TITLE: Little red riding dragon
NAME: Gilles Tran
COUNTRY: France
EMAIL: tran@inapg.inra.fr
WEBPAGE: http://www.oyonale.com
TOPIC: Spirit of Asia
COPYRIGHT: I SUBMIT TO THE STANDARD RAYTRACING COMPETITION COPYRIGHT.
JPGFILE: gt_asia.jpg
RENDERER USED: 
    Povray 3.5 beta

TOOLS USED: 
    Poser 4, Amapi 4, 3DWIN, UVmapper, Picture Publisher

RENDER TIME: 
    5 days, 1 Gb RAM used, 250 Mb of scene file and associated objects

HARDWARE USED: 
    PIII 733, Windows NT4

IMAGE DESCRIPTION: 

Asia is such a vast continent, made up of so many cultures and peoples that to
say that it has a ?spirit? sounds a little colonial to me. I don?t think
there?s a spirit of Europe, Americas, Africa, or Oceania, so why Asia? That
said, for the average non-Asian person, Asia surely evokes a incredible number
of clich_s and stereotypes, and while we can?t take these at face value (like
these kids who asked Tibetan monks about their kung-fu skills) we must
recognise that these clich_s and stereotypes do exist and bear with them.

So what?s Asia today? A good place to start is Hong Kong, as seen through the
eyes of movie directors like Wong Kar Wa_ or Tsui Hark. WKW was actually the
main inspiration for this picture, as he symbolises, through his work, how
Asian cultures are now facing / assimilating / reinventing other cultures
(European, American, Muslim etc.). In ?Chunking Express?, Tony Leung and Faye
Wong have a mute meeting with Faye?s cover of the Cranberries hit ?Dreams? in
the background. In ?Happy together?, Tony Leung and his boyfriend end up in
Buenos Aires. In ? In the mood for love?, Tony Leung and Maggie Cheung split up
with Nat King Cole?s ?Aquellos Ojos verdes? in the background, and Tony finds
himself looking for her in Cambodia. In this IRTC image, this is not Chunking,
but a variation on Nathan Road and Kowloon, trying to express my own Asian
clich_: Asia is changing, and changing fast. And yes, the image on the TV set
was shown on CNN on Sept 11, 2001, and the hour is 23:10, Hong-Kong time. It?s
a small world.



DESCRIPTION OF HOW THIS IMAGE WAS CREATED: 

A great amount of time was dedicated to research: photographs, ideographs,
logotypes, signs, patterns. There?s a large number of image maps used in this
picture. Some were created from scratch (I ?wrote? Chinese by pasting Unicode
characters from web pages into Word), other were derived from photographs and
heavily processed with Picture Publisher. This meant a lot of 2D work, as there
are more than 40 different signs (and many other maps). Here are some links to
pictures I used:

http://www.la-grange.net/photos/2001/05/Hong-Kong/index.html
http://larve.net/people/hugo/pictures/2001/04/28/mongkok
http://www.china-on-site.com/gallery/index.html

The larger characters are Poser/DAZ ones (Michael, Victoria 2, Millennium kids,
various morphs, props and clothes packs, wedge hair, gel cut hair, Hollywood
hair, clock, sunglasses and related textures). The smaller ones are just bitmap
?cardboard cutouts? also made in Poser, posed and lighted to make them fit in
the picture. Cars are anonymous Internet ones or models by Miguel de Espona
(available at Turbosquid). The tricycle is the work of "Abraham" (Renderosity).
The beer cans were created by Fabien Mosen. The portable phone is the work of
Sebastien Loss (Hamapatch model). 

Things I?ve done: Amapi was used to model the bowl and the chopsticks. Regular
CSG was used for the bus, the restaurants, the buildings, the signs, some other
street elements, the Pepsi machine, the table, the TV set etc. The building on
the right is inspired by a real-life one which is actually in Shangha_ (it
lacks the air-conditioning). The noodle restaurant on the right was also
modelled after a real one in Cambridge (see
http://www.winfield58.co.uk/yingwah/yingwah.htm). Other CSG was lifted/adapted
from previous images of mine (the fire hydrant, a couple of shop windows).

There are over a 100 light sources, and a good number of light groups (whenever
I wanted a particular highlight). The reflections are too dark because I had to
lower the max_trace_level and raise adc_bailout otherwise both the rendering
time and the memory use (1 Gb...) were unbearable. It doesn?t use radiosity.

Different techniques were used to create the neon signs: media, plain ambient 1
image maps, height fields with transparent textures. I also played with DF3s
but these signs didn?t make it in the final image. In fact, a more realistic
neon tube was made but it required a too high max_trace_level. Several macros
were used to create the sign frames and support beams. All signs emit light. A
good part of the time was spent creating the signs and positioning them.

Rounded isosurface boxes were used whenever possible. The sidewalk is made with
rounded boxes with granite deformation (it?s still too clean and even). Rounded
iso boxes are the greatest Povray 3.5 addition since sliced bread (which was
added in Povray 3.1).

Many textures are mixtures of image maps and procedural ones. Dirtying,
particularly, is done in some cases by using the same texture with different
diffuse value within a texture_map. For hair textures, I?m presently favouring
the newer image_pattern with texture_map technique over the use of alpha maps.
Two good reasons: 1) the image_pattern image can be a relatively small jpeg and
2) transparent parts can be really transparent (no highlights).

Problems: many of them. It?s not a very good picture. I got somehow too
ambitious about this one and the result is less satisfying than it should be.
Absence of radiosity makes it not too realistic. This may not be a real issue
but still? The image lacks a real topic (and the general setup is too close to
my IRTC ?Wet bird? image of April 2000). The signs should have been the topic,
not the characters. The left part should have been brighter and the right part
darker to make the signs and the tricycle stand out better. I got to see this
before the final render, but I didn?t correct it enough. The little Poser
people just look like Poser people, i.e. poserish. I ran out of time to make
more of them, and to make them nicer. There really should be more people in the
street. There are also many little things that would have needed more
attention. Blurred reflection should have been used on the ground, but memory
use and render time decided otherwise.

I?m not releasing the code because it?s a big mess (2500 lines or so, not
including the meshes), requires 100 or so image maps and uses copyrighted
material. 

However, some of it will be available in the resources section of
www.oyonale.com somewhere in 2002: the bus, the noodle restaurant, the building
on the right, the neon macros and a few other things, when I have cleaned up
the code.




