View Full Version : UT2k3 Poly Range

03 March 2003, 06:22 PM
For player models what is the highest polycount you should go upto and still have decent gameplay? Assume the computers being played on are atleast 1 Ghz processor with 256 MB ram and a GeForce 3.

03 March 2003, 07:28 PM
i belive 2500-3000 is a nice safe number...

03 March 2003, 08:26 PM
hmm I thought it would have improved over Quake III.

03 March 2003, 10:34 PM
hmm.. if memory serves me correctly.. quake 3 models were actually around 2k....

03 March 2003, 11:48 PM
didnt they have this bullshit discution just a few threds ago?

oh one more thing, I belive they were around 800 in quake doh :scream:

03 March 2003, 11:52 PM
That's tough to say Guardian, because it highly depends on what you consider decent gameplay... 20? 30? 60 fps?

The models that come with UT2003 are within the range that monotypic mentions. Some go a bit beyond.

As for the Quake 3 poly counts - the models that Steed made (that king among men!) were in the area of 1200 polys.
I don't know how high the actual end user models went (they could have gone up to 2k). Of course, you're going to pay that price on the engine, and your computer.

03 March 2003, 03:30 AM
Interesting the Gundam Universe mod uses models between 2k and some upto about 3k. From what I understand it runs smoothly. For allowable I consider 30 fps at 1024x768 on a low end system such as the specs i gave .

03 March 2003, 11:15 AM
much depends on how many players are on the same map!
(and on the map itself)
i had a perfectly fluent gameplay with 4.5k poly but on CTF-MAGMA i wouldn't have such a performance even with 800 poly characters!

anyway... to use more than 3k is forgivable mostly if the result is a masterpiece!

03 March 2003, 01:58 PM
I heard lots of things about how that map is a comp killer. Something about the particle effects I think? :hmm:

Another thing.. The downscaling of poly such as in the mini player mutator has no benefit to speed right? No matter the size the polygon count is still the same and takes just as much processing power?

03 March 2003, 02:19 PM
You can sure use 3k models with the Quake3 engine like JK2 did, at the expense of the level design, which will have to remain a boxy BSP environment.

With UT2 you can have 3-4k models AND detailed maps, or 5-6k models and crappy maps again. It's really up to how you juggle it. The proper thing would be sticking to 3-4k models and have decent maps too with custom static meshes.

03 March 2003, 03:01 PM
Is there really a need to even go beyond 2-3k for FPS games at the moment? Crap I would think the ideal would be 1.5-2k polygons per Character. Characters are moving around and jumping so fast most detail gets lost in a blur of gibs and weapon effects.

Most people who need that many polys too tend to use them inefficiently. Like some will slam 500 just in the face alone. Were even if the screen size is being drawn at 1024x768 the average faces you will see is around 64x64 pixels. Thats more polygons that pixels.

I personaly think more detail should be spent on level design and detail. Take Doom3 models for example. They are actually not that uber high if you look closely. Most of the details can be done with unique use of shaders and Normal Mapping.

I would think good speed would be 60fps. If you can have 32-64 people in an Map fighting at 3-5 polygons a peace. All players in that scenario would have to have way uber machines to run at 60fps.

03 March 2003, 03:23 PM
I'm pretty much just curious what the max poly count should be limited at. Lowest acceptable gameplay would be the stuff I mentioned before. 32 players at the max in a game. The maps ranging from decent to detailed but all preferablly large in size for lots of movement. But onto my main reason for asking setting a limit of 3.5k reasonable limit. I expect decent systems to be running this not scrap boxs. Thinking of how long mod development takes the computers of current will be pretty sub-grade to whats available then.

I'm wonder what the map size limitations are for ut2k3. They must have a size limit in dimensions.

03 March 2003, 07:51 PM
For most systems, the recommended max poly count for UT2k3 is 3,000. You can go higher than that, and in time I would expect most systems to handle 5k or 6k models without a problem, but right now stick to sub 3000 if possible, with an absoloute max of 4,000 say.

Yes, you can go higher than that right now if you wanted, but it'll irritate people when they have to go and adjust their settings in order to use your model smoothly, especially with the spec machine quoted at the beginning of the thread, and on certain high detail maps.

Sic1 - I think if you go look at the polycounts in Quake 3 you'll find they are all under 1,000 polys. Xero is one of the highest at slightly over 900.

You may be thinking of Team Arena, where Id decided the average machine had increased in power, and adjusted the polycounts to reflect that. The counts in Team Arena are indeed about 1200 per character.

Oh, yeah, the map sizes. They are limited, but it's an insanely huge size. Can't remember the exact size, but it's not recomended that anyone actually use the highest possible size right now.

03 March 2003, 09:57 PM
Thank you everyone for all your information and views on polycount. Mecha generally seem to use abit of polies if you go detailed and make something from anime. for now I think I'll say 3.5k triangles will be the limit. As things go along and the next gen video cards come out I'll have to take note at how big maps get.

Currently I am now trying to find out about a coding issue dealing with different scale player models and the difference in perspective each player would have. I'm not sure if its possible to have alil guy move around at a speed to his scale and see everything larger. FOV works like a zoom but things actually appear to be closer "zoom" than they are instead of being big.

CGTalk Moderation
01 January 2006, 02:00 PM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.