The Game Technology discussion thread *Read first post before posting*

something really bugs me is the lack of AA in DOA5, they game could have look so much better, and some of the stages have some god awful texture.
 
something really bugs me is the lack of AA in DOA5, they game could have look so much better, and some of the stages have some god awful texture.

Indeed it could have looked better, even DOA4 had AA. Probably they had some compromises with the engine, some low denominator, and they want to keep both versions as similar as possible, which can guarantee more revenue.
 
Last edited by a moderator:
with stuffs like postAA out since DOA4, I would have thought that they would look into those. Should have build the game from ground up with postAA, I don't mind little blur since they texture aren't exactly top quality.
 
with stuffs like postAA out since DOA4, I would have thought that they would look into those. Should have build the game from ground up with postAA, I don't mind little blur since they texture aren't exactly top quality.

Ultimately, a decent post-AA is going to take about 1-2ms on the GPU. I'm not so sure it'd be the best use during normal gameplay when they're already doing dynamic resolution reduction during heavier moments (explosions and what not); not really sure what you mean by ground up with post-AA as it's just a post process. Something is going to have to be sacrificed to hit 16ms (or even less to account for all situations).

On the other hand, perhaps if they had more time for polishing, they ought to have enabled some form of AA for cut-scenes and character close-ups that are 30fps. The character selection screen uses 2xMSAA FWIW. An extra 16 ms for AA/MSAA should certainly be more than enough - at least on 360 the memory footprint should remain the same.
 
Ultimately, a decent post-AA is going to take about 1-2ms on the GPU. I'm not so sure it'd be the best use during normal gameplay when they're already doing dynamic resolution reduction during heavier moments (explosions and what not); not really sure what you mean by ground up with post-AA as it's just a post process. Something is going to have to be sacrificed to hit 16ms (or even less to account for all situations).

On the other hand, perhaps if they had more time for polishing, they ought to have enabled some form of AA for cut-scenes and character close-ups that are 30fps. The character selection screen uses 2xMSAA FWIW. An extra 16 ms for AA/MSAA should certainly be more than enough - at least on 360 the memory footprint should remain the same.


by from the ground up, I mean always with the constrain included within the 16ms with post AA included. Which also means you can do less but at least it got AA......

added: I mean with a budget set aside for postAA right from the beginning.
 
Indeed it could have looked better, even DOA4 had AA. Probably they had some compromises with the engine, some low denominator, and they want to keep both versions as similar as possible, which can guarantee more revenue.

when you look at the compromises being involved for this title, i'm thinking about how much the developers were probably trying to cut back on in their minds. for a while i ran a bunch of theories of what was workable and what wasn't. it's got plenty of bloom and warm shadder fx, though, that was all in the alpha build as far back as i can remember and it looks to be in the same fidelity.

but first off, post AA is something that can work well for systems that are in constant need of memory, and it can come in handy at anytime of a game's development as it's just a shadder technique.

It helped outstandingly for Soul Calibur 5, a game that which just about brings DOA5's level of detail and animations to well....pretty much it's knees.


(In game, not photo mode.)
1280x-1

1280x-1

360_lighting_1.png

PS3_lighting_2.png


with beautiful facial animations to boot as well, even though they are not in these shots. in a sense namco couldn't have possibly been a better influence within DOA5's development time.
 
Last edited by a moderator:
what soul team can pull off always puzzle me that why can't tekken team do the same. T5 came out after SC2, it looked worse than SC2 IMO. SC3 on PS2 easily has some of the best graphics on the system.
 
what soul team can pull off always puzzle me that why can't tekken team do the same. T5 came out after SC2, it looked worse than SC2 IMO. SC3 on PS2 easily has some of the best graphics on the system.

It seems Tekken team puts more effort of their graphics engine on the emphasis of object based motion blur.

I remember some people praising Tekken 6's graphics over SCIV back then. I just don't understand it... It looked rough. DOA 2 Ultimate looked better IMO.

928302_20090819_screen002.jpg


928302_20090819_screen003.jpg








914836_20041027_screen004.jpg


914836_20041027_screen011.jpg


914836_20040709_screen007.jpg


Or even DOA4 to put it into perspective.

928260_20050909_screen005.jpg
 
Couldnt agree more. I was dissapointed with Tekken's visuals since Tekken4 excluding TTT2. Soul Calibur3 is the most impressive fighting game on the PS2. T3, TTT and TTT2 are probably the only games in the series that appeared to have set high standards and pushed the hardware to max potential. Tekken 5 looked like a downgrade in some areas. Character polys are lower, and effects have been taken a step back. Tekken 6 is indeed very rough looking. SC3 on the PS2 was a very rich visual experience. Like a reinassance paibting coming to life
 
SCIV is also incredibly good looking on PSP (and even more if you look at that on the Vita's display).

477220_388985604444862_334371729_o.jpg
 
for example forza 5 versus GT 5?

That is a prime case of better artwork instead of better programming.

Although it does include the very careful balancing of each element, for example how they construct their lighting and shaders is probably very closely tuned for the single purpose of rendering good looking cars. Choosing the best computing models and data types and such.
 
That is a prime case of better artwork instead of better programming.

Although it does include the very careful balancing of each element, for example how they construct their lighting and shaders is probably very closely tuned for the single purpose of rendering good looking cars. Choosing the best computing models and data types and such.

Being able to construct an engine which can handle realtime reflections on all cars is considered art instead of programming now?
Higher precision lighting? Real shadows and lighting to allow for 24h day cycle?

Better artwork to me would be creating tracks to look visually appealing, or modelling the cars based on the real life counterparts instead of a toy scale model. On that I fully agree :)
 
Being able to construct an engine which can handle realtime reflections on all cars is considered art instead of programming now?
Higher precision lighting? Real shadows and lighting to allow for 24h day cycle?

Better artwork to me would be creating tracks to look visually appealing, or modelling the cars based on the real life counterparts instead of a toy scale model. On that I fully agree :)

The reflections are probably a cube-map, something many games do today and even games like PGR3 had at the launch of this generation.

I don't think we know enough about the lighting models in both games to really compare them and how they differ. Though I agree GT's model is more impressive, same with the shaders on the car, but that probably is a benefit of more experience over Turn 10. GT games looked better last gen even though the xbox was ~2x more powerful than the PS2. If that's not enough to show how great art can effect a racing game, I don't know what is.

Honestly, even bringing up that comparison was foolish to begin with. Unless we have enough technical knowledge on the games being compared (which we never really do) and the person making the comparison can do a comprehensive breakdown of each engine (including trade offs and strengths) then we should really refrain from making baseless comparisons.

All my opinion of course.
 
What I'm saying is that the number of instructions, the amount of data moved, and the hardware utilization/efficiency of it all, is probably not superior in GT5 at all; they're not getting better results just because they're good at programming to the metal. But they do know a hell lot more about how to best approximate the lighting conditions and material properties to get convincing results.
 
What I'm saying is that the number of instructions, the amount of data moved, and the hardware utilization/efficiency of it all, is probably not superior in GT5 at all; they're not getting better results just because they're good at programming to the metal. But they do know a hell lot more about how to best approximate the lighting conditions and material properties to get convincing results.

I agree with the general sentiment and FWIW I think some of that is a conscious decision by turn 10's art team.
 
Yeah they probably don't want to look just like GT5, it makes sense to aim for a different art style.
 
Being able to construct an engine which can handle realtime reflections on all cars is considered art instead of programming now?
Higher precision lighting? Real shadows and lighting to allow for 24h day cycle?

Better artwork to me would be creating tracks to look visually appealing, or modelling the cars based on the real life counterparts instead of a toy scale model. On that I fully agree :)
Don't forget the physics, A.I., and dynamic weather. I would like answers to all those things including the HDR and day night cycles.
 
^ Answers for what?

If PD have released any documents or papers on GT5, that's probably where you can find anything you're looking for.
 
What I'm saying is that the number of instructions, the amount of data moved, and the hardware utilization/efficiency of it all, is probably not superior in GT5 at all; they're not getting better results just because they're good at programming to the metal. But they do know a hell lot more about how to best approximate the lighting conditions and material properties to get convincing results.

I think one part where the PS3 should have an advantage in both AI and lighting in these racing games is raycasting, which you should be able to afford much more of, and really helps, especially with the changing lighting in something like Nurburgring with day/night and weather cycles. The way evening light lits up the cars sometimes is really something special.

There are clear advantages for the 360, but I'm not seeing some of them in Turn 10s games - I am surprised for instance that Forza 4 (though pretty good looking, mind!) is weaker with framebuffer effects, where this is traditionally a weakness of the PS3. So yeah, I'm expecting the quicker turnaround and stronger adherence to 60fps (which is to be applauded) is holding back (if you can call it that) Forza technically a little, or they may well have been able to pull of some cool framebuffer effects still (though I don't know if there's just something that makes this hard to do at 60fps anyawy)
 
Back
Top