View Single Post
Old 11-02-07, 01:05 PM   #28
Resident Tire Destroyer
FastRedPonyCar's Avatar
Join Date: Aug 2004
Location: Montgomery, AL
Posts: 3,012
Send a message via AIM to FastRedPonyCar
Default Re: My massive DX9/10 graphics and tweaking analysis of Crysis demo

Have you tried not using the autoexe and just chaning the game.cfg file with the info?

I'll have to wait until I get home but I want to try changing what I've got to these following values and the ones I don't have in the config file, adding them with the values listed below.

r_DetailDistance = 12 (can go from 1 to 20, higher is better)

r_WaterReflectionsQuality = 3 (Apparently the value of 4 doesn't jive with XP)

r_WaterUpdateFactor = 0.01 (can go all the way down to zero indicating water texture reflections get updated every frame. I may try this at zero also to see if it drops FPS).

r_useedgeaa = 2 <- still not sure about this one. I know 16XAA looks the best by far, but edgeAA helps it look better than no AA with zero impact on performance. A value of 1 blurs the edges Supposedly a value of 2 steps it to true anti aliasing. I've seen the screen shot gif that goes from no aa to 16Xaa to level 1 edgeAA and then level 2 edgeAA and judging from the gif, I don't konw if I like the added thickness of surface edges that level 2 creates but it's one of those things I want to see for myself before making a final judgement.

sys_flash_curve_tess_error=1 (creates a more precise per pixel - when set to 1 - wave)

e_lod_max = 1 (not used in previous config but can potentially increase the LOD of distant objects when I'm further away from them... sort of how the distant mountain side texture gets better as you get closer. this one is the one I think that could potentially crush the FPS). XL spreadsheet says the lower the number, the better the quality.

e_lod_ratio = 8 (Not sure about this one. 6 is default for high and very high, 8 is supposed to be ultra high)

e_terrain_lod_ratio =0.2 (another one not used yet but I think I've tried it before and noticed no difference in FPS.)

I've scoured the cvars settings spread sheet and compared what I'm using currently (settings in post 19) to what the excel spreadsheet says very high values and ultra high values are and from what I can tell, the options listed above are the only noticeable changes that could be made aside from the sun beam slice variables that I know are different.

Some of the commands above though could be real deal breakers though.. especially those LOD settings.

I'll definately post an update when I get home and get a chance to play around with it and run some benchmarks. As I see it now, if I'm benching mid 30's, I have about 5 fps I'm willing to sacrifice if my eyes can see a real tangible difference in the graphics. If not though, and I do lose 5 fps, there's no point in having them at more demanding values if my eyes get nothing in return...which is the whole point to all of this.
- Desktop -
4770k @ 4.45 ghz - ASUS Gryphon Mobo - 16 gigs Corsair Vengeance Pro 1866 RAM - XSPC Rasa 750 RS240 H2o Cooling - EVGA GTX 680 @1325mhz - 128 gig Crucial RealSSD C300 - 150 gig Velociraptor +2.75 Terrabytes of WD Sata 3.0 storage - 910 watt PC P&C PSU - Corsair Obsidian 650D case

- Server -
Core I7 D0 revision 920 @ 3.75 ghz - ASUS P6T Deluxe - 12 Gigs Mushkin DDR3 1600 - Corsair HX 620 PSU - Noctua NH-U9B cooler - Coolermaster WaveMaster Case

- Laptop -
MSI 16F2-012
- i7 2630QM - GTX570m @ 750Mhz - 8 gigs HyperX 1866 - 120 gig OCZ Vertex 3 SSD- 750 gig Scorpio Black - BluRay - 95% Gamut Screen - IC Diamond goop
FastRedPonyCar is offline   Reply With Quote