Atomic Systems

Hi,
After 10 years on a moderate PC I bought some time ago a new PC with 16 Gb memory and a Geforce GTX 1660 Ti with I5-9600 3.7 Ghz processor.
So I set al TS2020 settings to maximum in the expectation that this would be stable.
It was not.
I started to play around with the settings and found that using in game anisotropic 2x is very stable where ingame anisotropic 4x gives a very instable game/simulator.
No OOM errors anymore with the anisotropic 2x settings.
I then set the configuration of the videocard on anisotropic 16x and all is stable still.

From the visual standpoint i'm fine, but is there a difference between these settings ?
Is one better than the other ?

thx
Albert

You know of this I presume?

http://forums.uktrainsim.com/viewtopic. ... 1&t=152120

I employed the suggestions and they work well.

Can you spot any difference in how the game looks when using it's Anistopic settings?

Hipper wrote:You know of this I presume?

http://forums.uktrainsim.com/viewtopic. ... 1&t=152120

I employed the suggestions and they work well.

I saw those but it applies to a GSync Monitor - which I don't have and cannot afford

I did have something from a few years back that Peter had done for normal monitors but I was running Windows 7 at the time

Peter

EDIT
I think this was it

TS2018 settings nVidia Control Panel [2918040 bytes] - Setting Up TS2018 with a single nVidia card.zip File ID: 38190 Date: 20 Aug 2018 - 841 Downloads

Hi Peter,

I have used peterfhayes NVidia Control Panel & TS2020 Settings "Set Up" values that are posted in the 'Link' http://forums.uktrainsim.com/viewtopic. ... 1&t=152120 (that you have given in your post above. I also do not have a GSync monitor and I just ignored the "settings" that refer to the GSync in my (GTX970) Nividia Control Panel and the improvement to my TS2020 has been amazing (smoother graphics & less stutters) and I would highly recommend that everyone at least gives them a try.

I suggest that you note down your 'original' settings before making any changes, then you can always go back to where you were before you started if you do not see the improvements that you require.

Thanks Rob - I will give that a try tomorrow.

Peter

I can't understand why a AA of 2x in TS20XX would cause issues.
AA is not well modelled in TS and although I use 16x AA in NVCP - I turned it "off" the other day and just used the TS 8x setting and I really couldn't see much difference.
However, leaving it on does no harm and it has no impact on the operation of TS.
I guess that in NVCP you are using settings for both Railworks.exe (32-bit) PLUS settings for Railworks64.exe?
pH

Peter,
Yes , I do.

The point of my question was : should I expect differences from in game AA vs Nvidea AA.
Like you I do not see much ( if any) difference from the visual aspects, but low AA seetings in game bring far less OOM errors on my PC.

Albert

jalmar wrote:should I expect differences from in game AA vs Nvidea AA.

So we're really looking at Anti-aliasing, not Anisotropic filtering?

You really shouldn't/wouldn't see a huge amount of difference on the monitor--both do the same job, with slightly different routines.

Yes, I messed this up a bit.
Anisotropic 'in game' 2x works OK, anisotropic 'in game' 4 x gives a lot of OOM errors.
Where anisotropic 8x in Nvidea configuration does not give OOM, this confuses me.
Although from the optical point I think there is no or hardly a difference.

jalmar wrote:Although from the optical point I think there is no or hardly a difference.

Working on the principle of "if it looks right, it is right" just use the 2x in game option. How much VRAM does your 1660Ti have?

Vram = 6 Gb

jalmar wrote:Yes, I messed this up a bit.
Anisotropic 'in game' 2x works OK, anisotropic 'in game' 4 x gives a lot of OOM errors.
Where anisotropic 8x in Nvidea configuration does not give OOM, this confuses me.
Although from the optical point I think there is no or hardly a difference.

Question for Gary or Peter then - would this be a case of a more capable program handling the anisotropic? The nvidia application would almost certainly be more sophisticated than the game engine, which is not exactly a spring chicken.

sbowness wrote: Question for Gary or Peter then - would this be a case of a more capable program handling the anisotropic? The nvidia application would almost certainly be more sophisticated than the game engine, which is not exactly a spring chicken.

My opinion is to let the game deal with things at the basic level, and use the nVidia enhancement routines (if needed/wanted) as just that--the final polish to the quality.
All personal preference, but I can't dee the need (on my system, for how I use the game) to *complicate* things. If using the nVidia tweaks makes a pronounced enough difference then obviously go with those, but the trick is to get it as good as it'll go in game, and then see how it looks with the nVidia stuff enhancing it, and also overriding it.

Sbowness
I agree with Gary - IMO the Nvidia driver is always going to be "better" for AA than the inbuilt one, BUT, we are dealing with DirextX 9.0c so the "gains" as such could be minimal. Even with DirectX 9.0c I'm not sure why DTG/RSC have limited AA to 8x in TS20XX - it doesn't make sense unless they know that a higher setting eg 16x, would not make any difference.
The AF settings in TS are reasonable and you can see differences between the various settings.
So, perhaps the AA/AF settings in TS20XX are adequate for the game engine used?
I set items like fps limit (if needed) in NVCP as that is most efficient, enhance others (AF, VSync, etc) but have to allow TS control of PhysX, it would have been much more efficient to let NVCP handle the latter.
Its indicative that TS now gives us an ambient occlusion option but it can't be set in NVCP - may show how the code may be working.
pH