> One comment, one suggestion and two questions. First the comment, the
> face on the cat looks a little strange.

It is looking up at the sign. Modelling a face using blobs isn't that easy,
but on a higher res the viewing direction of the cat can be discerned more
easily. Since the cat *does* look like a cat, I wanted to improve on the
overall image before getting stuck with it...

> I suppose this could be accomplished with a bump map or some averaged
> normals or something, but I was trying to think of a better way to do
> it. There seems to be a general sentiment running that bump-mapping is
> cheating somehow, and I don't like using normals to simulate actual
> shapes (personal preference, probably completely unfounded). So I
> wondered...would it be possible to use Surcoat to do the general snow
> layer with a good blob factor, then to do it again on top of that and
> shrink the blobs to a really tiny size so that they don't join at all,
> but rather look like little crystals sitting on top of the snow layer?

Yes, that is possible. You'd just have to do a two-pass with Surcoat: one to
collect the initial, rough data, and then a second pass onto the blob. You'd
have to either render an orthographic view of the blob (to speed up the
sampling of the second pass) or just shoot samples at the entire boundary
box, but that shouldn't be a too big problem, right? ;-)

> Second question: how does Surcoat handle the snow layers?

Actually, all samples, along with their respective surface normal, are
simply saved into a file. A simple loop can then parse through the file and
while the File is defined, like this:

#fopen File Surcoat_Data read
#while (defined(File))
#read (File, Sample_Position, Sample_Normal)
//do some stuff
#end
#fclose File

I've prepared a simple set of macros for visualization, e.g. placing objects
at the samples' positions, or create a simple, spherical blob, or just place
spheres with a given radius. It should be easy to write your own
visualization macro, as the read-in (as seen above) is quite simple.

For the snow, I've written another set of macros which actually trace() the
surroundings of a sample to add some data like possible height, closest wall
and such, to generate heights for the snow. Instead of spheres, I use
cylinders in a blob to visualize the samples (along with the added data).
The processing of the original data into an extended set is of course more
scripting-biased, but that's the way it always is, right?

> Lastly, the suggestion...it seems likt there's a high "blue factor" in
> your snow. With such strong shadows I think it would be a tad more
> realistic if it were more blindingly white. Maybe it's that "artistic
> license" again...just a thought.

Hm, I'll think about that. Need to inspect some more winter photographs. :-)

> Overall incredible work. You're in my HOF book now. :)

Thanks for the praise. :-)

-- 
"Tim Nikias v2.0"
Homepage: <http://www.nolights.de>

Post a reply to this message