mako ( deferred framework )

[Shadowsong]

This might be totally off topic but I didn't want to create a new topic on this 

I found this patent on deferred shading:

http://www.macsimumnews.com/index.php/archive/apple_patent_is_for_deferred_shading_graphics_pipeline

What does this patent actually mean? I mean, are there any consequences of using a deferred shading approach?
Can I use a deferred shading system in a commercial product I develop? Or does this conflict with the patent in some way?

Sorry, I have never had to do with such things before.

[Lenn]

I wouldn't worry about it. Millions of people are using it, or the form of deferred shading. The patent probably has little to do with the deferred shading we are talking about. Even if it doesn't there are many large companies that'd get sued before it was your turn.

[Mithrandir]

That's only US patent, ... Who cares 

[RuntimeError101]

Wow, I just read an article recently about the problems with software patents and Apple's abuse of them.  This is almost funny.  Don't worry though.  I think the patents are Apple's backup plan and extra funding, so you probably won't be a target.  Look at these other patents that they try to claim: http://www.engadget.com/2010/03/02/apple-vs-htc-a-patent-breakdown/

[AriusEso]

That's only US patent, ... Who cares 

Quoted for truth.

[AriusEso]

Redid HDRR. This is some screens of the same scene but with different settings applied.

The function being changed is defined as:

Code:

setHDRRParameters(bool bEnableVignette, float fExposure, int iFoggingColour, float fDeFog, float fGamma, float fBlurMix);

( click for bigs )

Code:

setHDRRParameters(true, 2, RGBA(0.5f, 0.5f, 0.5f, 1), 0.2f, float(1 / 2.2f), 0.25f);

Code:

setHDRRParameters(false, 3, RGBA(0.5f, 0.5f, 0.5f, 1), 0.3f, 0.62f, 0.25f);

Code:

setHDRRParameters(true, 3, RGBA(0.5f, 0.5f, 0.5f, 1), 0.3f, float(1 / 2.2f), 0.25f);

Code:

setHDRRParameters(false, 2, RGBA(0.5f, 0.5f, 0.5f, 1), 0.1f, float(1 / 2.2f), 0.25f);

Code:

setHDRRParameters(false, 2, RGBA(0.25f, 0.25f, 0.5f, 1), 0.1f, 0.62f, 0.25f);

Code:

setHDRRParameters(false, 2, RGBA(0.25f, 0.25f, 0.5f, 1), 0.1f, 0.62f, 0.75f);

[AriusEso]

Pseudo fake gi stuffs.



Needs refining and some halo removal.

Vid coming soon, looks a little better with the actors idling away.

[AriusEso]

Here: http://www.youtube.com/watch?v=8WVxknc-6uQ

720p and 1080p versions available there also. I capped it at 1920x1200.

[AriusEso]

Started work on the actual composite.






You can see there is a lot of halo there that I'll need to remove. But otherwise he looks pretty cool.

[AriusEso]

rancor http://www.youtube.com/watch?v=k9kXXIDN15c

[AriusEso]

Added minimesh to the AO( now half-res ) pipeline.

[click for big]

[AriusEso]

Obligatory sponza.

[AriusEso]

Working properly in the ambience.

[click for big]

[ZaPPZion]

Looks good, though there's one thing that worries me, what's going on with the gamma/brightness? The colors seem to be really flat 

[Toaster]

I think maybe thats just the way he set it up. I'm sure he could possible fix it by tweaking the settings? Maybe lol

-Toaster

[AriusEso]

Looks good, though there's one thing that worries me, what's going on with the gamma/brightness? The colors seem to be really flat 

Code:

struct MAKO_HDRR_PARAMS
{
bool  bEnableVignette;
float fExposure, fDeFog, fGamma, fBlurMix;
int   iFoggingColour;
};

That last screenshot was with a gamma value of 0.62f. More accurate would be 1 / 2.2f. Although, hmm, also note that the close geometry in that shot is lit purely be ambience: 0.12f( user-defined value ) * AO. The bumpmapping does not take affect in non-lit areas, due to there being no light vector to speak of -- that ambient material is in affect on geometry outside of the light sphere. Actually took me a great deal of thought to get that to work given deferred is projected through the spheres. Therefore usually what you get is only the buffer output contained inside the spheres. I am working on a way of faking this so the bumping is taken into account, maybe a simple light directional vector would work well. You can see in this shot that in the actual sphere it's not as flat, look at the cloth bumping for example.

[click big]


Having said all this, the colours are slightly washed out and the HDRR is not of the quality I would like yet. You're certainly right that it needs some work done. I haven't finished the HDRR yet. It's one of those things that is easy to do but I've always had trouble getting it perfect. I intend to work on it more .

[ZaPPZion]

I'd thought the colors would be more like this (did some color correction in photoshop)

[AriusEso]

Yes, they're definately being washed/whitened out slightly. Probably the AO layer. Let me tinker around and see if I can produce something closer.

[AriusEso]

It is the specular colour causing it. The sponza here has no specular applied, but the actors do.

]

That is with gamma @ 1.0f / 2.2f.

Obviously my specular is screwed somewhere.

[AriusEso]

[click for big]
]

I think much better now?