactive lighting and possibly destruction as well.
Game environment artists are actually the unsung heroes of the game art industry, they have to deliver a LOT more than their movie VFX brothers, and their work isn't appriciated as much as it should be.


All in all, more dynamic environments and more geometry/texture detail are still going to be very, very expensive to produce. And there's little room left for improvements of any other kind in graphics IMHO - with proper linear HDR lighting implemented, the looks are lacking only in the level of detail, object density, and of course the dynamic aspect.


Regarding graphics programming, image quality will become the other big issue - right now the two main offenders are the unrealistic handling of geometry (very little detail, dynamics, and bad skinning/deformations) and the lots of various kinds of aliasing and texture filtering related issues. Coincidentally these two require the most resources.
Okay, shadows could still use some improvement, too

Thanks for your insight into this as well Laa-Yosh!

What do you think of the concepts I've outlined WRT 100% in-house development vs smaller teams relying on specialists (or specialist teams) which are employed by an independent source?

Do you think it would be a detriment to the overall game? Would there be ways to counteract this such as having them sub-contract directly into the game dev studio, or having teams within the specialty studio which stick together on projects?

What are your thoughts on ways to counteract the growing number of studios which financially cannot sustain themselves?

 

[Cyan]

Asset production costs will increase, there's no escape from that.
Today's game character assets lack
- hair in general, and the necessary dynamics
- proper cloth and its dynamics
- proper geometry/silhouette detail
- proper deformations
- realistic shaders
- geometric detail

Datasets in the movie VFX industry are still at least 10-100 times higher. That's right, a Gears 3 character uses like a single 2K texture in the game, with about 10K vertices. An everyday movie character nowadays would use like 20-100 2K texture layouts, and 100-500k vertices, without the hair.
Sure, Epic builds 20-40 million polygon source models, but that part of the workflow is far from the most complicated in VFX production - building the model you render, creating the deformation systems, and painting those detailed textures is the really time consuming part.


I've got a little less insight to environments, but there is another very important issue - movies cheat a LOT. Most of the environments can be handled through 3D matte paintings, where an image is projected onto simple goemetry so the scene allows for a level of camera movement. This is not possible in a game environment where you have an interactive camera and you also want interactive lighting and possibly destruction as well.
Game environment artists are actually the unsung heroes of the game art industry, they have to deliver a LOT more than their movie VFX brothers, and their work isn't appriciated as much as it should be.


All in all, more dynamic environments and more geometry/texture detail are still going to be very, very expensive to produce. And there's little room left for improvements of any other kind in graphics IMHO - with proper linear HDR lighting implemented, the looks are lacking only in the level of detail, object density, and of course the dynamic aspect.


Regarding graphics programming, image quality will become the other big issue - right now the two main offenders are the unrealistic handling of geometry (very little detail, dynamics, and bad skinning/deformations) and the lots of various kinds of aliasing and texture filtering related issues. Coincidentally these two require the most resources.
Okay, shadows could still use some improvement, too

Maybe next gen we will see graphics like this (the video shows a recreation of the Titanic using CryEngine 3)

 

[Laa-Yosh]

What do you think of the concepts I've outlined WRT 100% in-house development vs smaller teams relying on specialists (or specialist teams) which are employed by an independent source?

It's a complicated issue. Movie VFX studios rely on hiring the majority of people on a per project basis only, and train them for a few weeks to fit into their production pipelines. But VFX asset creation is a bit more generic and less dependent on custom tech and hardware constraints, so it's easier to get people productive.
It's also easier there to build up the infrastructure for temporary work force, wereas a game studio would have a little more trouble if they had to suddenly establish 50+ workstations, all the office space and software and hardware.

Dealing with off-site teams is a lot more complicated, costs more, requires more time, has more possibility for problems, so it's not a viable alternative. Sure, all the big studios like id, Naughty Dog or Guerilla Games utilize outsourcing, but mostly for non-critical assets. This model is quite different from the VFX business and has its limits.
There's also the question of profit sharing, the one thing smaller studios can offer to employees is a percentage from the sales income. This is probably why id, Epic and Valve can hire the best talent, although the fact that Microsoft managed to grab Kenneth Scott and Sparth from id suggests that they can offer similar bonuses

There's also a very significant difference between VFX studios and game developers - the former rely on external clients and their projects, most movie jobs last 6-8 months and they have to bid for the gigs all the time and hire temp staff depending on the workload.
Game projects on the other hand last for about 2-3 years, and it's also a lot easier to plan human resource requirements. Game dev salaries are also somewhat smaller as far as I know.

What are your thoughts on ways to counteract the growing number of studios which financially cannot sustain themselves?

The game industry already went through a lot of consolidation in the past decade and there are very very few independent studios left. Valve is pretty much the only one I could list from the top of my head - ND and GG are basically Sony 2nd party, id's a part of Zenimax, Bioware and Dice belong to EA. Most of the others have either been bought - or folded like Factor 5 or Team Bondi.

The vast majority of AA games today are produced by the super-sized internal studios of Activision, Ubi and EA. R&D costs are spread between projects, technology is shared, developers are assigned to projects from a central resource pool, and so on. I think the changes have already happened and the nextgen hardware will simply mean an average, say, 50% increase in costs and not as many changes as we've seen in the past 5 years. Beyond that the changes will be even more insignificant.

 

[TheChefO]

http://www.eurogamer.net/articles/2011-11-25-5th-cell-boss-the-USD60-boxed-game-is-a-broken-model

Another Developer taking issue with the current development model ...

Can anyone remind me what the definition of insanity is again?

 

[almighty]

I'm no game developer or programmer but expect a mega increase in geometric detail through tessellation.

And the hardware these machines might have better ship with tessellation units that far out perform what top end PC graphics cards can handle.

Heaven 2.0 swamps most tessellation units on PC and it has no were near the tessellation detail that games running on the next set of consoles could/should have.

 

[Commenter]

The only way I see it working is with very advanced third party tools at the disposal of developers. This doesn't necessarily entail genericness in games though if the tools are good enough to allow for unique design to show through.

 

[tuna]

http://www.eurogamer.net/articles/2011-11-25-5th-cell-boss-the-USD60-boxed-game-is-a-broken-model

Another Developer taking issue with the current development model ...

Can anyone remind me what the definition of insanity is again?

I do not see where the article claims the "current development model" is broken. It only claims that there is no market for crappy games at $60. I think that is a good thing....

 

[TheChefO]

I do not see where the article claims the "current development model" is broken. It only claims that there is no market for crappy games at $60. I think that is a good thing....

As Cliff Bleszinski said, the middle tier games industry is dead.

Something has to give.

When we see studios like Bizarre Creations, Pandemic Studios, among others shutting down, and most games this gen not turning a profit, the need to do something to change the development/retail process is immediate.

Especially in the face of even greater detail required for xb720/ps4.