Where does one look for evidence of the -1.5 LOD Bias? Are there particular features that will show what that setting does? I am having difficulty finding any difference between 0 and -1.5, though it may be some tracks benefit more from this than others depending on how the textures are constructed and implemented...?
Yep, you guys are both correct. See this short summary (old article, but parts are still relevant)... http://naturalviolence.webs.com/sgssaa.htm or here ... http://www.overclock.net/t/1250100/nvidia-sparse-grid-supersampling In "Configure Sim", I use the following: View attachment 14709 In-game, AA is set to 16x. In player.json: "Heat FX Fade Speed":0, "Max Framerate":60, "Texture Sharpening":0, I don't know why 8xQ works with 4xSGSS, but it does (YMMV). I have cataracts in both eyes so my goal is to obtain a sharp image, particularly of cars in the distance. Also, I never see lag, experience stutters, or tearing. On occasion, I do see some minimal tear in replays. Note that my GFX card has 4gb.
From day one , "any settings" , rF2 has had the best distant detail in a sim period, isn't that obvious. AC pCARS and the rest, cars in the distance are just blobs on the horizon. Nothing to do with drivers or settings.
8xMSAA still works with 4xSGSSAA but it's probably not as good IQ-wise as 4x and 4x. Or it may just be the same IQ but at lower framerates (8xMSAA is obviously more demanding than 4xMSAA). 8xQ is essentially just regular 8xMSAA (just clarifying that for others); have you tried 4xMSAA instead of 8xMSAA? 4xMSAA is theoretically the best in your case (when using 4xSGSSAA).
IQ from best to worst on my rig is (with rating related to jaggies and other typical ISI annoyances): 8xQ and 2xSGSS - 8/10 8x and 2xSGSS - 5/10 4x and 4xSGSS - 5/10 4x and 2xSGSS - 4/10 And I choose not to use 8xQ and 4xSGSS because it is virtually indistinguishable from 8xQ and 2xSGSS (I would rate it 8.25/10), but has a significant performance hit.
I find it odd that 8xMSAA + 2xSGSS looks so much better than 4xMSAA + 4xSGSS, I'm starting to wonder if all those AA articles are incorrect.
The railings and fencing in the pits at Portugal and Silverstone are the best comparison points for anyone who wants to judge for themselves.
Useless. You have to see the moving image to notice the major part of the improvement. And, no, it likely wouldn't show-up in a crappy-quality YouTube video either. Try it for yourself. It's not difficult to judge for yourself and pick the settings you prefer and that work best on your set-up.
Marc, I hadn't tried 8xQ and 2xSGSS so I decided to give it a go. IQ ratings are very similar to what you listed. Using 8xQ and 4xSGSS is only marginally better, but at a significant FPS hit and probably not worth it. I did end up setting "LOD Bias (DX)" to '-0.5000' when using 2xSGSS. Spinelli, like Marc, I found IQ of 4x and 4xSGSS to be less than either of the above. I know this is contrary to what is supposed to happen, but it does (at least for Marc and I). I don't know why it works. I run everything on max except road reflections (low) and shadows (quality). Also, I'm not running triples (GPU is only 760 OC). Driver version = 344.48
Carlo, can you answer my question from above about negative LOD bias? Someone suggested -1.5. You are suggesting -0.5. 0 is default. How/where do you detect the difference(s) this setting makes?
You'll only notice the difference if you have texture sharpening at zero. In that case the further negative you set the bias the sharper the textures will be in the distance, most noticeable on painted lines. Conversely if you set a positive bias the textures will blur out closer to the camera. Basically the more negative bias you add the sharper the textures will be but the sharper textures also come with more aliasing (jaggies).
Thanks for the very simple-to-understand answer!! Now for a follow-up. What does the "auto" setting for texture sharpening do/how does it work? I can understand the specific settings, or off, but not auto. Mine is set to auto, so that likely explains why I couldn't see any impact of changing the negative LOD bias from -1.5 to -0.5 to 0!!!
Auto means use the bias set in the materials by the track creator. I would have called the option default rather than auto, somewhat misleading! Sent from my SM-G900F using Tapatalk
So using 0 should give a more consistent result in case the track creator used inconsistent applications of bias (or more likely track converter was stuck with them and unable to fix)? AND it also allows us to override with a global setting from the driver?
The LOD bias adjustment has been "broken" for around 4 or so years now and only works ATM with SGSSAA enabled, correct?
Marc, TechAde nailed it. Also, note that I have (player.json): "Texture Sharpening":0, This is what Inspector looks like ... View attachment 14716 LOD Bias -.50 results in a bit sharper image of vehicles out in front (look at decals) without a notable increase in jaggies. Anything more and IQ goes down. I prefer this, but others may prefer a bit softer look.
Spinelli, nVidia "clamp" is broken, but LOD Bias itself does work for me (SGSAA). Just be aware that it is non-selective in that it affects the whole image. I haven't tried it without SGSAA enabled.
