Visual differences between Pixelshader 2.0(b) and 3.0...

Thats just it, 2.0b might need to do two passes to do hyper realistic water than 3.0 can do in one pass. It will look the same though.

Realistic fire might take one pass in SM4.0 that might take two in SM3.0, it never ends... But I think its heading in the wrong direction, there are only less than a handful of visual effects that actually need a shader model that is that *deep*.

Its sort of like the Intel Prescott pipeline, it *can* do more, but the long multi-instruction pipes start to become cumbersome for simple tasks. Thats why there are SSE specialized instructions. I'm thinking Nvidia and Ati should be concentrating on specialized particle effects rather then generalizing larger and longer pipes that will mostly go unused.
 
ZenOps said:
Thats just it, 2.0b might need to do two passes to do hyper realistic water than 3.0 can do in one pass. It will look the same though.

Realistic fire might take one pass in SM4.0 that might take two in SM3.0, it never ends... But I think its heading in the wrong direction, there are only less than a handful of visual effects that actually need a shader model that is that *deep*.

Its sort of like the Intel Prescott pipeline, it *can* do more, but the long multi-instruction pipes start to become cumbersome for simple tasks. Thats why there are SSE specialized instructions. I'm thinking Nvidia and Ati should be concentrating on specialized particle effects rather then generalizing larger and longer pipes that will mostly go unused.
Click to expand...

Right. But it seemed you were bakcing up terras (erroneous) statements that there is a visual quality difference between 2.0b and 3.0, which there isn't, unless they aren't coded for correctly within the game engine.
 
I have not read this entire thread, I just wanted to chime in though.

A pixel shader effect written with PS 1.1, 2.0/b, 3.0 will look exactly the same if that is all you do is code the same effect. The difference will be performance, with 1.1 and 2.0 you may have to do more passes to achieve the effect since the code length may not fit within the limits. With 3.0 it may be able to do it all in one pass saving performance. It is then with this performance increase that a game developer could offset the difference by coding even more effects into that pixel making it look better than 1.1 and 2.0. So you'd then get the same perf as 2.0 but have it visually better, or they could opt to leave the effect the same and go for the perf increase.
 
Brent_Justice said:
I have not read this entire thread, I just wanted to chime in though.

A pixel shader effect written with PS 1.1, 2.0/b, 3.0 will look exactly the same if that is all you do is code the same effect. The difference will be performance, with 1.1 and 2.0 you may have to do more passes to achieve the effect since the code length may not fit within the limits. With 3.0 it may be able to do it all in one pass saving performance. It is then with this performance increase that a game developer could offset the difference by coding even more effects into that pixel making it look better than 1.1 and 2.0. So you'd then get the same perf as 2.0 but have it visually better, or they could opt to leave the effect the same and go for the perf increase.
Click to expand...
Yes. I pretty much said it a couple posts back, just with a lot less words.

sm3.0 offers NO visual quality enhancements over sm2.0b. Period.
Only PERFORMANCE increases are available in sm3.0.
Click to expand...
 
Terra said:
I just fell over this little "show and tell" display regarding the visual differences between PixelShader version 2 and 3 in real world games:

PixelShader Version 2 And 3 Compared

I know it's from Tom's, but dosn't alter with the fact that games are starting to show differences between the 2 versions.

Terra - And thus making the SM/PS 3.0 cards a more futurepsoof invesment if you like IQ in games...
Click to expand...


Error404 - Document Not Found!
Sorry! The Requested Page was not Found!!

Dear Visitor. The page you requested couldn't be found :( - If you followed a link from another Website please inform their Webmaster. If you happen to get this message while browsing Tom's Hardware Guide please inform the Webmaster of Tom's Hardware Guide.

Please check the URL in the address field of your browser. All pages end with .html, maybe it is .htm. You might want to use the Search Engine to find what you are looking for.

You can also browse Tom's Guides (see box below). They list articles according to their topic. Or you might want to check the Historical Archive. Here you can find all articles in a chronological order. The Keyword Index provides an alphabetical listing of products, product groups and companies.

RIGHT.....
 
And by the time the differences really become apparent, the 2.0b cards will be to slow to be any good anyways. Good trollery though Terra. Thought you had given up on posting flame bait material.
 
eno-on said:
Right. But it seemed you were bakcing up terras (erroneous) statements that there is a visual quality difference between 2.0b and 3.0, which there isn't, unless they aren't coded for correctly within the game engine.
Click to expand...

same could be said for shader model 1.4 too ;) his isn't really erroneous if the programmers want to take it that far. Instead of writting one or two shaders for an effect, they will have to write possibly 5 or 6 different shaders ;). Not the easiest of tasks. Plus they would be running too slow for older hardware to run anyways.
 
razor1 said:
same could be said for shader model 1.4 too ;) his isn't really erroneous if the programmers want to take it that far. Instead of writting one or two shaders for an effect, they will have to write possibly 5 or 6 different shaders ;). Not the easiest of tasks. Plus they would be running too slow for older hardware to run anyways.
Click to expand...
Thas not the point.
The point is that there are NO IQ differences if both are programmed correctly.
 
Terra said:
Fact is no matter how much you try and blame the developers, the moon or my trained spelling penguin games are starting to show a differnces between SM2.0 and SM3.0...

Terra - Like it or not...
Click to expand...

I quote myself, because no matter how much you flame, there IS a visual difference in eg. AOE between SM2.0 and SM3.0...

Terra - No you can flame on..the pics don't lie :D
 
Terra said:
I quote myself, because no matter how much you flame, there IS a visual difference in eg. AOE between SM2.0 and SM3.0...

Terra - No you can flame on..the pics don't lie :D
Click to expand...
Get this through your head :
There is no difference in IQ between 2b and 3. Only in developers usage of them. Period. You can't argue against a FACT like this without coming across as totally clueless.
 
He wont ever admit hes wrong, eno. No matter how much evidence is stacked up. You'll notice this common trend in his topics. Its better to just ignore him and his rantings.
 
Does anyone else think the "Low" quality water looks better than the "Medium" quality water?

Other than there being no shadow in the low quality.
 
eno-on said:
Get this through your head :
There is no difference in IQ between 2b and 3. Only in developers usage of them. Period. You can't argue against a FACT like this without coming across as totally clueless.
Click to expand...

The fact still remains that there ARE visual differences showing today i games between SM2.0 and SM3.0!
Blame it on the developers, but the visual differences are still there, so my statement is still true.
It dosn't matter if you can the do the same (with more passes ) in SM2.0 as in SM.30, there is STILL a visual difference, like it or not! :p

Terracide - Flame on :D
 
fallguy said:
He wont ever admit hes wrong, eno. No matter how much evidence is stacked up. You'll notice this common trend in his topics. Its better to just ignore him and his rantings.
Click to expand...

Are you saying to me that you se NO differences between the SM2.0(b) and the SM3.0 pictures from AOE? :rolleyes:

Terra - I'm not the one ranting now :p
 
Terra said:
Are you saying to me that you se NO differences between the SM2.0(b) and the SM3.0 pictures from AOE? :rolleyes:

Terra - I'm not the one ranting now :p
Click to expand...


Right.... Let's try this one more time....




















The differences you see in AOE are not from sm2.0b's inability to do those effects, only the programmers unwillingness to program those effects for 2.0b. Comprende? Begrijp? Omprenez? Verstehen Sie? καταλάβετε? Do you understand the words that are comin' out of my mouth?

If not, then I'll get to use the ignore function for the first time :)
 
Terra said:
The fact still remains that there ARE visual differences showing today i games between SM2.0 and SM3.0!
Blame it on the developers, but the visual differences are still there, so my statement is still true.
It dosn't matter if you can the do the same (with more passes ) in SM2.0 as in SM.30, there is STILL a visual difference, like it or not! :p

Terracide - Flame on :D
Click to expand...


You argument is as valid as the artists from Doom3 and the artists of Stalker, Prey, HL2, or whatever not looking the same or worse than (insert game name) here :rolleyes:
 
eno-on said:
If not, then I'll get to use the ignore function for the first time :)
Click to expand...

Actually right now would be a perfect time for you to try it out.
 
eno-on said:
Thas not the point.
The point is that there are NO IQ differences if both are programmed correctly.
Click to expand...


That is the point, parrallex bump mapping can be done on sm 1.4 hardware. Did Crytek program thier parrallex bump mapping for that? How about Monolith for Fear?

It can be done, but is it worth it? Not really. Its too slow.

Then you have normal mapping, it can be done on dx 6 and 7 hardware. Was it done with Crytek? How about do Doom?

This is the same argument, many things can be done on sm 2.0, 2.0b that can be done with sm 3.0 but are they playable.
 
ivzk said:
Actually right now would be a perfect time for you to try it out.
Click to expand...

It is glaringly obvious that no matter what you say, Terra will still claim a visual difference using that link as proof, arguing with him/her is a futile effort.
 
This is like trying to argue that S.M 1.1 is nearly as good as s.m 2.0.

Really.. why do any of you people care... ? Shouldn't you be playing games or something instead of argueing about S.M .. ?
 
dagon11985 said:
This is like trying to argue that S.M 1.1 is nearly as good as s.m 2.0.

Really.. why do any of you people care... ? Shouldn't you be playing games or something instead of argueing about S.M .. ?
Click to expand...


No time for gaming :D
 
razor1 said:
That is the point, parrallex bump mapping can be done on sm 1.4 hardware. Did Crytek program thier parrallex bump mapping for that? How about Monolith for Fear?

It can be done, but is it worth it? Not really. Its too slow.

Then you have normal mapping, it can be done on dx 6 and 7 hardware. Was it done with Crytek? How about do Doom?

This is the same argument, many things can be done on sm 2.0, 2.0b that can be done with sm 3.0 but are they playable.
Click to expand...
No one is arguing that there are (or can be) performance differences. That's a given.
Trerra is stating that there are IQ differences between them. Which there are not.
 
razor1 said:
This is the same argument, many things can be done on sm 2.0, 2.0b that can be done with sm 3.0 but are they playable.
Click to expand...

The trouble, however, is that the SM 3.0 generation parts that compete with SM 2.0b parts are not quite playable either, or if they are, it is at a low enough resolution that it is at odds with the effort of higher IQ.
 
Pretty much what i've gotten from this thread is that there are performance differences, 1.1 and 2.0 take more passes etc... And that there can be a visual difference if the programmer feels that there is enough performance gain to add better textures or something to that effect. Is this correct?
 
ChronicTrees said:
Pretty much what i've gotten from this thread is that there are performance differences, 1.1 and 2.0 take more passes etc... And that there can be a visual difference if the programmer feels that there is enough performance gain to add better textures or something to that effect. Is this correct?
Click to expand...
Sort of. If the developr decides to program for 2.0b to look the same as 3.0, they will look EXACTLY the same, but there can be (will be) performance differences, with 3.0 being the better performer.
 
eno-on said:
Sort of. If the developr decides to program for 2.0b to look the same as 3.0, they will look EXACTLY the same, but there can be (will be) performance differences, with 3.0 being the better performer.
Click to expand...

Ok, so any visual differences would be a developers call. SM x vs SM x will have no visual differences unless the developer implements something under one path and not another.
 
ChronicTrees said:
Ok, so any visual differences would be a developers call. SM x vs SM x will have no visual differences unless the developer implements something under one path and not another.
Click to expand...
Correct.
 
truffle00 said:
The trouble, however, is that the SM 3.0 generation parts that compete with SM 2.0b parts are not quite playable either, or if they are, it is at a low enough resolution that it is at odds with the effort of higher IQ.
Click to expand...

Sort of. If the developr decides to program for 2.0b to look the same as 3.0, they will look EXACTLY the same, but there can be (will be) performance differences, with 3.0 being the better performer.
Click to expand...

Thats the thing you can have parrellex bump mapping that looks exactly the same with ps 1.4 cards and 2.0 cards, and 3.0 cards, It won't run on 1. 4 cards well at all, 2.0 cards resonably well, but still not that great, and much better with 3.0 cards. This is only because of processing power not that one or the other can't do it. But ATi said the same thing there is no need for sm 3.0 last year cards. But it is definitly usable on the gf 6's. The turn around is the same thing with nV saying that sm 2.0 wasn't needed when they didn't have a sm 2.0 card. There is an IQ difference if a card you can program an effect for can or can not run it in real time.

Ya its great that the G7 demos can run ambient occlusion, but do that in real time.

Same with the HDR demo ATi had for the 9700.

Same with the normal mapping demo they had for the gf 2's.

by real time I mean game situtaion.
 
No, you can't really argue it either way.
They are both capable of the SAME IMAGE.
The performance is the difference.

There is an IQ difference if a card you can program an effect for can or can not run it in real time.
Click to expand...

Technically speaking, their IQ is identicle if programmed to display the same image. Hence, no IQ difference.
No one wants to play at 5 FPS though.
I certainly don't.
I'm not saying 3.0 isn't superior to 2.0b. It clearly is. But to say that the differences in those AOE shots are because 2.0b CAN'T do something is plain wrong. The differences in IQ there are STRICTLY programming, not limitations of 2.0b hardware.
 
eno-on said:
No, you can't really argue it either way.
They are both capable of the SAME IMAGE.
The performance is the difference.



Technically speaking, their IQ is identicle if programmed to display the same image. Hence, no IQ difference.
No one wants to play at 5 FPS though.
I certainly don't.
I'm not saying 3.0 isn't superior to 2.0b. It clearly is. But to say that the differences in those AOE shots are because 2.0b CAN'T do something is plain wrong. The differences in IQ there are STRICTLY programming, not limitations of 2.0b hardware.
Click to expand...


It really depends on how you look at it. I would say they can't do certain effects due to speed reasons. And you would say they can do the effects but just not playable. :p
 
The differences in IQ there are STRICTLY programming, not limitations of 2.0b hardware.
Click to expand...

Right, I guess what I meant to say by arguing both ways on it is that having sm3 vs 2 would be a variable in the Image Quality but it is not directly responsible for any visual differences.
 
razor1 said:
It really depends on how you look at it. I would say they can't do certain effects due to speed reasons. And you would say they can do the effects but just not playable. :p
Click to expand...
Razor, if it can draw it, it can draw it.
If I can walk to my house but it would take me 25 years, well guess what, I can still do it. Would it make sense to? No, but the fact remains that I can, in fact, walk home.
Just as SM 2.0b can render those same effects. Meaning, IQ is identical. Hence, the statement that IQ is different or that it 'can't do it' is wrong.
 
ChronicTrees said:
Right, I guess what I meant to say by arguing both ways on it is that having sm3 vs 2 would be a variable in the Image Quality but it is not directly responsible for any visual differences.
Click to expand...
Fair enough assessment.
 
Netrat33 said:
YOU! You're a troller. YOu don't even acknowledge B&W2 screen shots!

suddenly you're bring HDR support into SM3. Why don't you bring ultrashadowing while your at it :rolleyes:


whatever I'm done with you.
Click to expand...

not reading the whole post but i thought id mention that EARLIER someone said that it would just take mroe passes for the same effect (like B&W2) anyone look at the benchmarks and compare an x800 xt to a 6800 ultra? (for performance loss...)
 
razor1 said:
This is the same argument, many things can be done on sm 2.0, 2.0b that can be done with sm 3.0 but are they playable.
Click to expand...

I think your really as streching this. Yes some will run slower, but to say that every effect done in SM3.0 will be too slow to run in SM2/2.0b is not correct as your post is making it sound...
 
0mega said:
not reading the whole post but i thought id mention that EARLIER someone said that it would just take mroe passes for the same effect (like B&W2) anyone look at the benchmarks and compare an x800 xt to a 6800 ultra? (for performance loss...)
Click to expand...

Looks like an x800 spanks the 6800ultra in AoEIII and would have more room for image improvement if disired.

In B&W it's still beating the 6800ultra but a couple of frames.
 
Terra said:
Are you saying to me that you se NO differences between the SM2.0(b) and the SM3.0 pictures from AOE? :rolleyes:

Terra - I'm not the one ranting now :p
Click to expand...

Only now am I reading the rest of the thread, and I have to say, you don't even know what you're arguing anymore.

Visual differences in AOE3 are not the issue at hand; the original issue you brought up was that there are visual differences between SM2.0(b) and SM3.0. As has been stated several times throughout this thread (and even reinforced by Brent) there is absolutely no visual difference between SM2.0(b) and SM3.0 themselves. ANY visual difference in a game is due to developer laziness.

Sure, there is a difference between the screenshots in AOE3, but the exact same effects can be done in SM2. There may be a slight performance delta, but that is irrelevant to the argument which you are presenting.
 
You must log in or register to reply here.
Back
Top