TITLE: Cartoon Bears
NAME: Guillermo Sanz Romero
EMAIL: famrom@ran.es
WEBPAGE: http://www.geocities.com/SiliconValley/4774/index.html
TOPIC: Childhood
COPYRIGHT: I SUBMIT TO THE STANDARD RAYTRACING COMPETITION COPYRIGHT.
JPGFILE: gr_carto.jpg
ZIPFILE: gr_carto.zip
RENDERER USED: 
    POVRay 3.02.msdos.wat-cwa

TOOLS USED: 
    DOS Edit, FrGen 1.4 and PP7

RENDER TIME: 
    16m 25s

HARDWARE USED: 
    PC iP55-166MHz(MMX)


IMAGE DESCRIPTION: 


How many cartoons have you seen in your childhood?
Well, here is another, but this time it have been raytraced.


DESCRIPTION OF HOW THIS IMAGE WAS CREATED: 


The topic Childhood was not appealing for me from an typical point of view.
But after some "random" testing of textures, I get the idea: maybe I can do
cartoons with POVRay. I was bored of talking with people, always remembering
their special software to create cartoons, and that POV can not.

Many texture and lighting tests were necesary, but finally I think I got a
"cartoonish" scene. The camera also is "retouched", to improve the effect.
This scene took more time for tech things that for the rest, and it rendered
quickly, luckly, otherwise I would abandon the idea during the tests.

I know that the scene is not one of the better, but... hey!, I was putting
the efforts in the tech side, not in the artistic side (and for cartoons,
IMHO the scene is not poor, just "cartoonish", a photorrealistic cartoon is
not a cartoon). It tells a story, the bears happily playing in the forest
(the start or the end of the child tale, you decide), and I feel OK with it.

If you only want to know the cartoon tricks for POV, here are the results of
the tests: use a special texture with low brillance (F_Cartoon9 is the better
at this moment). Put one light with shadows, the key light. Then add one or
more to get the color zones of cartoons, this time without shadows. These new
lights must be displaced from the key one. With two light (key plus one), you
will get a good looking cartoons, with three defined zones of color. The
amount of light must be around 100%, do no over expose the objects (I used
two lights, each with 500f intensity). The camera should have a normal
statement, but do not force the effect, the objetive is to distort a bit the
image, so it looks like hand drawn.

I recommend you to read the source code and visit my homepage, there I will
publish new improvements in the technique. Hope some day I can get cartoons
with lines without recoding POV (the border lines of the objects).

To avoid copyright problems, I created my own characters, three bears (after
all, some child tales have bears in them). The setting was easy, their house
in the middle of the forest.

All the objects were build with the "cartoonish" look in mind. They are soft,
rounded, maybe even deformed. See the chimney or the walls, for example.

The bears are blob based, first I make one, and the other two are variations
of the first. I know that the modelling can be improved, for example the
shoulders, or make them more "flexible", so they can be positioned easily.

The house is based in spheres, yes, portions of spheres, so the walls do not
look flat. The window and roof too. The door is a bit more normal (boxes).

The trees are generated with blobs, one for the trunk and another for the
tops. They look poor without the cartoon "tricks".

The grass are spheres and cones with some loops to place random copies. The
stones are the same object, but rotated to look different. It is generated
with FrGen, an really interesting program that randomices meshes.

All the objects were coded by hand except the stones, maybe I should have
used a NURBS or similar modeler for the bears (but the "new things counter"
for this two months was full). The text and jpeg conversion were done with
Picture Publisher 7.