Quake3World

Well, yeah the topics says it all.
(Given it's 2011 and people have better computers now than when Q3 was realeased.)

I always try to keep it below 20k.

Trinity goes to somewhere above 30k in some places.

What number are you guys referring to?
What are the maximum visible shaders in your maps?
I've noticed that can be a significant fps factor

Could be cool if people posted some screenshots with speeds and fps included and maybe a bit about their system here

I don't have the level any more, but I think this was some 1M triangles.

In these days it is normal (my opinion) r_speeds about 30k. Even on my old unstable and dust-covered PC it was playable level (that with 30k).

Hipshot - was it playable?

I think the limit to r_speeds should be about as high as Focal Point in QL, unless you are building a terrain map, in which case you can`t help the fact that the map essentially has to be boxed because the terrain is detail brush...

You should also take into consideration how well the map performs with the recommended number of players for that map. You could have a map that averages 10k tris for 1v1 or 8-16 players. When the map is actually at full capacity, the overdraw from smoke and blood, plus all of the characters and weapon models may drastically affect fps.

r_speeds are just part of the overall picture so it should never be the definitive figure to judge performance. That will always be FPS. If you can get reasonably stable FPS on a 4 year old computer, I think you are good for your target audience for Q3.

obsidian wrote:r_speeds are just part of the overall picture so it should never be the definitive figure to judge performance. That will always be FPS. If you can get reasonably stable FPS on a 4 year old computer, I think you are good for your target audience for Q3.

Assuming that most people still have 4 year old computers, yes. But I don't think that's the case.

What are reasonable FPS? 100? (And don't say 24 because that doesn't allow for precision aiming :P )

Anyway, r_speeds should be a better measure, because r_speeds + info of system => FPS.

I think you should aim for having a stable 125 FPS, never lower.

The reason for aiming for 125 fps is the dependency of quake 3 movement of the fps count. IIRC there are a few fps numbers that "work" for q3. However, higher numbers increase jump height or something like that... The most important thing is having consistant fps because inconsistent fps inconsistent movement.

IIRC 333fps is even better. And with todays quad core set ups... why not?

333 is more or less unplayable. It lags from high FPS.

Keep in mind there's a difference between FPS and SV_FPS which is what you guys are talking about, in relation to player movement, etc. FPS is just the video card crunching the scene, SV_FPS is how often the actual entire game cycle is updated per second.

Not sure if my explanation is entirely correct, but as I understood it, Quake 3 updates the game state for each rendered frame. Whatever the explanation, the fact is that the amount of frames rendered by the videocard has influence on the game's physics.

To add another thing to the topic's title "r_speeds". You can usually count 800 additional tris for each bot/player that is added, don't forget about that. Especially if you're aiming for a 16 people ctf this tends to get important no matter how fast the system Q3 is running on might be.

stop bumping all the old threads o_o

Dammit, so it's been bumping after all, sorry man. Should have just been the normal "back for 1 day once every two months and quickly commenting everything that has happened so far", but that might have gone too far this time. I just saw that it's been 15 topics or something or so, that looks indeed So that's already it for the next few weeks again, I'm off into other real-life obligations.
Sorry again.