Questions about PS2

the amazing water was all geometry, no normal maps.
Yes, I know that. You can see it in game very clear! Probably most amazing water tech in 6th gen of consoles.
T IIRC the place we saw 'normal maps' were the bumpy, shiny floors, and that's more fakeable as you have a static camera orientation so can make assumptions (ie. more bump mapping than normal).
But does PS2 support Bump Mapping?
 
PS2 relies on software solutions, not hardware. So there's no fixed hardware bump mapping or normal mapping built in. However, the capable maths performance meant devs could choose to implement whatever algorithms they could get to work within their frametime and targets, and bump-mapping is one of those options. As a technique, yes, PS2 supports Bump Mapping. Whether it was too costly to practically use is something one ofthe PS2 devs will need to answer.
 
Last edited:
Sorry about double post.
PS2 relies on software solutions, not hardware. So there's no fixed hardware bump mapping or normal mapping built in. However, the capable maths performance meant devs could choose to implement whatever algorithms they could get to work within their frametime and targets, and bump-mapping is one of those options. As a technique, yes, PS2 supports Bump Mapping. Whether it was too costly to practically use is something one ofthe PS2 devs will need to answer.
Great explanation! Thank you so much.
 
Dou you thint same is with Matrix Path of Neo (link above) or other games?
Well, since the hardware doesn't have the capability, it stands within reason that other games also fake the results. So, the answer would be yes, it's the same for other games as well. :)

But does PS2 support Bump Mapping?
No, it does not. PS2 rasterizer (Graphics Synth) is very bare-bones, feature-wise; it doesn't even have a full set of alpha blending modes IIRC, but what it does have is shitloads of fillrate (comparable to hardware of its time, and even for years afterwards too), due to lots of pixel rendering pipes (16 total; 8 that can texture) and the eDRAM with its double 1kbit wide read/write ports and 512bit wide texture read port, giving it absolutely staggeringly massive aggregate on-chip bandwidth for a turn of the century era piece of hardware. So you could pile on layers and layers of textures to create a specific effect using brute force and not really slow that thing down.

IIRC, John Carmack once matemathically proved that using Boolean logic you could perform any pixel shading op just with texturing and alpha blending - theoretically. In reality, advanced operations would require massive amounts of blending passes and 8-bit integers just don't have the precision needed. Everything would literally turn into mud. :) But you can fake stuff without having to actually perform the operation itself - you just do something simpler which gives a close-enough visual result, under the specific conditions of your game. Under other conditions, the result might look completely wrong, but that doesn't matter for this particular game.

So it's not really important if PS2 supports this-or-that feature; if the end result looks the same - or sufficiently similar - then it doesn't really matter. :) Almost every game out there doing advanced rendering effects fake things to get the game to run fast enough and/or look the way the developers want. Faking it is a time-honored tradition in the gaming business! :)
 
Indeed. And talking of Carmack, I seem to recall Champions of Norrath used mega textures.


Verzonden vanaf mijn iPhone met Tapatalk
 
PS2 relies on software solutions, not hardware. So there's no fixed hardware bump mapping or normal mapping built in. However, the capable maths performance meant devs could choose to implement whatever algorithms they could get to work within their frametime and targets, and bump-mapping is one of those options. As a technique, yes, PS2 supports Bump Mapping. Whether it was too costly to practically use is something one ofthe PS2 devs will need to answer.

Morten Mikkelsen (IO Interactive) wrote a paper on DOT3 normal mapping on the PS2. With (four passes) and without (two passes) normalization.

Cheers
 
Well, since the hardware doesn't have the capability, it stands within reason that other games also fake the results. So, the answer would be yes, it's the same for other games as well. :)
I got it! :D
IIRC, John Carmack once matemathically proved that using Boolean logic you could perform any pixel shading op just with texturing and alpha blending - theoretically. In reality, advanced operations would require massive amounts of blending passes and 8-bit integers just don't have the precision needed.
But you said that PS2 doesn't even have a full set of alpha blending modes. So how it can use it for advanced effects?:sly: :D
So it's not really important if PS2 supports this-or-that feature; if the end result looks the same - or sufficiently similar - then it doesn't really matter. :) Almost every game out there doing advanced rendering effects fake things to get the game to run fast enough and/or look the way the developers want. Faking it is a time-honored tradition in the gaming business! :)
Now, I realy start to understand how PS2 worked. Thank you a lot for great and big explanation.
Also can you tell me what Graphic Synthesizer 16 pixel processors are used for if they don't support shaders?
 
Really? No way!:D Too good to be truth. :D
Yes, TGTBT. It had a load of prebuilt tiles on disc which it streamed in. An issue with reading the second layer meant you could get blacked out tiles for a few seconds. It didn't have megatexture in the true sense of a single texture atlas for all scenery and tile-based textures.
 
To quote a post from Shifty Geezer from 2011:
Code:
Champions of Norrath was fabulous on PS2. Dynamic shadow casting lights, zillions of particles, fluid dynamics and supersampled IQ, all at 60 fps (especially RTA that didn't have the dual layer DVD streaming issues).
 
Yes, TGTBT. It had a load of prebuilt tiles on disc which it streamed in. An issue with reading the second layer meant you could get blacked out tiles for a few seconds. It didn't have megatexture in the true sense of a single texture atlas for all scenery and tile-based textures.
WOW, that's impressive! Truly. Watched some videos from game. I think I need to play it now. :D

To quote a post from Shifty Geezer from 2011:

Very good find! Thank you.

Wasn't also Champions Return to Arms even more impressive?
 
WOW, that's impressive! Truly. Watched some videos from game. I think I need to play it now. :D
Very, very entertaining game. Real shame the genre climaxed with RTA, although some D3 fans will disagree.

CON had procedural dungeons which RTA lacks, making CON the more replayable. RTA is more of an expansion in terms of content.
 
But you said that PS2 doesn't even have a full set of alpha blending modes. So how it can use it for advanced effects?:sly: :D
You don't need a full set of blending modes to do advanced effects. Some of the GS's deficiencies could be worked around by doing things in a less obvious (and typically slower) way. Another piece of hardware might have been slowed down, but PS2 had so much fillrate you usually didn't have to care about that at all. Some PS2 dev posted on here years ago that as a test he'd made a demo that re-drew the entire screen a silly amount of times per frame (like, a dozen layers, maybe several dozen, I can't recall), at 60 frames/sec.

Any other piece of hardware at the time would have just rolled over and died from trying to do that. :)

Also can you tell me what Graphic Synthesizer 16 pixel processors are used for if they don't support shaders?
They're not 'processors' in the sense they're freely programmable. They're fixed-function pipelines with some knobs you can turn to switch on and off certain features. What happens is that the GS reads display list data and set the pixel pipelines to perform certain operations (read texels from a texture for example at certain coordinates, filter them a certain way, maybe add interpolated vertex colors for lighting or whatever, maybe fog, then draw resulting data to certain screen coordinates. Then repeat until polygon is fully drawn.

What exactly happens or how isn't that important, unless you're like either a high-level programmer or a hardware engineer. It's beyond my level to explain anyhow as I'm neither of these things! :D
 
CON had procedural dungeons which RTA lacks, making CON the more replayable. RTA is more of an expansion in terms of content.
Interesting. So now I should try both games and compare.
You don't need a full set of blending modes to do advanced effects. Some of the GS's deficiencies could be worked around by doing things in a less obvious (and typically slower) way. Another piece of hardware might have been slowed down, but PS2 had so much fillrate you usually didn't have to care about that at all. Some PS2 dev posted on here years ago that as a test he'd made a demo that re-drew the entire screen a silly amount of times per frame (like, a dozen layers, maybe several dozen, I can't recall), at 60 frames/sec.
Is it because of that so many games on PS2 were 60fps? Also I've heard that on PS2 were games where was a lot of fog, smoke, dust etc. Is that was possible because of so high fillrate? And when entire screen was re-drew many times, doesn't it increase geometry count?
They're not 'processors' in the sense they're freely programmable. They're fixed-function pipelines with some knobs you can turn to switch on and off certain features. What happens is that the GS reads display list data and set the pixel pipelines to perform certain operations (read texels from a texture for example at certain coordinates, filter them a certain way, maybe add interpolated vertex colors for lighting or whatever, maybe fog, then draw resulting data to certain screen coordinates. Then repeat until polygon is fully drawn.

What exactly happens or how isn't that important, unless you're like either a high-level programmer or a hardware engineer. It's beyond my level to explain anyhow as I'm neither of these things! :D
But just to be clear, did they made some effects in games, postprocessing?
 
Interesting. So now I should try both games and compare.
RTA would be the better game perhaps if you haven't played CON. CON was the more impressive and memorable game when it released because it was a sequel to BGDA introducing four players. No harm playing both though! If only Sony allowed disc-based BC on PS4.

Also I've heard that on PS2 were games where was a lot of fog, smoke, dust etc. Is that was possible because of so high fillrate?
Yes. You draw, draw, and draw again!
And when entire screen was re-drew many times, doesn't it increase geometry count?
Yes when you draw the same polygons over and over. PS2 had high fillrate and polygon draw, but in comparisons this was misleading because other consoles with less triangles per second could do more with each triangle.
 
RTA would be the better game perhaps if you haven't played CON. CON was the more impressive and memorable game when it released because it was a sequel to BGDA introducing four players. No harm playing both though! If only Sony allowed disc-based BC on PS4.
Ok, but still I can get better both of them.:D So I'll play both.
Yes. You draw, draw, and draw again!
I remember in WRC games there was a lot of dust after car then you drive it on off-road. Very impressive. Even on Xbox 360 and PS3 in Dirt series games and WRC series games wasn't so much dust. Did it because of lack of fillrate? Also great example is The Lord of the Rings The Return of the King. In level there Aragorn goes to speak with ghosts he goes through cave and in some places there is a lot of fog. I remember on PS2 there was just the same frame rate as in other places of the game. But on my old PC frame rate drops to some 10fps. This was amazing win of PS2. My PC was Athlon 1,4 Ghz, Radeon 9600 Pro with 256 Mb and 1Gb of RAM.
 
Yes when you draw the same polygons over and over. PS2 had high fillrate and polygon draw, but in comparisons this was misleading because other consoles with less triangles per second could do more with each triangle.
But was it used in real games? And wasn't PS2 capably draw so many polygons what it was possible to do it? And one more question did multi pass rendering required twice geometry calculations with two passes and four times more geometry calculations with four passes?
 
I think I remember an interview with Kazunori where he laments about the PS2's read-mod-write capabilities and how he wasn't able to do effects like the heat wave from the race track in GT5-6 like he could do in 3-4.

If you are interested, John Stokes wrote a nice technical overview of the Playstation 2 Emotion Engine http://archive.arstechnica.com/reviews/1q00/playstation2/m-ee-1.html

A comparison of the PS2 system to a PC at the time
http://archive.arstechnica.com/cpu/2q00/ps2/m-ps2vspc-1.html
 
Yes, I've heard about GT. Very interesting but also I don't understand why they can't do similar effect in GT 5-6 with shaders. There wasn't resources of power on PS3 anymore?
Thanks for links.
 
Back
Top