nVidia AA problem - Black screen & Main menu/Readables switch problem - by bikerdude
bikerdude on 2/4/2009 at 12:57
Afternoon
This may or may not be know to everyone, with regard to the odd issue where when in game with T1/T2 you frob a readable or hit escape to enter the main menu but the game stays on the game screen or you start the game, can hear sounds but just gte a black screen.
I have noticed this "issue" since my first nVidia DX10 card and never got around to finding a fix for it. But after some testing, I found this "issue" only happens when 4xAA mode is enabled either globally or per game profile. The following AA modes are all unaffected - 4x, 8x & 16x.
I dont know how/Why 4xAA effects the dark engine like this, but maybe someone with more hardcore knowledge of the dark engine & Direct3D will be able to offer an explanation.
So far this problem has been found on the following hardware:
GTX280 - (
http://www.ttlg.com/forums/showthread.php?t=126009)
9800GTX - (
http://www.ttlg.com/forums/showthread.php?t=126486)
At some point I will see if this "issue" effects DX10 ATi cards aswell.
baeuchlein on 3/4/2009 at 17:35
I have tried it with two (older) ATI cards. So far no problem.
My Radeon 9550 uses Catalyst 5.2 (old but reliable for me) with Windows XP. The Control Center is installed as well. I have used some Thief Gold original missions and all available anti-aliasing settings (2x, 4x, 6x and (I think) 8x). No problems with readables (books, scrolls) or with the in-game menu (pressing ESC always works).
An ancient ATI Xpert 2000 pro (a cheap 64-bit version of the Rage 128 Pro) with the latest drivers and Windows 98 SE as well as the same Thief Gold version can only activate or deactivate AA (possibly 2x, but I am not certain). No problems here either.
So, at least older ATI cards with older drivers are unaffected by this issue.
Since I am unable to successfully install Catalyst 8 drivers with Catalyst Control center (and without, I cannot change or even see the card's AA settings), I cannot verify whether the issue would appear with newer drivers. On the other hand, these old cards have no DirectX 10 hardware anyway.
bikerdude on 3/4/2009 at 19:49
Quote Posted by baeuchlein
So, at least older ATI cards with older drivers are unaffected by this issue.
Since I am unable to successfully install Catalyst 8 drivers
with Catalyst Control center (and without, I cannot change or even see the card's AA settings), I cannot verify whether the issue would appear with newer drivers. On the other hand, these old cards have no DirectX 10 hardware anyway.
I still need to test this on my Hd3850 and come back here with the results...
bikerdude on 7/5/2009 at 22:13
Can admins please ad this to the FAQ as it effect all DX10 nVidia cards.
bikerdude on 25/5/2011 at 22:37
Sorry to dig up an old thread, but as things hardware wise have moved on since I started this thread I though I would post an updated result etc.
Running DDfix 1.5.11, W7x64 and a GTX480, couldn't force AA to work in T2, so doing a google search came up with this thread so then did an uninstall/reinstall of the driver and that fixed it.
I found best settings were 4x or higher AA, but crucialy 0xAS (AnisoTropic filtering). I dont understand why, but T2/DDfix1.5.11 dosent like any AS. I think is because its enabled in the ddfix.ini file already, if you do have any AS enabled you get that annoyimng sparkle effect on distant objection/surfaces.
jermi on 26/5/2011 at 11:50
Dark uses textures which contain several lightmaps, and often empty space between the lightmaps. The empty space is uninitialized, so the lightmap edges often touch uninitialized data, which shows up when the GPU multisamples the texture near polygon edges. The result is brightly colored artifacts at the edges of polygons viewed at a shallow angle. This is why ddfix doesn't use anisotropic filtering for the second (lightmap) texture stage. But if aniso is forced in the driver, it gets forced to the second texture stage as well.
bikerdude on 26/5/2011 at 14:52
ah thanks for that Jermi. On a side note T2 looks better with AS off in-driver, as it add's a bit of distance depth due to the distance texture beinbg blurred ever so slightly.
jermi on 27/5/2011 at 14:20
That reminds me that I never added mipmap bias to ddfix. Oh well.
But anyway, the best texture filtering quality with ddfix can be achieved when it's "not set" or "application preference" (or whatever) in the driver settings. Anisotropic shouldn't be forced on or off. ddfix needs to be able to set anisotropic filtering for the first stage, and trilinear for the second stage.
