Introduction
This is my project website for the Final CS563 Project.
Here you will find details and progress of the Final project assignment
in
CS563: Advanced Computer Graphics
This was an attempt to get realistic skin rendering. Unfortunately,
there is no final image, just a series of test screenshots.
This is the model I decided on working with, I named him Ming.... :
Here he is rendered, floating in the now famous Cornell box. (Hey at
least there is some color bleeding on his face)
Next, I set about trying to get participating media working... After
some hours of work I started to see something that looked like a
spherical cloud of smoke:
After gaining *some* confidence that I was seeing some good results, I
then moved on to try and render a smoke box...
As can be seen in the picture, there is an issue with shadows. Since the
photons bouncing off surfaces end up hitting the
'volume', they never deposit themselves onto the ground, so this false
shadow appears.
There is no shortage of knobs to turn. scattering, absorbsion, scaling,
etc, etc.... almost too many variables to tweak in order to get
something decent looking.
Next, after playing with different parameters.
The above pics were using an isotropic scattering pattern. Bascically,
photons were able to bounce off in any direction once they entered the
volume and scattered.
A typical pattern looks like this for photons hitting a sphere:
I then moved on and added in the Henyey-Greenstein function for
scattering. Below are a few patterns with different G value. For
testing purposes I set the photon to
all enter the medium at the same point, and then injected a few thousand
photons to get a pattern. You can see how the pattern gets funneled
tighter as G approaches 1.
G = 0.1
G=0.99
Next, I tried to shoot photons straight through a sphere, just to see If
I could see a column of bright light:
Finally, I adjusted some more parameters and fixed a few more bugs, and
ended up with this. This illustrated the light moving
closer to a flat blue partition. As the light gets closer, more white
bleeds through. Also, the edges of the participating medium get darker
as the light is focused more on the center of the wall.
This is probably closest to a final image, since I still don't have
everything working... and I've actually taken a step back and can't
reproduce the following image.
Its sort of a participating medium modelling a translucent Blue piece of
glass.