View Single Post
Old 12-30-07, 08:46 PM   #18
Gaco
L4D & DoW2 ftw!
 
Gaco's Avatar
 
Join Date: Jul 2004
Location: Denmark
Posts: 3,143
Default Re: Tweaking Crysis with Gaco Retail_Tweak_Config_v1.0 Can it get any better on Vista?

Naturally I'm still interesting in improving and optimizing my config having spend so much time into tweaking Crysis back at release. However I feel it somehow does more harm than good when somebody just comes around and says "lookie here I have a tweak which is better in performance and quality than Gaco 1.01" without specifying why. No offense but it's highly unscientific and just adds to the confusion in the config jungle. I know it's more work but you should really hold all parameters but one constant and test each one's impact and visuals and performance individually instead of just throwing a little of this and some of that in there. So lets try and make a direct comparison so we in the end eventually will end up with two configs based on preferences for various quality/performance assessions i.e. prioritisations, not "this is better than that definitively". To do it properly we still have some testing to do. I don't know when or if I will have the time to test, but here is the preparatory work I've done (comments in red italic):




PRESENT FOR BOTH, SAME VALUES:
-------------------------------------
sys_spec_GameEffects = 3
sys_spec_ObjectDetail = 3
sys_spec_Particles = 3
sys_spec_Physics = 3
sys_spec_PostProcessing = 3
sys_spec_Quality = 3
sys_spec_Shading = 3
sys_spec_Shadows = 3
sys_spec_Texture = 3
sys_spec_VolumetricEffects = 3
sys_spec_Water = 3

r_TexturesStreaming=0
r_colorgrading = 1
r_UseEdgeAA = 1

r_WaterGodRays = 1
r_sunshafts = 1

e_water_ocean_fft = 1
e_water_tesselation_amount = 10

e_view_dist_ratio_vegetation = 45
e_detail_materials_view_dist_xy = 4096
e_detail_materials_view_dist_z = 256

g_battleDust_enable = 1

r_TexturesStreaming=0

q_renderer = 3
q_ShaderGeneral = 3
q_ShaderPostProcess = 3
q_ShaderShadow = 3
q_ShaderMetal = 3
q_ShaderHDR = 3
q_ShaderVegetation = 3
q_ShaderIce=3
q_ShaderFX = 3
q_ShaderTerrain=3


PRESENT FOR BOTH, DIFFERENT VALUES:
------------------------------------------
r_MotionBlur = (Gaco=0, Nano=4) - you prefer motion blur? Fine, but as far as I recall, it takes a good chunk of performance, we can test it again. This is a setting of personal taste, I had it on in 1.0 and off in 1.01
q_ShaderGlass = (Gaco=3, Nano=2) - all of these q_Shader settings set to 2 means that it's equivalent to medium instead of high, but I take that you're well aware of that and have made this choice. Do you reckon this downgrades to 2 instead of 3 have a relatively big impact on performance? Otherwise I want them just left on 3 for myself..
q_ShaderSky = (Gaco=3, Nano=2)
q_ShaderSky=(Gaco=3, Nano=2)
e_vegetation_sprites_distance_ratio = (Gaco=1.5, Nano=1.2)
r_UsePom = (Gaco=1, Nano=0) - I just want to note that this is a relatively major player in both eye candy and performance - turning it off won't give you parallax occlusion mapping aka. "3D-textures" that adds depth to sand, rock and other stuff. Eats up around 2-4 FPS in scenes where it's used I think


GACO EXCLUSIVE:
--------------------
es_DebrisLifetimeScale = 1 - This and the next two is just minor stuff like ragdoll physics being applied to bodies and physics being applied to debris a little bit longer
g_ragdollDistance = 40
g_ragdollMinTime = 20

q_ShaderWater = 3 - I doubt that you can convince me to downgrade it from high to medium but you can try if you have a good reason?

NANOSUITGUY EXCLUSIVE:
-----------------------------
sys_budget_videomem = 512
sys_budget_sysmem = 2048 - Is this really necessary?

sys_physics_CPU = 0 - Isn't this the setting used when you want to capture a massive physics movie but how is this relavant to realtime gameplay?

r_TerrainAO_FadeDist=1
r_TexAtlasSize=2048 - I remember this one in particular as one of those where I didn't see enough difference to the single FPS I lost or something of that sorts. We can test it again if you think it's worth checking out
r_BeamsDistFactor = 0.02 - This is the beam setting below downgrades the beams a bit quality-wise from the high setting - how much does it give? Is it worth it? If you do not know the answer, this is once more a thing that needs testing
r_BeamsMaxSlices = 175
r_DepthOfField = 2 - Interestingly this looks like one of the few that gives a very high setting by making distant mountains blurry as well. I assume that this doesn't take up any real performance, so this might go in my 1.02 if I make such, but I won't make it for this setting alone

e_view_dist_ratio = 60 - I guess we need to test these e_view_dist_ratios as well but I think I remember tampering with them with no satisfactory result for performance at least. I could give it another try.
e_view_dist_ratio_detail = 30

e_particles_lod=0.7 - I'm pretty sure this is the one I saw Mad Boris' screenshots and performance comparison and there was barely a difference at all so I don't see any reason to downgrade it from high-default 1.0 to .7, correct me if I'm wrong
e_vegetation_min_size=1.5 - I think another one of those that I thought cost too much performance, but I'm not 100% sure about this one, can may have to test again

e_gsm_lods_num=4 - I don't remember having any experience with this and the next five parameters so this needs explaining or checking out
e_shadows_from_terrain_in_all_lods=0
e_lods = 1
e_obj_quality=4
es_OnDemandPhysics = 1
s_MaxChannels = 24


NANOSUITGUY EXCLUSIVE, OBSOLETE:
----------------------------------------
sys_spec_Sound = 4 - as far as I know you can't set any sys_spec commands to 4 in DX9 mode, so what is the deal?

r_MultiGPU = 0 - this defaults to auto detection (setting 2), why change it?

r_Flares = 1 - already the default value at high settings
r_Coronas = 1 - already the default value at high settings
r_Beams = 3 - already the default value at high settings

e_shadows_max_texture_size=1024 - already the default value at high settings




Now I remember fiddling around with those view dist values but at the time I thought it gave me unacceptable performance, some good 3 FPS lower than the 1.01. But Nanosuitguy it is simply wrong when you state that my tweak has performance equivalent to that of very high settings, it's VERY untrue. Lets work together with the tweaking and do the proper testing and in the end we can split on key settings that will come down to personal preference, but there's no reason to (unintentionally?) lie about anything. But we got some good testing ahead of us now if we want to get to the bottom of this
__________________
---MAIN GAMING RIG from MID SEPTEMBER 2007, MOBO+CASE+FANS UPGRADED APRIL 2009, GRAPHICSCARD UPGRADED MARCH 2010---
Case: Antec P182 case + 3x 120mm Scythe Slipstream SY1225SL12M @ ~900 RPM
Motherboard: Asus P5Q Deluxe
CPU: Intel Core2 Q6600@3.3ghz w. Noctua NH-U12P
RAM: 4Gb Crucial Tracer PC-8500
Graphics: XFX Radeon HD 5970 Black Edition
Harddiscs: 2xRaptorX 150gb 10.000 RPM in RAID0
Soundcard: X-Fi Xtreme Gamer
Sound: Sennheiser HD600 Headphones + headamp
Keyboard: Logitech DiNovo Edge
Mouse: Razor Orochi
Monitor: Samsung 2232BW (22" 1680*1050)
OS: Windows 7 Professional 64bit
Gaco is offline   Reply With Quote