Le 2020-03-15 à 18:24, Anthony D'Agostino a écrit :
> I wanted to see how close Radiosity could get to pure ambient occlusion. 
> I found out that it's very close, and that there are only some minor 
> differences. The bottom line is that radiosity (using one bounce and a 
> white background) is about the same as regular AO.
> 
> For this experiment, an object with lots of nooks and crevices was 
> helpful, and that's where my inverted quintrino came to the rescue. The 
> only adjustment I did to the image was an inverse gamma correction.
> 
> You can see the results on my webpage by following link below. You can 
> drag the slider back and forth, or click "AO" or "RADIOSITY" at the top 
> and it will slide across automatically. Or you can click anywhere in the 
> image and the dividing line will slide to that location.
> 
> https://scorpius.github.io/compare.htm

Well, radiosity > ambient occlusion.

Here's why : Ambient occlusion only takes into account the proximity of 
the occluding object. It can't give you the effect of a flame 
illuminating it's surrounding, the effect of a mirror nor that of a 
crystal sphere projecting other objects onto a near surface. Also, it 
neglect the effect of a strongly coloured object have on it's surrounding.

Radiosity do take into account the reflectivity, emission capacity, 
colour and transparency of the objects, and can take into account more 
than one bounce.

So, one bounce radiosity where all objects are white will get you 
essentially the same result as using ambient occlusion.
Ambient occlusion is essentially a subset of radiosity.

Post a reply to this message