The first picture took about 20 minutes to render and the second one
about 1 hour.

Post a reply to this message

Attachments:
Download 'satyr1.jpg' (39 KB) Download 'satyr2.jpg' (39 KB)

Preview of image 'satyr1.jpg'

Preview of image 'satyr2.jpg'

Warp wrote:
>   The first picture took about 20 minutes to render and the second one
> about 1 hour.

The first looks like an area light but what is the lighting of each?  Is 
one more experimental than the other?

Charles

Post a reply to this message

Warp <war### [at] tagpovrayorg> wrote:
> The first picture took about 20 minutes to render and the second one
> about 1 hour.

I like the model.  It looks like the first one is just radiosity and the second
one has scattering media (with some reflection?).  Not really a tossup which
one looks better.  So here are some ideas, as if I know what I'm talking about:

First of all, some smoothing on the mesh should get rid of the artifacts.  Had
the same problem but it worked great for me.

Maybe increase the media samples to some outrageous number?  The arm picked up
the wrong colors on the highlights unless I increased the samples to 500 or so.
 It looks like that might be what's happening on the hair where it turns blue.

You'll notice that the twisty dodecahedron I posted has the largest area light I
can afford to render without much noise, and the hand has no self-shadowing on
the small features.  Seems to me this method can only work well with
translucency that falls off at no farther than the smallest length scale, if
that makes sense.  Maybe higher extinction?

I also had to add plenty of absorption and emission to get the overall coloring
right, but that adds some undesirable ambient light, clearly visible on the
arm.  I've posted an example with textures from both scenes in p.t.s-f.  It
renders in about fifteen minutes at 600x800 for me although it's probably about
linear in the number of area lights.  But increasing it to just 2x2 on the first
line helps the appearance a lot.

Despite all these points, it still has potential though, if you ask me.  I don't
even know that this hasn't been done ad nauseum since my pov-ing has been more
off than on for quite some time.  Just don't see it much though.

On another note, I got the crazy idea the other day that I knew enough math and
programming to attack the subsurface scattering problem once and for all.  So I
googled "bssrdf" and got pretty much all the way through the paper with a decent
understanding of what was going on, although I stared at Wikipedia for a while
trying to figure out the difference between luminance, radiance, irradiance,
etc.  Never even heard of a steradian before (and neither has my spell-checker
by the look of it).  But when it comes down to it, I hardly know where to
start.  Shame, since the media method doesn't exactly strike me as robust.  Oh,
well.

 - Ricky

Post a reply to this message

triple_r wrote:
> First of all, some smoothing on the mesh should get rid of the artifacts.

  It's the mesh self-shadowing problem. There's little I can do about
that (short of making the mesh shadowless, in which case it wouldn't
cast any shadow, obviously).

Post a reply to this message

Warp wrote:
> triple_r wrote:
>> First of all, some smoothing on the mesh should get rid of the artifacts.
> 
>   It's the mesh self-shadowing problem. There's little I can do about
> that (short of making the mesh shadowless, in which case it wouldn't
> cast any shadow, obviously).
If you subdivide the mesh before using it in POV-Ray it dramatically 
decreases the self-shadowing problem. In some cases it eliminates it if 
the mesh is well formed and subdivided.

FlyerX

Post a reply to this message

FlyerX <fly### [at] yahoocom> wrote:
> Warp wrote:
> > triple_r wrote:
> >> First of all, some smoothing on the mesh should get rid of the artifacts.
> >
> >   It's the mesh self-shadowing problem. There's little I can do about
> > that (short of making the mesh shadowless, in which case it wouldn't
> > cast any shadow, obviously).
> If you subdivide the mesh before using it in POV-Ray it dramatically
> decreases the self-shadowing problem. In some cases it eliminates it if
> the mesh is well formed and subdivided.
>
> FlyerX

Also, might want to mention the crease angle setting in PoseRay. :)

Post a reply to this message