RE: [GD-Consoles] Nobody on this list?

[Tom F. <to...@mu...>]

> -----Original Message-----
> From: Steve Rabin [mailto:St...@no...]
> 
> 
> >>> "Jason G Doig" <jas...@py...> 07/03/02 
> 12:23PM >>>
> >Obviously I don't speak for other platforms where you might 
> still be nuked from orbit.
> 
> Roger that. The nukes are positioned and standing by. 
> Awaiting target acquisition...
> 
> (that was a joke)
> Yep, unfortunately many things are covered by NDA and it's 
> hard to talk too candidly in a public forum like this. 
> However, there are some good cross-platform subjects that are 
> worth discussing. I'll throw out some tasty ones:


I'll bite. We're doing PS2 and XBox versions, no GC at the moment. But I can
make wild extrapolations based on friend-of-a-friend rumoured GC specs if
you like... :-)

> Issue #1
> An important issue for cross-platform developers is how the 
> same art/textures will appear on each of the platforms. My 
> sources tell me that Microsoft recently released a paper 
> comparing XBOX to PS2 in terms of color/display output. I'd 
> like to hear from developers what their experiences have been 
> across all three platforms with respect to texture 
> appearance. Have textures needed to be changed in any way 
> when porting from one platform to another?

Fairly simple. PS2 mipmapping is expensive and approximate. So only the
textures that really need it get mipmapped, so the artists need to be
careful with the high-frequency stuff. XBox also has bigger textures because
of the greater memory and better compressed texture support, so for the PS2
we just ditch the top mipmap level of everything.

Also, since the surface effects available on XBox are much fancier than on
PS2, things like specular maps and so on get pre-composited in together to
one or two textures rather than the half-dozen we have on the XB.

> Issue #2
> Consoles have limited memory when compared with PCs. What 
> general compression algorithms (if any) are people using on 
> their titles (custom/zlib/lzo/platform specific). Do you hold 
> any data in memory and then decompress it on-the-fly when it 
> is needed? How concerned are you with patent issues when 
> considering a compression algorithm on your title?

We're using LZSS in places. We generally store compressed stuff on disk
(hard drive for XB, DVD for PS2), stream it in as we need it, and decompress
as we get it. The cost of LZSS decompression is far far less than the speed
boost we get from having to load half the data (typically ~50% compression
from LZSS).

We haven't tried storing stuff in memory compressed - I'm not sure that
would work terribly well. We'd rather use the memory to hold more stuff
cached off disk.

> Issue #3
> For cross-platform games: What platform is your primary 
> platform and at what point do you start porting to the 
> others? During development, what methods do you use to 
> guarantee quality and stability across all platforms? Do you 
> design/code your game for the lowest common denominator or do 
> you build it on the best (perhaps on a top-notch PC) and then 
> gracefully scale down?

Our primary platform is the PC! Our artists aim for PC quality, and we then
scale down in all sorts of ways to the target platform. It's pretty easy to
scale down to a target - bin mipmap levels, merge textures, reduce shader
complexity, reduce vertex count using VIPM, reduce bone counts using
progssive bones, etc. But going upwards in quality is very hard without
actual artwork to base it on.

Some of the artwork was started at PS2 quality before we realised we could
render 2-4x as many polys on XBox, and bumpmaps weren't done early enough.
Inventing that data without redone artwork has been ... tricky. I wouldn't
want to do it again. But if you have bumpmaps, then compositing them down to
"prelit" stuff for the PS2 is pretty easy.

> Issue #4
> Each console maker has a (different) plan for how online 
> games will work. Are you excited by any of these plans and 
> how will it affect the direction of your games? Are you going 
> to jump onboard or sit this one out until the next generation 
> of consoles?

Not done any console games this is applicable to yet, so it's not something
I've thought about much.

> Grab an issue and run with it. Issues #1 and #4 are timely 
> ones and issue #2 holds a certain fascination for me (I was 
> bit by the compression bug about a year ago).
> 
> Steve Rabin
> SDSG - Nintendo of America, Inc.
> Editor of "AI Game Programming Wisdom" (www.aiwisdom.com)


Tom Forsyth - purely hypothetical Muckyfoot bloke.

This email is the product of your deranged imagination,
and does not in any way imply existence of the author.