Graphics options

[rubberduck]

Yesterday i played SR the first time, i started to learn how to drive in sr and i tested first functions of the track-editor.

i also have a question:

which graphics-settings do you have in smaller and in bigger tracks

are there some tricks to set graphics higher without getting slower (on standard settings shadows were "edgy")

i also read on https://code.google.com/p/vdrift-ogre/wiki/Running "Graphics options", (my computer is a new one with a good GPU (Linux))

[CryHam]

Well yeah the section Graphics Options already tells everything.
But in short: I use the default settings (the same as Higher), just without bloom (I never use any effect, they are cool but drop Fps too much).
And this on all tracks. Settings in general should work so that you don't need to change them for big or small tracks. But yes, I get like 100 Fps on small and 60 on big ones.
Better shadows also drop Fps. (I reduced default to 2 shadowmaps, and size to 2048).
Shadows aren't smooth here. We tried smoothed shadows but had some errors and didn't finish.
Also in editor there is a weird issue that they are very blocky if minimap is shown.
And also, if you look in direction close to sun light direction, they start being quite blocky for a small angle range, can't fix it.

[rubberduck]

I have some questions now.
First i see, that i don't get Fps of 60 on these settings
My computer has a GeForce 9500 GT (not the GeForce 9600 GT, what "minimum" on the running page says),
I also have the original driver for my gpu,

how much is the difference here?

The next thing is, that i play it under Linux (Xubuntu), is there a difference in other os?

My next question is about shader model?, what is that, have i to install something to make this work.



I also noticed, that for me the track infinity (extreme) is easier to drive than some tracks that are only hard or very hard (for example MntnTrail or Hole)

[CryHam]

There is no difference between running on Linux (with GL) and Windows (with Direct3D9).
At least from what I've seen on my hardware with Win7 and the same build on Kubuntu. And generally IMO there shouldn't be a difference (on same hardware).
Shader model is just something that is directly connected to GPU (its core). Generally SM 3.0 is present in all recent GPUs since few years (I think more than 5 years).

Next, I've wrote which GPUs I have, here on hardware requriements. So, on my GTX 560 Ti, I get 60 fps on big tracks and about 90 or more on smaller, using Higher preset (default) and without any effects.

Then, as for the cards, I compared with some benchmarked values,
from this site (And I'm guessing now that this list would have all of the cards that can handle the game well):

GTX 560 Ti - 3530 (my primary, Higher preset, >60 fps)
GT 640 - 1292 (my second gpu, didn't test on it a while, but should do 40 fps on High or so)
9600 GT - 753 (so called "minimal requirements")
8600 GTS - 307 (probably less than 9500 GT I hope)

Not sure if anyhow useful, links to compare GPUs: GT 640 vs GTX 560 Ti, 9500 GT vs GTX 560 Ti.
So I think if you see your GPU here (high) (or here (low)), read the benchmark value and simply dividing it with those above will tell you how well should it perform.

I'm not sure, but basing on those values, your card would do 10x less than mine, 4x less than my worse, and about 2,4x less than "minimal requirements". No idea if this is true. I would start trying Lowest, Low and maybe medium settings. See how it does. But 30 fps is probably max to have with it (on medium tracks?).

Yeah, I should probably decrease the difficulty for Infinity, few people already said that it's not so difficult.

[CryHam]

This is now a good post for everyone to post their performance.
Just put your GPU name (maybe already that benchmark value) how much Fps you get, and on which preset (and if you did change anything after setting that preset).

[Wuzzy]

I don’t think that would be a good idea. The results would not be comparable at all. The FPS greatly depend on the track and screen resoultion. On 0W13-MtFog you generally can expect a higher FPS than on, let’s say, 0W2-Wet, because less stuff is rendered on it. For me, this makes a difference of 20 FPS. We should at least define some common base before we can compare anything.

I suggest that we at least choose two so-called reference tracks, one with complex terrain/many models/, one with a near-flat terrain and a few simple models.
Complex: J20-JungleMaze
Simple: D1-simple

Here are my results:
Complex, Impossible (what a funny name for a preset ;-)): ~13 FPS (unplayable)
Complex: Ultra: ~18 FPS (unplayable)
Complex: Very High: ~20 FPS (unplayable)
Complex, Higher: ~40 FPS (ok)
Complex, High: ~62 FPS (good)

Simple, Impossible: ~20 FPS (unplayable)
Simple, Ultra: ~30 FPS (ok)
Simple, Very High: ~45 FPS (ok)
Simple, Higher: oops, I forgot.
Simple, High: ~52 FPS (good)

Card: GeForce GTX 460 from NVIDIA.
Resolution: 1680×1050

I consider everything below 24 FPS unplayable, everything between 30 and 50 FPS is playable but could be better, and everything above 50 is good.
I stopped testing as soon as I got over 50 FPS, since everything above 50 FPS does not make any difference to me.

Still my so-called benchmark has flaws. For example, the FPS are just estimated averages. It would be nice if SR could do this job, since humans tend to be pretty bad at this.

[CryHam]

Good call.
Of course: you need to quit game after each preset change, I hope you did, and that everybody will know this.
I would just add those:
- camera view: Follow close (9/17)
- car: ES
- both ghosts disabled (tab Replay - view)
Cars have a big impact on batches cound which does drop Fps (IDK if anything does that more, it's the 3rd value on left top bar).
I also like testing on J13-Tropic and F12-HighPeaks, on start position. J20 has too many objects which are 1 batch each still.