New articles and demos

Humus

Crazy coder
Veteran
I've added an articles section on my site. I've written two articles, which will be published in the ShaderX2 book later this year.
The topics are:
Fragment level Phong illumination
Mandelbrot set rendering


To each of these articles there are a related demo illustrating the topics in the articles.

phongillumination.jpg

mandelbrotset.jpg


Check it out, http://esprit.campus.luth.se/~humus/
 
Lol....this is my favorite quote from the Phong article:

There are several shadowing techniques to choose from and some of them exist in several different forms. Unfortunately, they all suck in one way or the other.

:D
 
Joe DeFuria said:
Lol....this is my favorite quote from the Phong article:

There are several shadowing techniques to choose from and some of them exist in several different forms. Unfortunately, they all suck in one way or the other.

:D

Hehe, let me nitpick momentarily. : ;)
That really ought to read "Unfortunately, they all suck in one way or another."

ok, with that out of the way, Congrats on getting published Humus! :D
 
Humus,

Great demos again, as usual! Before you hit the road, would it be possible to have the Soft coronas demo, without the music? ;) Not that I don't like the music, but every machine I try it on, is giving me a "can't find the damn vorbisfile.dll, and you 'ain't gonna get any!"

A pretty plllllease! :D
 
Ilfirin said:
Is there any particular reason why you are doing multiple rendering passes to a cube-map for omni-directional lights vs dual-paraboloid texture maps?

Yes. Dual-paraboloid isn't linear, it wastes space in the shadow map and requires two texture lookups for each sample since we don't have any data-dependent branching in the pixel shaders. AFAICS the only time dual-paraboliod would be advantageous is when we're transform limited.
 
Ichneumon said:
Hehe, let me nitpick momentarily. : ;)
That really ought to read "Unfortunately, they all suck in one way or another."

True, sucks that the spelling/grammar check can't pick up errors like that. Anyway, it's fixed now.
 
worm[Futuremark said:
]Humus,

Great demos again, as usual! Before you hit the road, would it be possible to have the Soft coronas demo, without the music? ;) Not that I don't like the music, but every machine I try it on, is giving me a "can't find the damn vorbisfile.dll, and you 'ain't gonna get any!"

A pretty plllllease! :D

Well, I could compile and include a no-audio .exe too. Otherwise, I assume you have access to a compiler (think it's fairly safe to assume there's at least one within the walls of futuremark ;)) so you could compile a non-audio version yourself. Just uncomment the NO_AUDIO define at the top of the main file and remove all audio related files from the project.
 
Ingenu said:
Still boost sound to 100% with latest SBLive! drivers and Win2k SP3

I assume you're talking about the Soft Coronas demo too (the new demos don't have sound).

It gotta be a driver bug or some really odd system setting. Cause the last time I heard about a bug like this I was using the bass library. Now I'm using OpenAL. There's no shared code, so I can't see a it being my bug. Especially since I don't touch any volume settings at all.
 
Humus said:
Well, I could compile and include a no-audio .exe too.
Tackar och bockar! ;)

Humus said:
Otherwise, I assume you have access to a compiler (think it's fairly safe to assume there's at least one within the walls of futuremark ;)) so you could compile a non-audio version yourself. Just uncomment the NO_AUDIO define at the top of the main file and remove all audio related files from the project.
But but.. You do it so well! :D No really, sure I can ask our "bois" to help me out. They might do it, someday. :rolleyes:

Oh, and about your new job, Congrats!
 
Humus,
great work, sadly I can't watch the demos.

However, in the mandelbrot paper you're using the code
Code:
mad r2.xy, r0.x, r0, t0
mad r1.x, -r0.y, r0.y, r2.x
mad r1.y, r0.x, r0.y, r2.y
which seems to me like a waste. You're actually using only four 4 mad while you'd get 12 'for free'. The problem is that you have to invert Yi once. The easy but not-so-obvious solution is to calculate negative Yi along with Yi. That results in:
Code:
mad  r1.xyz, r0.zyx, r0.yxz, t0.xyz
mad  r0.xyz, r0.xxx, r0.xyz, r1.xyz

(Note that I'm not sure this kind of swizzling works on R300, the DXSDK documentation is a bit vague on this). Now all you have to do is put -Cy in t0.z, which you can do easily by using 3D texture coordinates.
 
That kind of swizzling isn't allowed, this is what the compiler says:

(52): error X5202: (First source param) Invalid source selector: .bgrr/zyxx. The only available source swizzles in this shader version are: .rgba/xyzw (same as not specifying swizzle), .r/x, .g/y, .b/z, .a/w, .gbra/yzxw, .brga/zxyw and .abgr/wzyx.
(52): error X5202: (Second source param) Invalid source selector: .grbb/yxzz. The only available source swizzles in this shader version are: .rgba/xyzw (same as not specifying swizzle), .r/x, .g/y, .b/z, .a/w, .gbra/yzxw, .brga/zxyw and .abgr/wzyx.
(52): error X5202: (Third source param) Invalid source selector: .rgbb/xyzz. The only available source swizzles in this shader version are: .rgba/xyzw (same as not specifying swizzle), .r/x, .g/y, .b/z, .a/w, .gbra/yzxw, .brga/zxyw and .abgr/wzyx.
(53): error X5202: (Second source param) Invalid source selector: .rgbb/xyzz. The only available source swizzles in this shader version are: .rgba/xyzw (same as not specifying swizzle), .r/x, .g/y, .b/z, .a/w, .gbra/yzxw, .brga/zxyw and .abgr/wzyx.
(53): error X5202: (Third source param) Invalid source selector: .rgbb/xyzz. The only available source swizzles in this shader version are: .rgba/xyzw (same as not specifying swizzle), .r/x, .g/y, .b/z, .a/w, .gbra/yzxw, .brga/zxyw and .abgr/wzyx.
 
Humus said:
Well, I've included an no audio .exe now.
Thanks! It looks great man.

question; would it be possible to sort the coronas so that they don't show in places where they actually shouldn't? Are there any resons why it is not done? Dunno how else to explain it.. :?
 
worm[Futuremark said:
]question; would it be possible to sort the coronas so that they don't show in places where they actually shouldn't?

Are you talking about the Unreal Editor's bugs?
 
Back
Top