TITLE: That's the end of that.
NAME: Richard Sutherland
COUNTRY: United States
EMAIL: rich@brickbots.com
WEBPAGE: http://supermegamulti.com
TOPIC: The end of...
COPYRIGHT: I SUBMIT TO THE STANDARD RAYTRACING COMPETITION COPYRIGHT.
MPGFILE: rpaz_teo.mpg
RENDERER USED: 
    Maya 5.0

TOOLS USED: 
    Maya, Photoshop, After Effects, Audition (Cool Edit), Premier,
TMPGenc, LenseCare (AE Plugin), ReelFX:MotionBlur (AE Plugin)

CREATION TIME: 
    Rendering 28hours (~920 frames, ~1 30 per frame, 5 min per scene
background, 8 scenes), Compositing: 30minutes (~2s per frame to composite)

HARDWARE USED: 
    P4 2.4ghz, 2gb Ram

ANIMATION DESCRIPTION: 


Man + Woman + The End Of 'That' = Funny Animation

I've been looking forward to entering the IRTC animation contest for a while.  I
was finally ready for this one, and then the topic was released.  It's a fine
topic, and will probably bring in some great entries, but it was really counter
to the style I like to work in.  I like funny, cartoony type work.  Then it hit
me, how I could have the end of something be funny!  Aha!  Hopefully people
will see how it fits the theme.


VIEWING RECOMMENDATIONS: 

This animation has sound. I recommend the Quicktime viewer, not because I am
biased, Microsoft Media player sometimes displays this animation darker than it
should (actually any MPEG1 file).  Quicktime always plays MPEG1 correctly. For
reference it should look like the poster image, bright.

For MS Media Player: I have found that if you select the file in a folder with
web content enabled (the default, it shows a preview in the left hand side)
then play it through Media Player (which also has generated the small preview
on the side) it has a much, much higher chance of playing correctly (bright
with the right contrast).  I have seen this on several machines and lots of
MPG's but never figured out how to avoid it.  If you know anything about this,
please drop me a line.  


DESCRIPTION OF HOW THIS ANIMATION WAS CREATED: 


This description will be a bit sparse, just enough to help the judges judge. 
Mainly because I gotta get my entry in.  There is a good chance that by the
time you read this I will have all my 'making of' materials posted on my
website at http://supermegamulti.com/teot, most or all of the pre-production
art, much better descriptions of the processes described below and
mini-tutorials should be there if you are interested.  Also full size (720x216)
versions with more sophisticated codecs.

This was my first attempt at a full production pipeline...

Concept -> Script -> Dialog Recording -> Storyboards ->
Modeling/Rigging/Texturing/Environments/Lighting -> Animatics and Layout ->
Animation -> Rendering -> Compositing/Effects

It all looks nice and linear on paper.. I went back and forth quite a bit. 
Probably my inexperience.  Following are some technical notes on each computer
related step, and then a few notes about scene 6.

Modeling/Rigging/Texturing:

Characters - Not much to say, polygon modeling with joint/skin rigging.  The
faces were done with a combination of clusters (groups of translatable
vertices) and blend-shapes (morph targets).  The skin and hair materials both
have a special setup that links their incandescence (self-illumination) with
the face normals.  Faces that are almost perpendicular to the camera have a
glow to them.  This is meant to soften things up and to simulate to some extent
the light that refracts off of the small hairs that cover our bodies.  This
effect is much more pronounced on both characters heads.  A side benefit is
that is helps to define the character silhouette so a backlite is not needed,
and I did not use one.

Background - I wanted a simple background, so I could actually finish the
project.  I also wanted everything to be a bit more primitive than reality, so
I used geometric shapes as a starting point for the trees and just deformed
them.  There are some simple house size boulders and a large structure I
imagined was a gym in the distance.  All very simple. The textures are all
procedural (fractal and noise), with several layers.  I love using one
procedure to control the blending of two others.  This gives both small scale
texture and large scale changes.  You can see this on the grass and concrete
where some large patches look different than others.

Sky - The sky is a sphere with a procedural texture.  I used a gradient with the
cloud color to make them get fainter the closer to the horizon they are.  This
helps give the illusion of a much larger space.  At the same time, I think the
background covers most of that up, so it might have been overkill.  

Lighting -  I love to simulate sunlight, but it is so challenging.  Clearly real
global illumination and skylighting would just take too long to render.  I have
had a lot of success with a script called GI_Joe, which basically just sets up
a whole bunch of lights arranged in a sphere.  I used this script with very
high values for the individual shadow map resolution.  It took longer to render
(~30 seconds a frame just for the shadow maps) but the subtle shadows are nice
and the illumination has depth even in the sunlight shadow areas.  Since there
was such a large amount of 'space' consumed by the background there was no way
I could get the subtlety on the characters that I wanted if I rendered it all
at once.  The scale of the shadow maps would be all wrong.  Rendering the
background and foreground separately meant that the shadow maps only had to
cover the bench area for the foreground shots so I could get the effect I
wanted with reasonable sized shadow maps.  The skylight is pure white, the
sunlight is yellow.  The sunlight shadows are not black, but are more purpleish
to compliment the yellow sun.

Animation:
I blocked out all the shots to match the storyboard and then started adjusting
everything until it flowed right.  Lots of playblast (openGL viewports saved to
disk for a quick render) renders cut together in premier to see that everything
worked okay.  Then I started scene by scene doing the animation.  Whenever I
would advance one scene, I would cut it again with all the rest to continue to
check for continuity and flow.  I wanted to refine things further, but I ran
out of time.  I got all the big changes that I wanted done, but I could have
worked on the motion curves some more.  The woman's hair is dynamic, but I had
to hand tune it in parts with the help of a couple of cluster deformers to get
it to lay right when she looks down/away in scene 7 (the crash scene).

Rendering:
There was a single frame sky render and background render for each shot.  I
generated a depth map (z-map) for the background pass.  Then the foreground,
basically the two characters, the bench and the concrete pad the bench is on,
were rendered as a series of TGA stills.

Compositing/Effects:
These elements were brought into After Effects for compositing.  I applied a DOF
effect (the LenseCare plugin, which is great and inexpensive BTW) for the
background and sky.  The foreground had the motion blur applied (the ReelFX
plugin, also fantastic and inexpensive).  I could have done these in the
renderer, but I never would have been able to render the whole thing.  I
suspect it would have take upwards of 10 minutes per frame as I would need to
render everything for each frame and the DOF takes a long time.  Having
everything separated also gave me the opportunity to do
color/brightness/gamma/contrast correction on each element, turns out in this
case I did not need it.  Oh and doing the DOF and motion blur in post let me
fine tune it quickly.  A little more DOF, little less motion blur, all without
hours and hours of re-rendering.  

On top of the actual laying of the pieces and the DOF and motion blur I applied
two effects layers.  The first was a very blurred version of the scene
composited in screen mode.  Screen just lightens everything.. it is sort of an
addition of the luminance of the layers.  This has a very low opacity and gave
a soft sort of blown out look like a camera in a very bright environment. 
Everything just sort of glows.  The second pass was very much like the first,
but not nearly as blurred and selecting just the brightest colors.  This makes
any bright spot (like the man's shirt, or the woman's arms/hands at some
points) bloom.  Since it is not blurred much the bloom is very localized.  It
is a nice set of effects and I like to use them a lot.  Finally I added my
watermark to the lower right.

Sound:
I suppose I should mention something about this.  Yep, that's my voice (the
man), and an old friend of mine voiced the woman.  I recorded it straight to my
computer with the best mic I have, which is not very good.  Only after I was
almost done did I learn one of my friends has access to a recording studio. 
Ahh well... next time I will have better quality sound.  All the effects (the
kids playing, water splash, bang, crash, birds chirping, wings flapping) are
from http://www.a1freesoundeffects.com/.  I recorded the 'thud, thud, thud'
myself by beating on the mic in various ways.  Some of the sounds were mixed in
Audition, others right in Premier while I was editing the video.  

Scene 6 - "Well!  I don't know where you went to college..." - Diagonal shot of
bench, man in foreground, woman in background

This scene was extra tough.  I wanted the man to be out of focus, then as he
sees what is going on and exclaims to come into focus.  All the other shots I
only rendered and DOF'd the background.  Since the characters were in the same
focal plane, they were just in focus.  For scene 6 I needed to render a
depthmap for the foreground and background in one pass and DOF the whole thing.
 So there was a sky pass, background pass, foreground pass, and then the
depthmap pass with both foreground and background.  Then I could animate the
focus on the LenseCare plugin.  Whew, all that and the effect is very subtle,
not sure anyone but me will notice it.

If you have read this far, I hope you enjoyed it.  Visit my website at
http://supermegamulti.com/teot for videos of the various stages and more
explanation and example frames.

