• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Doom 3 shader mod for ++ performance

Status
Not open for further replies.
ThisMonsterLives said:
That's nice. Had a GT been available when I got my X800 2 months ago I would have considered one. That's one of the benefits of not playing favorites, having choices.

There was no games that were grabbing my immediate attention 2 months ago. This month we have D3. Next month we will HOPEFULLY have HL2. Another month or two after that we should HOPEFULLY get Vampire and Battle for Middle-Earth. Some people might say there was FC 2 months ago and I'll just point out that FC was released before the X800s. Summertime is typically the death-toll for gamers.

Edit: I found this thread over at Beyond3D. Maybe the 'snow' effect some people have seen with the hack might not even be from the hack. Regardless, it's an interesting read.

http://www.beyond3d.com/forum/viewtopic.php?t=14787
 
KakimotoR said:
Original Poster DaveBauMann

Utilisings normalisation via math will decrease your your texture fill-rate, which should increase the performance under AF situations (expectially if texture sampling and math instructions are interleaved effectively by the hardware).

You do know that was in response to Demirug's method right =/?
 
Tried it on my 9800 NoN Pro, Specs in Sig

1024x768 High
Went from 43.9 to 46.1 after this hack
Not to bad...
 
people on that rage3d forum are like OMG I GET 20 MORE FPS>> then they list thet settings and have shadows turned off :gay: why the hell would you turn off shadows in doom3. the game isnt even worth playing w/o shadows
 
Not to bust anyone's chops but in essence wouldn't this be similar to changing from high quality to medium quality and gaining (wow) a performance boost because you change the quality settings? I mean, my 9700 pro runs d3 just peachy, but I don't see what the big deal is about this. If they say you can barely tell the difference between high and medium, and now you can barely tell the difference between the original and this hack, it just seems like a fair comparison where you get negligible quality loss for a little fps gain.
 
depends if the shadows are on you don't get much of an increase, but there is a bit of drop with the specular level, and that sometimes will irrate a player, "wish I could have the same IQ with the same performance thing" Alot of people won't even notice it, some will. Its not a big deal.
 
god been on these boards way too long today lol, got to get out and play some golf! cya guys around!
 
Just FYI this patch sacrifices IQ for performance, if you don't want to lose IQ I wouldn't apply the patch. It also causes artifacting in many places.

from the B3D thread:
Chalnoth said:
Chalnoth wrote:
After some looking, I don't think it's mathematically equivalent, Humus (I found a place similar to the one kyleb posted, and compared how different specular exponents affected the wall and the chaingun...the wall with the texture lookup always looked dull, with the chaingun shiny...but if I made the wall dull with a low specular exponent, then the chaingun looked way too bright).

There may be a different per-material or per-object specular lookup table.

kyleb said:
nasty bright white areas with humus' code:

danzg7 said:
here's another example of this tweak not working correctly

micron said:
got some pretty serious tearing while using this tweak. Vsync helped a little, but it was still pretty bad.

Smurfie said:
This could the start of a bad precedence. ATI employee comes in with IQ lossy tweak. Nvidia might see ATI doing IQ lossy tweak and introduce their own IQ lossy tweak.

Then IQ lossy tweak wars start, and we get flat shaded polygons in the end.

I just hope ATI remembers their mantra on IQ.

CrunchTime said:
well upon further inspection, the shaders with this patch look terrible.

Almost all of the doors have this ugly aliased white look to it, basically that is why we get a FPS increase, in my case, 8 fps...

The IQ is dropped dramatically which is why JC knows what hes doing

Setting your filtering to Bilinear & reducing AF level will likely cause less artifacts than ATI's new shader replacement code, and give you a performance boost.
 
trungracingdev said:
Since you don't own a ati card, you can stay away from this thread. This only benefits ati cards. I find Humus to be a very smart guy. He writes these on his own time and shares it with the gaming community.

NO ONE is forcing anyone to use these codes. Some people are actually benefiting from this and thats good enough.

By the same token you can stay away from my posts...nobody is forcing you to respond...shithead

Whether I own an ATi card or not has nothing to do with the matter...the problem is this guy stirred up a whole bunch of shit and has a whole bunch of people talking about NOTHING...this hack is just that...a HACK...I appreciate that the guy wants to help out ATi users and that's great...but I think he should leave the matter in the quite capable hands of the ATi driver team and id software instead of posting IQ reducing hacks and claiming they're the ATi/Doom3 performance holy grail...

I say again, irresponsible and unprofessional...this is now posted on every major hardware website I know of...and for what? 3-4fps and some artifacts? thanks Humus, just what ATi needed, more image quality issues...
 
^eMpTy^ said:
but I think he should leave the matter in the quite capable hands of the ATi driver team and id software instead of posting IQ reducing hacks and claiming they're the ATi/Doom3 performance holy grail...

I say again, irresponsible and unprofessional...this is now posted on every major hardware website I know of...and for what? 3-4fps and some artifacts? thanks Humus, just what ATi needed, more image quality issues...

i concur, leave it to the people who know (well i'm not saying humus doesn't know anything, he sounds like a smart-guy however has he really benefitted the community?)
 
I have to agree with empty for the most part. also If we let all the ati peoples alone in this thread there would be no one to tell you the draw backs. thats the wonderful thing about fan boys. they are always aware of their cards and the cons to the other cards and this helps differentiate claims vs actuality.

hence theres always 2 sides to every story/hack.
 
Hey empty, maybe humus just found his tweak.. "interesting"? :rolleyes: If some people with ati cards are getting appreciable gains, without a percievable iq loss (and that's a completely relative opinion btw) then I think it's an awesome find, and something that's definately worth sharing. Why don't you just calm down, you're blowing things way out of proportion.
 
^eMpTy^ said:
By the same token you can stay away from my posts...nobody is forcing you to respond...shithead

Whether I own an ATi card or not has nothing to do with the matter...the problem is this guy stirred up a whole bunch of shit and has a whole bunch of people talking about NOTHING...this hack is just that...a HACK...I appreciate that the guy wants to help out ATi users and that's great...but I think he should leave the matter in the quite capable hands of the ATi driver team and id software instead of posting IQ reducing hacks and claiming they're the ATi/Doom3 performance holy grail...

I say again, irresponsible and unprofessional...this is now posted on every major hardware website I know of...and for what? 3-4fps and some artifacts? thanks Humus, just what ATi needed, more image quality issues...

feel free to discuss the merit of the tweaks, hacks, or whatever you want to call them. however, namecalling is not allowed, and will get you banned from the forum
 
GeniusInABottle said:
Hey empty, maybe humus just found his tweak.. "interesting"? :rolleyes: If some people with ati cards are getting appreciable gains, without a percievable iq loss (and that's a completely relative opinion btw) then I think it's an awesome find, and something that's definately worth sharing. Why don't you just calm down, you're blowing things way out of proportion.

there is an iq loss in almost every case, did you not read the posts on beyond3d?
 
I did the patch on my 9800pro system and it made everything so dark I couldnt even see anything lol anyways reverted back to original.
 
Xenozx said:
there is an iq loss in almost every case, did you not read the posts on beyond3d?

Yes, I did read the posts. Did you read mine? I do believe I clearly stated some.
 
If you've ever used an ATI card you know that DX7/8 shadows/fog are just plain slow. Bah, forget 30 percent, I'd say all the way up to 50 percent speed difference by just turning off the shadows or using simple "blob" shadows. Don't know about OpenGL, but I would not assume it to be any different.

ATi explained why they did not want a per-pipeline shadow ability in DX7/8 a long time ago, they said it would obscure or abstract too much of the picture and would be too dark and confusing to the end-user (sound farmiliar?). In most cases there would be one overriding main light anyways, the sun and it would never cast a perfectly shaped or perfectly black shadow. It was their same argument for not doing per-pipeline fog, it is an "obscuring" technology, not an visually enhancing technology.

This does not affect DX9 however, as it can use MRT's with full shadows with no real peformance hit. I'm not sure, but there may be something in OpenGL 2.0 that is similar.
 
any IQ loss is unacceptable especially the way humus presented it. "oh heres a fix for some more FPS" "john carmack sucks i cant believe he missed this" "by the way your gonna notice a drop in IQ and some rainbow things"

play the game the way the creater intended it to be played. dont go off and lower the IQ in sake of getting more FPS. that is called HACK plain and simple. im sure we could hack the code to give nvidia cards a boost, but that again is a hack and should not be done.


hack does not = tweak...

im sure john carmack knew what he was doing, and for something that simple i highly doubt it would have been overlooked unless there was some unwanted drawback for it.
 
I've seen the two side by side now. I'm going to go out on a limb and say the new hack looks slightly more appealing to me than the default :confused:

Things are a little "brighter", "glossier" and tiny bit easier to see. Its like everything just stands out a little bit more. Metal edges shine like metal edges are supposed to (even as dark as it is)
 
ZenOps said:
If you've ever used an ATI card you know that DX7/8 shadows/fog are just plain slow. Bah, forget 30 percent, I'd say all the way up to 50 percent speed difference by just turning off the shadows or using simple "blob" shadows. Don't know about OpenGL, but I would not assume it to be any different.

ATi explained why they did not want a per-pipeline shadow ability in DX7/8 a long time ago, they said it would obscure or abstract too much of the picture and would be too dark and confusing to the end-user (sound farmiliar?). In most cases there would be one overriding main light anyways, the sun and it would never cast a perfectly shaped or perfectly black shadow. It was their same argument for not doing per-pipeline fog, it is an "obscuring" technology, not an visually enhancing technology.

This does not affect DX9 however, as it can use MRT's with full shadows with no real peformance hit. I'm not sure, but there may be something in OpenGL 2.0 that is similar.

Very true, you are right, its closer to 50% increase of polys that need to be rendered. pretty much when a shadow is casted the polygons double or more then double, but when frame rates are low, you really wont' see the 50% decrease in the frame rates, just the way hardware is set up as they are stressed more they work more efficiently.
 
I believe Doom 3 is OpenGL 1.5, it may be patched to 2.0 sometime in the future (Hence, I'm wondering if something can be done about the speed of the shadows)
 
Wow this is quite a thread. Well i just wanted to give my two cents to it.

I had many High End systems in the past. There was the 3.9GHz Vapochill, The dual 3.2GHz watercooled Dual XEON, and the lovely 2.6C @ 3.8 :) I spend over 3 grand on my last computer .....and you know what ?!? It wasnt worth anything. Because 1 month later came a new game around, and a new graphics card. And man was i mad cause my recently bought 500€ video card was 10FPS slower ..than then the latest one. :mad:

As far as i know its like this.

Doom3 = nvidia is faster
Far Cry = ATI is faster
UT2004 = ATI is faster
UT2003 = ATI is faster

and there are many many more games that run faster on ATI than on nvidia. So what the hell !! I do not care if nVidia is currently faster. Ill just wait another 6 month to play the game on the new ATI card thats out by then. Why would i spend 400 bucks on the nvida card now, only because its faster in Doom3.

And by the way, to all who call this a "cheat" or a "hack" ...so what. I remeber nvidia cheating like crazy on all of 3DMark. The even cheatet back in the Quake 3 Arena days :) LOL go on and wear your pride up high. We dont care, we still beat ya on the rest of the games.

I had the GeForce 1, the GeForce 2 Utra, The GeForce 3 TI500, the GeForce 4 TI4600. Then i had the 9700 Pro, then the 9800 Pro and during that time nVidia had absolutly nothing to stick against it. The entire 5800 and the entire 5950 range is BS compared to ATI.

just my opinion

Greetings from Germany
 
well ogl 1.5 to 2.0 there isn't any change with stencil shadow rendering. still takes 2 passes. Really only benfits soft stencil shadows where branching will help it.
 
fugu said:
feel free to discuss the merit of the tweaks, hacks, or whatever you want to call them. however, namecalling is not allowed, and will get you banned from the forum

sorry...one slipped out...won't happen again...
 
Mr.Magneplanar said:
Wow this is quite a thread. Well i just wanted to give my two cents to it.

I had many High End systems in the past. There was the 3.9GHz Vapochill, The dual 3.2GHz watercooled Dual XEON, and the lovely 2.6C @ 3.8 :) I spend over 3 grand on my last computer .....and you know what ?!? It wasnt worth anything. Because 1 month later came a new game around, and a new graphics card. And man was i mad cause my recently bought 500€ video card was 10FPS slower ..than then the latest one. :mad:

As far as i know its like this.

Doom3 = nvidia is faster
Far Cry = ATI is faster
UT2004 = ATI is faster
UT2003 = ATI is faster

and there are many many more games that run faster on ATI than on nvidia. So what the hell !! I do not care if nVidia is currently faster. Ill just wait another 6 month to play the game on the new ATI card thats out by then. Why would i spend 400 bucks on the nvida card now, only because its faster in Doom3.

And by the way, to all who call this a "cheat" or a "hack" ...so what. I remeber nvidia cheating like crazy on all of 3DMark. The even cheatet back in the Quake 3 Arena days :) LOL go on and wear your pride up high. We dont care, we still beat ya on the rest of the games.

I had the GeForce 1, the GeForce 2 Utra, The GeForce 3 TI500, the GeForce 4 TI4600. Then i had the 9700 Pro, then the 9800 Pro and during that time nVidia had absolutly nothing to stick against it. The entire 5800 and the entire 5950 range is BS compared to ATI.

just my opinion

Greetings from Germany

You are right 100 percent thats is changing and the next line too is going to be very tight, oh btw ya might want to check out the newer drivers for nV's cards cause ATi isn't faster anymore in Dx, pretty much equal, also when the 62 series drivers are ready for official release, and when ultra shadow is finally activated with the drivers, just more speed for nV
 
GeniusInABottle said:
maybe humus just found his tweak.. "interesting"?
:p it looks like humus jumped the gun a little by thinking he fixed the performance problem ("But this seems to be more or less plain Blinn shading"). His assumption was wrong and the only problem I have is that he hasn't admitted it yet. His original post wasn't exactly done with any sense of neutrality... it was very defensive and implied Carmack messed up.

Like I said a page or two back, who cares if someone want to run it? It's not a big deal and it will only work for the single player game (not pure servers). It affects how the game is rendered, so no one is going to use it to compare to untweaked peformance. I've run it and it doesn't look too bad, but it's definitely inferior to the way the default shader looks.
 
Mr.Magneplanar said:
Wow this is quite a thread. Well i just wanted to give my two cents to it.

I had many High End systems in the past. There was the 3.9GHz Vapochill, The dual 3.2GHz watercooled Dual XEON, and the lovely 2.6C @ 3.8 :) I spend over 3 grand on my last computer .....and you know what ?!? It wasnt worth anything. Because 1 month later came a new game around, and a new graphics card. And man was i mad cause my recently bought 500€ video card was 10FPS slower ..than then the latest one. :mad:

As far as i know its like this.

Doom3 = nvidia is faster
Far Cry = ATI is faster
UT2004 = ATI is faster
UT2003 = ATI is faster

and there are many many more games that run faster on ATI than on nvidia. So what the hell !! I do not care if nVidia is currently faster. Ill just wait another 6 month to play the game on the new ATI card thats out by then. Why would i spend 400 bucks on the nvida card now, only because its faster in Doom3.

And by the way, to all who call this a "cheat" or a "hack" ...so what. I remeber nvidia cheating like crazy on all of 3DMark. The even cheatet back in the Quake 3 Arena days :) LOL go on and wear your pride up high. We dont care, we still beat ya on the rest of the games.

I had the GeForce 1, the GeForce 2 Utra, The GeForce 3 TI500, the GeForce 4 TI4600. Then i had the 9700 Pro, then the 9800 Pro and during that time nVidia had absolutly nothing to stick against it. The entire 5800 and the entire 5950 range is BS compared to ATI.

just my opinion

Greetings from Germany

People are buying nvidia mostly because the XTPE is completely unavailable...it can win as many benchmarks as you like but if it's not available it doesn't really matter...so it's not just a Doom3 thing...it's a 12 pipe Pro vs a 16 pipe GT thing...

And I was just as annoyed by every other cheat hack or whatever...when nvidia cheated in 3dmark I was furious...when ATi did it with Quake 3 I called bullshit once more...and when people hack doom3 and say it's the answer to your X800 performance problems I get equally as annoyed...

This has nothing to do with ATi vs Nvidia...this is about one person's irresponsibility...
 
Well. If the nVidia cards will be faster soon. Then ill buy an nVidia card. If the ATI one is better, than ill buy the ATI one. I dont care who i give the 500€ too. But ill properbly give the 1Grand to Clearaudio and get myself a turntabel .... Muhahah :p
 
pxc said:
:p it looks like humus jumped the gun a little by thinking he fixed the performance problem ("But this seems to be more or less plain Blinn shading"). His assumption was wrong and the only problem I have is that he hasn't admitted it yet. His original post wasn't exactly done with any sense of neutrality... it was very defensive and implied Carmack messed up.

Like I said a page or two back, who cares if someone want to run it? It's not a big deal and it will only work for the single player game (not pure servers). It affects how the game is rendered, so no one is going to use it to compare to untweaked peformance. I've run it and it doesn't look too bad, but it's definitely inferior to the way the default shader looks.

ditto, I've used similiar optimizations myself just to show off a bit :) where I could show normal mapping has no speed loss over environmental bump mapping thats exactly how its done. It works great for some types of lights but by doing so, you loose some of the depth of the normal map.
 
Xenozx said:
any IQ loss is unacceptable especially the way humus presented it. "oh heres a fix for some more FPS" "john carmack sucks i cant believe he missed this" "by the way your gonna notice a drop in IQ and some rainbow things"

play the game the way the creater intended it to be played. dont go off and lower the IQ in sake of getting more FPS. that is called HACK plain and simple. im sure we could hack the code to give nvidia cards a boost, but that again is a hack and should not be done.


hack does not = tweak...

im sure john carmack knew what he was doing, and for something that simple i highly doubt it would have been overlooked unless there was some unwanted drawback for it.

So I suppose we should all be running in Ultra Quality 8xAA 16xAF at 2048x1536? Anything less and IQ has been reduced from it's maxim. I realize that I'm using "unrealistic" extremes, but I'm proving a point. The truth is, as I was trying to say, that IQ is relative to the person judging it. Very simply, beauty is in the eye of the beholder. Those not running in Ultra Quality have lost image fidelity due to compression, does this mean that the image quality they're playing at is unacceptable? Hardly. What humus did, to my understanding, is replace a shader with what should be a mathematical equivalent. It's not perfect just yet, but he never stated it was. I for one am appreciative of stuff like this, as opposed to condemning.
 
Mr.Magneplanar said:
Well. If the nVidia cards will be faster soon. Then ill buy an nVidia card. If the ATI one is better, than ill buy the ATI one. I dont care who i give the 500€ too. But ill properbly give the 1Grand to Clearaudio and get myself a turntabel .... Muhahah :p

LOL cool, it should be a pretty tight fight with the nv50 and r520 should be so nice fireworks.
 
quto thingy

This has nothing to do with ATi vs Nvidia...this is about one person's irresponsibility...

end quote thingy


Well i think that anyone who reads this in a Free Forum has to take it with care. Its the same as if i say. "Put ice cubes over your AMD CPU it will go 40% faster"..... if ya do it and the result is bad, then why should i be blamed ?!?!

If he is wrong, than just prove him wrong. But dont bash on the poor guy. :)
 
GeniusInABottle said:
So I suppose we should all be running in Ultra Quality 8xAA 16xAF at 2048x1536? Anything less and IQ has been reduced from it's maxim. I realize that I'm using "unrealistic" extremes, but I'm proving a point. The truth is, as I was trying to say, that IQ is relative to the person judging it. Very simply, beauty is in the eye of the beholder. Those not running in Ultra Quality have lost image fidelity due to compression, does this mean that the image quality they're playing at is unacceptable? Hardly. What humus did, to my understanding, is replace a shader with what should be a mathematical equivalent. It's not perfect just yet, but he never stated it was. I for one am appreciate of stuff like this, as opposed to condemning.

You're comparing apples and oranges...there's a big difference between switching between pre-determined game settings and opening up a pak file and screwing with shader code...

It's not that I don't appreciate his efforts...I just think the way he presented it was BS...his post read like a piece of flamebait, he should hold himself to a higher standard...
 
rancor said:
ditto, I've used similiar optimizations myself just to show off a bit :) where I could show normal mapping has no speed loss over environmental bump mapping thats exactly how its done. It works great for some types of lights but by doing so, you loose some of the depth of the normal map.
Did you see the plasma light tweak? It's supposed to hurt framerate but looks :cool:.
 
for those people who want a little extra performance i think its worth while, but again it should be labled a hack and not "something john carmack did wrong" its not gonna increase performance from no where. its sacrificing something the question at this point is what.

again can you tell me why john carmack did not use this shader path out of the box?
 
Mr.Magneplanar said:
quto thingy

This has nothing to do with ATi vs Nvidia...this is about one person's irresponsibility...

end quote thingy


Well i think that anyone who reads this in a Free Forum has to take it with care. Its the same as if i say. "Put ice cubes over your AMD CPU it will go 40% faster"..... if ya do it and the result is bad, then why should i be blamed ?!?!

If he is wrong, than just prove him wrong. But dont bash on the poor guy. :)

I don't have to prove him wrong...he's been proven wrong by everyone who's tried it and posted their results and complaints in that thread...and he's not just some random guy...he's a senior member at b3d and well known for his demos...people on that forum look up to him, if he posts stuff like this they are going to take it at face value...that's why it made it to the front page of [H] as well as xbit and several others...
 
Status
Not open for further replies.
Back
Top