Scalable Graphic FTW?

............................................________
....................................,.-...................``~.,
.............................,.-...................................-.,
.........................,/...............................................:,
.....................,?......................................................\,
.................../...........................................................,}
................./......................................................,:`^`..}
.............../...................................................,:........./
..............?.....__.........................................:`.........../
............./__.(.....~-,_..............................,:`........../
.........../(_....~,_........~,_....................,:`........_/
..........{.._$;_......=,_.......-,_.......,.-~-,},.~;/....}
...........((.....*~_.......=-._......;,,./`..../............../
...,,,___.\`~,......~.,....................`.....}............../
............(....`=-,,.......`........................(......;_,,-
............/.`~,......`-...............................\....../\
.............\`~.*-,.....................................|,./.....\,__
,,_..........}.>-._\...................................|..............`=~-,
.....`=~-,_\_......`\,.................................\
...................`=~-,,.\,...............................\
................................`:,,...........................`\..............__
.....................................`=-,...................,%`>--==``
........................................_\..........._,-%.......`\
...................................,<`.._|_,-&``................`\

One of the difficulties of PC gaming is the arms race that naturally occurs as newer technologies are released and game developers design games to take advantage of them.  Sure, we get gorgeous games like Crysis or Fallout 3, but the emotional and fiscal costs can be trying.

I've been having lots of email discussions with gamers lately, and Abdul sent me a long (but interesting!) email on this very subject.  He had some great points, basically boiled down to the fact that even when you do everything right and get the right parts, your game experience can still suffer.  What a frustrating experience!

Most of the time, when I upgrade my system, it's not because I'm particularly excited about the next new graphics card/sound card/whatever, it's because I want to increase the time it will take for games to overtake my system's capabilities.  I think of it as annoyance insurance.  And like most long-term solutions, it can be a tough pill to swallow. 

This is the price of being a PC gamer - you must be a little more hardcore, a little more technologically savvy and a lot more patient.  But does it always have to be this way?  To make PC gaming flourish, and to draw more people into the hobby, PC gaming has to be easier to get into.

Rudedog over at fpsadmin made the excellent point that the baseline PCs that are being sold these days are far more game-capable than they used to be, but Abdul made me think that there might be another solution - scalable graphics.

Now, almost every game has scalable graphics to some extent, usually revolving around tweaking all sorts of options and knowing all sorts of enigmatic terms.  What I'm talking about is making the low end lower, so that games can be played on a higher percentage of computers.  Sure, they may not look as good as the dev would have liked, but increasing your potential customer base would have to make up for that in the long run.

It would also make upgrading much more fun - now I'm not buying annoyance insurance, I'm unlocking new content!  Games that looked passable before become a new experience, and I can see the immediate effects of my upgrade. 

I've designed a lot of games, but I don't write code, so it's not clear to me how difficult this type of scalability would be to implement, but the possibilities it presents are certainly exciting to me. 

Comments

[RR]

I understand where you're coming from with this line of thinking, and I agree to a point.

But I believe what we really need, is two things:

1) The baseline configuration you can buy from any PC vendor needs to improve. Specifically, when it comes to the GPU. No more low-end, crappy integrated graphics! One of the few comments of wisdom that ever came out of Mark Rein's mouth at Epic, was that Intel's crummy graphics chips were holding back the entire PC gaming idustry.

I realize this will increase prices, and that business customers won't care and would rather save money on their purchases. But the price difference between a really poor GPU and a decent GPU can't be that significant.

AMD's current "AMD Game!" program is a step in the right direction, that's been long overdue.

Microsoft is apparently a part of the PC Gaming Alliance. I would hope that they will try to put pressure on PC vendors to improve their offerings. I know it's difficult because the PC is such an open platform and it's a free market. But something needs to be done.

2) Enforced quality standards on hardware device drivers. I know that Microsoft has been making an effort here, and I think it's helped. But they need to keep pushing harder. Poorly written drivers are probably the number one cause of crashes in games. Nvidia's early Vista video drivers were apparently notorious for causing crashes.

I think I read somewhere that Microsoft is planning to add a feature to Windows where manufacturers and their drivers will be rated by their reputation. That drivers which are known to be unreliable will be flagged as such, which will put pressure on those manufacturers. If so, then I can only praise the idea and state that this, too, is long overdue.

Oh, and automatic updating for games (also apparently planned) would also be a welcomed new feature.

[Mordenkainen]

On the topic of whether scalability with hardware is easy to implement, today's mainstream games target consoles and (sometimes) PCs so there's inherent scalability already because the consoles are now less powerful (even if more efficient) than your high end gaming PC where the game is put together.

Draw distance is something that has to be scalable because you don't know up until the last stage of game development what is the sweet spot of geometry/texture LoD required by the game's assets so you always have a slider functionality. This can then be used fairly easily for low-end through high-end pc scalability.

Also with textures there is inherent scalability because the art is authored at much higher resolution (and uncompressed) than what is ultimately used so it then becomes a simple extra step of processing the same common pool of art into x different bins of quality (resolution and compression format).

Even for geometry complexity scalability this is also inherently available because most games use geom-lod for the draw-distance so it becomes fairly easy to say that the LoD2-model of a certain prop is drawn at 100 feet away in the console versions but 50 feet away in a low-end PC and 200 feet away in a high-end PC.

The only major problem with scalability comes with A.I. routines and fragment/vertex shaders where the console's static hardware allows for simplyfied development whereas on PC you really have to write a lot of versions of the same "effect" for different hardware level (and sometimes hardware brand -ugg-).

Anyway, this is a completely solvable problem. I think games that don't take scale well on PC are mostly due to budget constraints and general indifference to the PC platform than any real architectural changes that need to be implemented to support this.

[Ryan Miller]

@RR:

I think that is a key part of the solution, however it won't do much for the millions of PCs already out there. I think in the long term, software and hardware have to meet in the middle. Scalable graphics helps in the short term.

[Abdul]

I don't think wishing the quality of baseline PCs will improve is the right way to move forward. Thats what the trend was in the past. But the trend changed, which is why MS did not anticipate that they'd still be licensing XP instead of Vista on low-end PCs. I do know that Intel is working on improve their graphics accelerators, but in general, you will rarely see a graphics card capable of playing today's game on an entry-level PC.

Not everything is scalable easily. A lot of work would have to go into scaling physics/AI down if CPU speed was an issue (which is why FIFA 09 for PC used a mix of the graphics from the 360 version and the gameplay from the PS2 version; the 360 version's AI routines were too CPU intense for the average PC). But CPU speed is not an issue in most games; I'll submit to you that we do have a decent baseline level for CPUs these days.

Instead, the biggest stumbling block for hardware requirements is the graphics card-- yet it is likely the easiest to scale down to. As for requiring a certain Shader Model capability for a few lighting effects, I think that smacks of laziness or a lack of understanding of the PC games market by game developers. Its impossible to regulate against something like that. But if MS is serious about PC titles, they need to lead the way with first party development of games that do take into account the broader base of PC gamers. You don't have to look further than Blizzard to see how effective a low entry level for hardware can be.

[Abdul]

@Mordenkainen: I agree with most of what you have said. I just look at what Bethesda did with their last 2 games and its quite sad. They do have a scalable engine, but their insistence on requiring shader model 2.0/3.0 cards for Oblivion and SM3.0 for Fallout 3 has limited their market greatly. I had a GeForce 3 at the time of Oblivion, so I paid close attention to this issue, and a hobbyist pretty much made a SM1.x patch (oldblivion) within a few weeks. And now the same thing is happening with Fallout getting a patch.

It just shows me that PC game developers aren't spending that extra amount of time to ensure that they can reach a large audience.

There are more headaches with PC gaming than just hardware requirements, but its definitely the first thing that needs to be fixed if the PC Gaming Alliance/Games For Windows wants to get on the right track.