Do ANY gamecube games use 8 texture passes?
- Thread starter Fox5
- Start date
Evil_Cloud
Veteran
Fox5 said:
That 'somehow' is the use of mapping techniques + a powerfull in-game engine
Grall
Could you post a quote for me, because I haven't seen Faf say that.
The very nature of a second/third ect pass, in the case we're discussing, is to re-transform and/or re-send geometry to give the GPU texture info that was not available to it in the first pass (info for the extra texture layers). That's why when you need an extra pass, for whatever reason, it can effect what sort of polygon counts you can achieve. So I can't see how using more texture layers can effect GC's poly pushing power if no extra passes are required.
I mean there would be a minimal effect for using an extra texture layer simply because your taking up more bandwidth ect that could be used to produce more polys. But not the sort of effect we're talking about here (polycounts dropping to Sonic Adventure 1 levels, which would happen if GC could only do one texture layer per pass).
Could you post a quote for me, because I haven't seen Faf say that.
The very nature of a second/third ect pass, in the case we're discussing, is to re-transform and/or re-send geometry to give the GPU texture info that was not available to it in the first pass (info for the extra texture layers). That's why when you need an extra pass, for whatever reason, it can effect what sort of polygon counts you can achieve. So I can't see how using more texture layers can effect GC's poly pushing power if no extra passes are required.
I mean there would be a minimal effect for using an extra texture layer simply because your taking up more bandwidth ect that could be used to produce more polys. But not the sort of effect we're talking about here (polycounts dropping to Sonic Adventure 1 levels, which would happen if GC could only do one texture layer per pass).
Do ANY gamecube games use 8 texture passes?
Do you mean at the same time or total?
Do you mean at the same time or total?
cthellis42 said:Do ANY gamecube games use 8 texture passes?
Do you mean at the same time or total?
Bleh.....
Ok, well if there is no or little performance hit(to having 8 texture layers), then how come developers don't make use of it? Heck, how many games even use 4 layers? I think Agent Under Fire and Nightfire only used 2 layers, and they had better textures/effects(the reflections on the car windows) than many other gamecube games I've seen.
Eh, how about someone explain to me exactly what a texture layer is first? Like a texture on top of a texture?
DeathKnight
Regular
That's a foolish notion. It does affect polygon counts just like with any other graphics processor. It still eats up clock cycles. Do you really believe that Flipper can still process a full 32.4mpps with 8 texture layers? It's actually more like 10mpps or less. Just adding 1 texture drops the polygon rate to 26.4mpps. The NV2A can apply up to 4 texture layers in a single pass (not resending geometry), yet it can't sustain the full 116.5mpps figure with that amount of textures.Teasy said:
DeathKnight said:
My comment wasn't supposed to be an exact technical statement. I was only trying to say quickly, in lamens terms, that GC is not effected by extra passes within 8 texture layers. So I meant that GC's polygon counts aren't effected purely in the sense that geometry doesn't have to be re-sent/re-transformed. As you can see later in the thread I breifly mentioned other things that could effect polycounts when using more textures (more bandwidth usage ect).
Ooh-videogames
Regular
If each gpu has a different architecture, would it be possible that each gpu cost when it comes to the use of features would be viewed differently. It seems as if people make the assumption what applies to the PS2 or XBox also applies to the GC and vice versa. I have yet to see any other third-party game come close to matching what F5 acheive on the GC texture and polycon count wise. I doubt Faf or ERP have attempted to use 8 texture in one pass. The textures in Rebel Strike are some of the sharpest textures I have seen in a game. The engine is built around the strenghts of the hardware and optimzed for optimal performance. Some one that just porting to a console, I beleive can't give you an accurate answer on what can and can't be done on a console.
Teasy said:
Yes, I could since he did say something like that, but that would require hunting through millions of back-page threads to find it. I just can't be arsed to do it.
Simply put, he explained extra tri setup would limit the amount of polys that can be rendered per second, despite the fact the chip can do 8 textures in just one pass.
Seems logical to me.
Do not joke around in this manner again.
32 milions is the number of lighted, flat shaded poligons (not texture mapping).
GCN is capable of apply 8 effects per texture (not multitexture effect) and it affects the fillrate a lot.
If the effect is a multitexture effect (Multitexture Mapping, Bump Mapping and SubPixel Antialising) because TEV is not capable of do Multitexturing in one pass the number of effects per texture go down.
GCN is capable of apply 8 effects per texture (not multitexture effect) and it affects the fillrate a lot.
If the effect is a multitexture effect (Multitexture Mapping, Bump Mapping and SubPixel Antialising) because TEV is not capable of do Multitexturing in one pass the number of effects per texture go down.
I believe that the only instance of this I have seen to date is the water in Rogue Squadron 2. And that really left a mark on the framerate...
I think it would be helpful if some people exactly explained why this isn't done more often considerig the much publicised "8 textures per pass"?
I think it would be helpful if some people exactly explained why this isn't done more often considerig the much publicised "8 textures per pass"?
Ooh-videogames
Regular
Li Mu Bai said:
Li Mu Bai, I totally agree, it doesn't make sense at all, they wouldn't even release raw polygon numbers but they publicise a feature. Neither console gpu architecture is the same and the cost for using certain features are the same but seem to believe this is so.
Li Mu Bai said:
well, that's the same question we should ask pretty much every PC graphics card manufacturer... PC Graphics cards are full of features which for the most part never get to be utilised because the architecture isn't powerful enough to substain perfect framerates.
Some features might be used in small portions, while others are just not used at all.
And let's not forget the art assets demands increase a lot when having to draw 8 textures for a single surface...
Another reason in the PC world is that some HW provides feature "X" while others have "X+Y" but a games developer may only have time to support the lesser feature set.london-boy said:
For example when multi-texturing was really taking off, some HW systems supported 2 layers, some 3(?), others 4, while Kyro could do 8. There were games that used up to 5 texture layers but always issued them as 2 passes because they assumed no HW could do it in one go. <shrug>
