Re: [Algorithms] [SPAM] Re: Crease Shading, a screen-based approximation to ambient occlusion

[<Pau...@sc...>]

gda...@li... wrote on 19/11/2007 
22:35:23:

> The biggest difference is that our method is effectively biased to
> care most about those neighbors closest to directly in front of the
> normal, and less about those further away from the normal, where that
> range is customizable with a variety of tweak factors.  It's an
> artistic approach.  His is ray-casting out from the point, finding
> neighbors in a set range, and attempting to assemble a visibility
> sphere from that data / boil that sphere down to an AO value.  If you
> think about it, our approach is like a variant of his where you
> balance the radii of his occlusive spheres such that they are
> dependent on the angle between the center normal and the
> center-to-sphere normal.

Actually, the core of your technique is very close to the utexas research:

They aren't actually ray-casting, they're just finding a number of sample 
pixels within a radius around the centre pixel, projecting said pixels 
into world space and then the use the same maths as you in their loop:

N = centre_normal
loop
{
        PC = normalize(centre_pos - sample_pos)
        ao += max( dot(N,PC), 0 )
}

They also have some code in there to account for the occlusion spheres 
(just a weighting on the accumulation) and some attenuation stuff as well. 
Its not the "same" per se, just fairly similar. :)

Cheers, Paul.

**********************************************************************
This email and any files transmitted with it are confidential and intended solely for the use of the individual or entity to whom they are addressed. If you have received this email in error please notify pos...@sc...
This footnote also confirms that this email message has been checked
for all known viruses.
Sony Computer Entertainment Europe Limited
Registered Office: 10 Great Marlborough Street, London W1F 7LP, United Kingdom
Registered in England:3277793
**********************************************************************