GS didn't support shaders and had plenty of limitations even compared to dx6/7 pc-cards, but it was Very fast and with a local memory you really could do some crazy things with it.
Also VUs were quite intressing and could be used as a very advanced vertex processors even with their limited memory. (VU0 8KB and VU1 16KB)
Not having modern GPU doesn't mean that you can't use the multipass and blending modes to do the math needed for all sorts of 'shader' effects like gaussian blurs and several methods of bump mapping.
Here's two methods for bump mapping using normal maps.
http://playstation2-linux.com/files/p2lsd/ps2_normalmapping.pdf
Shiny the developers of Matrix
ath of Neo claim to use normal mapping methods for many objects in the game including game characters.
It does look like they do use it from screenshots, but the art in the game really isn't too great and doesn't seem to have much in emphasis in lighting.
For nice blurs and spherical harmonics on ps2, I recommend to read this little gem.
http://research.tri-ace.com/Data/Pr...SH Lighting and HDR Rendering Full-Length.ppt
Also VUs were quite intressing and could be used as a very advanced vertex processors even with their limited memory. (VU0 8KB and VU1 16KB)
Not having modern GPU doesn't mean that you can't use the multipass and blending modes to do the math needed for all sorts of 'shader' effects like gaussian blurs and several methods of bump mapping.
Here's two methods for bump mapping using normal maps.
http://playstation2-linux.com/files/p2lsd/ps2_normalmapping.pdf
Shiny the developers of Matrix
ath of Neo claim to use normal mapping methods for many objects in the game including game characters.It does look like they do use it from screenshots, but the art in the game really isn't too great and doesn't seem to have much in emphasis in lighting.
For nice blurs and spherical harmonics on ps2, I recommend to read this little gem.
http://research.tri-ace.com/Data/Pr...SH Lighting and HDR Rendering Full-Length.ppt
B
Yes, the vu1 does the lighting coding for making the shading slither by slither, pixel by pixel change the darkness and lightness via a hieght map.
Accordding to the doc the if the normal maps layer are stored in a free alpha channel then it could be the same with EMBM or greyscales. In my own opinion though n maps take too many passes 2 is good enough, but I suppose a greyscale since it's not multi chanel but one channel could be accompleshed in one single pass :mrgreen:.
So i'd rather use EMBM greys if I made a ps2 game, as n-mapping look realy just the same in motion(and are less complicated in math and to the artist to make).
There is also the new geo-textures envented on the ps2 and GC SC:CT wich like motion blurs or real-time reflections render the polys on the fly as a full screen high rez texture image. And the best way to go is to use HIGH poly flat plane of a mesh, then use a displace map witch is like a hieght map only it molds the polys to it rather(GT4 uses this). On SC:CT they realy only used normal meshes as it was only recently invented, but since they say it uses no polys for those particular meshes then imagine them doing high res TRUE BUMPS!
EDIT:Also you judge the vu's real performance realy by thier calculation...both are 2.8 or 2.9 each...meaning it's like xbox where there are only about 3 Gflops for AI...and then you get rid of the vertex shaders or combine them and put them on the ps2's cpu. Wich means doing normal mapping effects no AI space as thats on the vu0. This makes xbox and ps2 very similar if you think about it. Though xbox may still be alittle in the lead, their games could of been at least similar looking.
Also if you noticed jlippone that PON if you have it that it has a lighting detail meter to mess with....also the reason it doesn't look like it uses it alot is because they dont' program it to interact with non specular lighting. Why? I don't know...but I do know that it may be to emulate a parallax mapping effect with the normal maps instead of the traditional using it to correct incorrect and still light and shadows in textures.
Accordding to the doc the if the normal maps layer are stored in a free alpha channel then it could be the same with EMBM or greyscales. In my own opinion though n maps take too many passes 2 is good enough, but I suppose a greyscale since it's not multi chanel but one channel could be accompleshed in one single pass :mrgreen:.
So i'd rather use EMBM greys if I made a ps2 game, as n-mapping look realy just the same in motion(and are less complicated in math and to the artist to make).
There is also the new geo-textures envented on the ps2 and GC SC:CT wich like motion blurs or real-time reflections render the polys on the fly as a full screen high rez texture image. And the best way to go is to use HIGH poly flat plane of a mesh, then use a displace map witch is like a hieght map only it molds the polys to it rather(GT4 uses this). On SC:CT they realy only used normal meshes as it was only recently invented, but since they say it uses no polys for those particular meshes then imagine them doing high res TRUE BUMPS!
EDIT:Also you judge the vu's real performance realy by thier calculation...both are 2.8 or 2.9 each...meaning it's like xbox where there are only about 3 Gflops for AI...and then you get rid of the vertex shaders or combine them and put them on the ps2's cpu. Wich means doing normal mapping effects no AI space as thats on the vu0. This makes xbox and ps2 very similar if you think about it. Though xbox may still be alittle in the lead, their games could of been at least similar looking.
Also if you noticed jlippone that PON if you have it that it has a lighting detail meter to mess with....also the reason it doesn't look like it uses it alot is because they dont' program it to interact with non specular lighting. Why? I don't know...but I do know that it may be to emulate a parallax mapping effect with the normal maps instead of the traditional using it to correct incorrect and still light and shadows in textures.
Likes:
NoUseMercenary
This is kinda pointless, as PS3 is right around the corner. If you want super graphics, its not gonna be PS2.
Likes:
NoUseMercenary
B
Still it's nice to know it can do it, also it gives alot of developers no ecxuse when making an xbox port like those 2k baseball games, or double agent and how it won't includ some good bump mapping while Hitman:BM and Neo does. It's not that hard to do as all you need is anothers normal mapping or EMBM engine. Sad as Double Agent could of been ps2's graphical standpoint gem. As of now Path of Neo is.
B
Inconveniently bumping an old post, but I found out some new things on normal mapping on ps2. On the pdf normal mapping on ps2 it shows the only thing the Vu1 does is tesselate the geometry using the normal mapping(tiling the geometry, meaning reapeat the per pixel code quicky for every geometry using normal maps). This means normal and bump mapping is nothing but code out of the same ol shading in a 3d game and that ps2 wasn't actually "emulating" it in software. It's just that it's simple coding(thogh they don't say if the per pixel is being done on vu1).
Vertex shaders are merely the polygon transformation and lighting tool is all, with built in already coded per-pixel things for normal and bump mapping. On ps2 the GS and Vu1 are the poly transformation and lighting tool, and with 2.9Gflops in each vu, the per pixel math is actually quite simple, as it was ony several small lines of code out of the already there shading. Also it's not quite software when they do that stuff, it's just that the polygon transformation and lighting part is on the cpu, the Vu0 or Ai and physics vector is untouched. Also there is this....
http://freespace.virgin.net/hugo.elias/graphics/x_polybm.htm
See all the way at the bottom it shows that the earlier(and hardly taxing) grayscale version of bump mapping can use a small 2 line code of checking the x and y of the gradient at the pixel to calculate vectors. Then another tiny 1 line code to incert those vectors onto the normal, where the everyday shading(as gaurod lighting can shade it too) x y cordinates and z direction of the light at that part of the shading or texture light has a new texture gradient look up based on that calculation of the position of light and the vectors calculated on every pixel from a two line code. I seen this myself on hitman:BM on ps2 and even on SCCT on my pc that that is how it works.
This all means bump mapping if developers weren't kids and knew how to code bump mapping themself then ps2 could had bump mapping on alot of the games it had out. Because actually the thing that made xbox so popular was Halo and bump maps.
If Killzone had grayscale bump mapping on all textures it would tax les hardware then it does now. Grayscale bump maps can be done in one pass and still be the same as your path of neo 2 or 4 pass normal maps. And killzone had tons of multi-pass multi-layer textures to emulate realistic quality. If it used bump maps then it would render normal non-detailed color textures, then instead of lots of passes for multiple layers on all the textures only one pass on all textures would be needed for the height map, then the small coding for the vectors and then you have a real halo killer(especually since the bump maps wont be plasticy from specular, but realisticly film like).
Sorry for bumpping when ps3 is now out. But I wanted to share as none of those developers deserve credit with the grayscale bump maps looking the same as normal maps and being so light in passes and easy to do, I just hate looking at those ports of xbox games to ps2, like the sega baseball game and how it's so small in size and looks worse then path of neo and scarfaceyet they put the xbox shots on the back.
Vertex shaders are merely the polygon transformation and lighting tool is all, with built in already coded per-pixel things for normal and bump mapping. On ps2 the GS and Vu1 are the poly transformation and lighting tool, and with 2.9Gflops in each vu, the per pixel math is actually quite simple, as it was ony several small lines of code out of the already there shading. Also it's not quite software when they do that stuff, it's just that the polygon transformation and lighting part is on the cpu, the Vu0 or Ai and physics vector is untouched. Also there is this....
http://freespace.virgin.net/hugo.elias/graphics/x_polybm.htm
See all the way at the bottom it shows that the earlier(and hardly taxing) grayscale version of bump mapping can use a small 2 line code of checking the x and y of the gradient at the pixel to calculate vectors. Then another tiny 1 line code to incert those vectors onto the normal, where the everyday shading(as gaurod lighting can shade it too) x y cordinates and z direction of the light at that part of the shading or texture light has a new texture gradient look up based on that calculation of the position of light and the vectors calculated on every pixel from a two line code. I seen this myself on hitman:BM on ps2 and even on SCCT on my pc that that is how it works.
This all means bump mapping if developers weren't kids and knew how to code bump mapping themself then ps2 could had bump mapping on alot of the games it had out. Because actually the thing that made xbox so popular was Halo and bump maps.
If Killzone had grayscale bump mapping on all textures it would tax les hardware then it does now. Grayscale bump maps can be done in one pass and still be the same as your path of neo 2 or 4 pass normal maps. And killzone had tons of multi-pass multi-layer textures to emulate realistic quality. If it used bump maps then it would render normal non-detailed color textures, then instead of lots of passes for multiple layers on all the textures only one pass on all textures would be needed for the height map, then the small coding for the vectors and then you have a real halo killer(especually since the bump maps wont be plasticy from specular, but realisticly film like).
Sorry for bumpping when ps3 is now out. But I wanted to share as none of those developers deserve credit with the grayscale bump maps looking the same as normal maps and being so light in passes and easy to do, I just hate looking at those ports of xbox games to ps2, like the sega baseball game and how it's so small in size and looks worse then path of neo and scarfaceyet they put the xbox shots on the back.
I didnt think it was. And havent seen anything even resembling it in any of the games I've played over the years.....until now. That's right. Its 2022 & I'm in CEX perusing some old PS2 titles (hoping to find god if war 1 or 2 or shadows of the collosus) when I saw Final Fantasy X 2. I bought it when it came out, and despite it not being as good as 12 is on the PS2, I bought it anyway for the nostalgia. So here I am playing it right now, and have just noticed that on the menus, all the little nicks and scratches and dents in all the parts of the menu screen when you hit triangle, all seem to have dynamic shadows, which I never noticed before. The cursor has "light" coming from it, and when you go on say, items, and move it around the screen, the shadows and light change and move like they would on the textures in any ps3 game. And when I just saw it I had to look up if any other games had bump mapping. The first thing that came up was this site. But I just had to mention what I saw on this game. It looks cool. I dont know if its true bump mapping, but whatever it is, I'm surprised they didnt use it for stuff in the game world. And now I'm wondering if they used, whatever it is they are using here, in any other games? I wonder if Silent Hill 2 used it anywhere? But anyway. Just wanted to let my ASHD ramble on as I tried to say I saw what looks like bump maps in the menus of Final Fantasy X 2
Likes:
NoUseMercenary