T2: So, fog is broken again. *sigh* - by EvaUnit02
falco216 on 27/9/2009 at 14:12
Fog is no longer working for me on 9.9, odd, it was working perfectly earlier, trying other drivers now...
I think I had 9.8 installed before but I had installed 9.9 today, seeing as how fog still worked while using 9.9 I assumed it worked in both versions.
I completely uninstalled 9.9 so I could install the 9.8's, I hadn't yet installed them, so I am currently using the default Microsoft drivers. Fogs works, who knew?
Fog works in the 9.8 drivers for me.
I don't have ATT installed and all CCC settings are set to default other than my overclock.
Edit: Wow, I specifically remember trying to get the fog to turn off by going to the menu AND reading a book as well as using the scouting orb, fog stayed perfect, after fiddling with drivers I have the same problem as you, I am now completely confused.
mudi on 27/9/2009 at 19:52
If you can get the fog to stay after entering a menu a lot of us with HD**** cards would really appreciate it :)
Dainem on 27/9/2009 at 20:37
On Win XP fog is fine. (I´m using Ati HD 4850 and Catalyst 8.12; I can´t use newer versions because Thief crashes completely.) After I was forced to install Vista I now get the same strange bug like the others.
@ falco: please try again to get perfect fog. If this does not work I´m not more interested in Windows 7 and will go back to XP. (Playing Broken Triad with no fog is not funny)
Dainem
bikerdude on 27/9/2009 at 22:31
Quote Posted by falco216
so I am currently using the default Microsoft drivers. Fogs works, who knew?
thats down to the fact the driver is an older version, at a guess id say 9.4
poroshin on 29/9/2009 at 18:02
Isn't it strange that we have such talented people that came up with the DDFix, and especially the Widescreen patch - something many swore was just not possible - and other tweaks, yet getting fog for T2 seems like it just can't be fixed.
MoroseTroll on 30/9/2009 at 06:38
poroshin: There is no absolute shield (hacks&patches) can be created against the weapon (drivers) that constantly changes itself. All we can is to keep using old proven drivers or to manifest our displease of new quirky ones to a GPU vendors. What about the new petition?
bikerdude on 30/9/2009 at 18:33
Quote Posted by poroshin
and other tweaks, yet getting fog for T2 seems like it just can't be fixed.
it could, but Timeslip dosent have the time anymore to work on DDfix. And the original sources files for the earlier version has been lost leaving on the current version 1.3.11
SiO2 on 11/10/2009 at 11:45
Having grabbed Timeslip's source and recompiled DDFix so I could do some investigation, I'm not too surprised that we're having such problems with fog. Thief2 passes 4D homogenous coordinates to a D3D3 device which I'm guessing is probably outside IHV's support path. These days, developers usually pass 3D object-space vertices to a vertex shader together with the transformation matrices needed to project the verts into clip-space. T2 effectively does the work of the vertex shader itself. Add the fact that modern GPU's don't have any fixed-function hardware anymore and convert fixed-function device state into runtime-generated programmable shaders and you start to wonder what some drivers actually do when confronted with a path as comparatively unusual as "old" Direct3D3 API calls with 4D homogenous vertices. Purely a guess, but I would not be surprised if the fog switch was being ignored by some drivers.
Hmmm. I guess it would be possible to change Timeslip's wrapper so it creates a D3D9 device internally, translates the D3D3 calls and uses programmable shaders - we could then do whatever we want in the pixel shader. The vertex shader would just pass-thru the data to the pixel shader and after the homogenous divide we would have a depth value we could use to apply our own fog calculation to the colour output from the pixel shader. A lot work just to fix fog!
BTW If we ever get the official T2 source code one of the primary areas of improvement would be to offload the transformations to the GPU.
bikerdude on 11/10/2009 at 13:12
Hi Sio2
You sound like the man/woman for the job, we/I would be very happy if you could take point on this, as the majority of us dont have anywhere near the level of skill/knowledge that you have just demonstrated.
This would also mean that I/others would be free to choose what gfx card to upgrade to and not be tied to ATi for fog support...
I would suggest either continuing this in the guild section or in my (
http://www.ttlg.com/forums/showthread.php?t=117616) thread
jtr7 on 11/10/2009 at 23:24
Quote Posted by SiO2
Having grabbed Timeslip's source and recompiled DDFix so I could do some investigation, I'm not too surprised that we're having such problems with fog. Thief2 passes 4D homogenous coordinates to a D3D3 device which I'm guessing is probably outside IHV's support path. These days, developers usually pass 3D object-space vertices to a vertex shader together with the transformation matrices needed to project the verts into clip-space. T2 effectively does the work of the vertex shader itself. Add the fact that modern GPU's don't have any fixed-function hardware anymore and convert fixed-function device state into runtime-generated programmable shaders and you start to wonder what some drivers actually do when confronted with a path as comparatively unusual as "old" Direct3D3 API calls with 4D homogenous vertices. Purely a guess, but I would not be surprised if the fog switch was being ignored by some drivers.
Hmmm. I guess it would be possible to change Timeslip's wrapper so it creates a D3D9 device internally, translates the D3D3 calls and uses programmable shaders - we could then do whatever we want in the pixel shader. The vertex shader would just pass-thru the data to the pixel shader and after the homogenous divide we would have a depth value we could use to apply our own fog calculation to the colour output from the pixel shader. A lot work just to fix fog!
BTW If we ever get the official T2 source code one of the primary areas of improvement would be to offload the transformations to the GPU.
:D:thumb:
