Visual Effects Artist, Software Developer and Filmmaking Enthusiast
|
|
 |
Bin Can Can - An animation by Steve Agland |
;)
A short animation by Steve Agland Running Time: 1':20" The Dancing Bins are the stars of an 3D computer-generated animation (now called "Bin Can Can") that I made as my major project for a subject at uni called, funnily enough, 3D Computer Animation. I was taught by a very clever and helpful man named Fillip Edson. In the four months or so the subject lasted I was able to learn enough of Houdini (a tiny fraction) to turn a hazy vision in my mind into a sixty-six-second three-quarter PAL-resolution CD-quality soundtracked almost-photorealistic video.
;)
The line-dancing bins of Berryman St have hit the big time. Suddenly, deciding they wanted to be famous, they went off and got themselves:
;)
I got the idea for this one day when I was walking home from my late-night Computer Graphics Rendering Techniques class wracking my brain for an idea for the animation I was going to do. It's rare for an idea to come at the right time isn't it, but this one did. I was walking along, at night, and the street was lined with Ottos, because it was garbage night. I also happened to be listening to my Classic Classics CD, which is like a collection of all the popular well-known "classical" pieces of music, and the track playing was Offenbach's Can Can - which is such a lively tune. Natually I thought of dancing bins. I had ideas like this all the time - I really wanted to animate something to music, its so cool.. like Fantasia. But the difference between this idea and most other ideas was that this idea actually seemed feasable. The two things I wanted most were good music-animation synchronisation and photorealism. Photorealism is extremely hard to acheive, but Otto bins are the perfect thing. Not too hard to model accurately; not exactly difficult to texture; and fairly easy to animate (compared to, say, a Tyrannosaurus). So I expanded the idea, selected the most practical chunk of music (which I decided was 3:15 until the end), imagined what could be happening during each bit of the piece, and just stuck to the idea (except that I ended up doing it in daylight instead of night). It was, of course, a lot harder than I thought.
;)
As required for entry into various festivals. A short but sweet animation shot in a raw home-video style, Bin Can Can follows the increasingly elaborate antics of a group of suburban "wheely bins". Their awkward but meticulously choreographed dance is synchronized hilariously with Offenbach's familiar soundtrack.
;)
As required for entry into various festivals. The animation was produced almost entirely with SideEffects Software's Houdini - including the compositing and editing. The background plates are still photographs taken on an inexpensive digital camera. CGI bins are animated on top, dancing to the music. The shaky illusion of a handheld camera was achieved by cropping and panning the final composites using motion information originally derived from mouse input. The entire animation was produced on a standard home PC.
;)
As required for entry into various festivals. The goals of the production of this animation were simplicity, humour, photorealism and "slavery" to the music. Generally perceived as being mutually exclusive, the goals of simplicity and photorealism can in fact be complementary. A low (virtually non-existent) student production budget and a one-person crew meant that a raw and realistic look would be practically preferable to an elaborate, abstracted visual style. The plan then was to create an animation that appeared to be just what it was: an amateur home video. The audience would instead be captured by the funny and unexpected energy in the bins and, in contrast, the satisfyingly appropriate nature of the dance, synchronised to the music. The classic soundtrack is in fact the primary creative force in the production, and all the animation, "story" and editing was dictated by the music: accompanying, illustrating or celebrating it rather than being supported by it. Since the music will be so familiar to the audience, the animation is designed to satisfy their expectation of it, using clichés and sensitive timing. It is this combination of delightful surprise and musical satisfaction that is intended to produce the desired result of the animation: a smile on the face of the viewer.
;)
This is the text document I hurriedly wrote and handed in to my lecturer to accompany the animation project files.
___________________________________________________________________
|BIN DANCE - A street of dancing garbage bins |
`-------------------------------------By Steve Agland---98075207----'
Overview
---------------------------------------------------------------------
This minute-long animation follows the antics of ten CG suburban
garbage bins around a street represented by twelve static
photographic images.
Compositing is used to place the computer-generated bins into the
photographed background. Fixed cameras were carefully placed to
accuately line up the real scene and the modelled scene. Panning,
zooming and general wobbling are simulated with COPs. Lighting is
arranged to reflect that of the real street. Shadows are cast on the
ground and motion blur is added to add to the realism.
The soundtrack consists of a well-known segment of Offenbach's
Can-can, to which the actions of the bins is closely timed.
The Music
---------------------------------------------------------------------
The driving force of this animation is the lively soundtrack played
by the New York Philharmonic Orchestra and conducted by Leonard
Bernstein. This is the closing section of Offenbach's Can-Can from
Orpheus in the Underworld and contains the most well-known and
exciting part of the tune.
The music is broken into several distinct sections which last for 4
to 8 bars each. These sections are also used to divide up the
animation itself into a number of different "scenes".
The Models
---------------------------------------------------------------------
There were a number of models made for this animation, ranging in
complexity.
--The Road-- ("ground" object)
The road is probably the simplest model, consiting of a number of
grids (some slightly altered). This is used as a geometic
representation of the ground which assisted in matching the positions
of the digital cameras to the real ones, and also allows the casting
of shadows by the bins on the ground.
--The Ottos-- ("otto" object - and otto1, otto2, otto3, otto4)
One model was created of an Otto bin. This is the standard garbage
bin used in North Ryde for all generic garbage. Called Ottos after
their mannufacturer, variations of these are found all over Sydney
in a variety of different designs and colours.
The "otto" geo acts as a master object and is never actually seen.
This is where, however, all the geometry is created. In building
this model a balance was struck between life-like accuracy and
time restraints, as only so much detail would be seen in the final
animation.
The objects "otto1" etc use Object Merge SOPS to bring in geometry
from the master geo (such as wheels, frame, lid) and merge them with
transform SOPs, allowing for control of individual moving parts in
each otto clone, as well as applying different texture maps for
visual differentiation. These transformation SOPs are described
in the Animation section.
The numbered ottos can now be animated individually but if the
master model can still be altered, the changes filtering down into
the clones.
--The Recycling Bins-- ("recy" object - and "recy1" etc)
These bins are organsized in much the same way as the Otto bins. The
master recycling bin borrows much geometry from the master Otto as
their designs are very similar. There are small differences in the
wheels, however, and the recylcing bins have two smaller lids
instead of one. In addition, their textures are quite different.
--The Trash Can-- ("can" object - and "Can1")
This is a simpler model of a small cylindrical trash can. Only one
clone of this is used in the animation, unlike the ottos and
recycling bins, with four each. In addition to the transform SOPs
used to animate it, if has twisting and bending for added character.
--The Dumpster-- ("dumpster" object)
A very simple polygonal model with a very detailed texture map.
--The Rose-- ("rose" object)
Another simple object used in a brief shot where two bins attempt
a Tango.
The Particle Systems
---------------------------------------------------------------------
Two particle systems were created for this animation. The first, a
fireworks simulation, isn't used in the final cut as it was later
decided not to set the animation at nighttime.
The second is a swarm of flies which surrounds the dumpster. The
flies cannot pass through the dumpsters walls, nor can they stray
too far away without being killed. Unfortunately the flies are too
small to be seen easily in the rendered version.
The Animation
---------------------------------------------------------------------
In this animation it is assumed that these enlivened bins have the
physical power to move themselves around in any way they want, which
includes rolling, flipping, leaping, spinning etc.
--The Ottos and Recyling Bins--
Various channels were created for each of the visible bins which
allowed for flexible control of their actions. These included rocking,
bouncing, rolling and lid-flapping.
Rocking is achived using a MasterRock SOP, with animated rotation
channels. This SOP is not actually in the network at all. However,
other SOPs use the values in MasterRock to rotate the model the
right amount in the right directions using the correct pivot points.
This means animating the rocking of the bin is as easy are adjusting
one channel and the pivot points look after themselves, so the
corners of the bins never go through the ground, and when they rock
backwards, their wheels remain stationary while the frame rotates
around the axel.
Expressions like these made the animation of the bins a lot easier
and more flexible. It was simple to make a bin roll along and stop
while rocking momentarily forward due to its momentum.
At some points some channels would use others as input. For exmaple,
a lid might get bounced by the persitant bouncing of the entire bin
or a Can-Can line recycling bin would follow the actions of the lead
dancer, with its timing slightly off to make it seem less robotic.
A null object called "Beat" was ofter used to provide a generic
bouncing channel that was used by various dancing bins in the
scene.
--The Trashcan--
This was animated in much the same was as the Ottos. It did have
extra Twist SOPs to allow bending and twisting as it danced
towards the end of the animation.
The calculation of the pivot points for this cylindrical can as it
rocked was more complex than for the other bins (the dumpster was
the simplest) and required an expression ("cylpivot in
Aliases/Variables") in order to find it. This means that the no
matter which direction the bin is leaning it will always pivot on
the part of the bottom rim which is touching the ground.
--The Cameras--
This pseudo-animation is achieved with a crop COP. The composite
of the rendered bins on the static backgrounds was stale and did
not look very real. With a crop COP is was possible to trim the
output image and then scale it back to the output resolution.
By animating the values in this COP it was possible to simulate
limited panning and zooming. As well as controlling the pan and
zoom an expression was created that simulates the waving and
wobbling associated with a hand-held camera. This added a lot
more realism to the animation and made it appear less flat.
(see programming function "camwave")
The Materials
---------------------------------------------------------------------
The materials in this animation are relatively simple in their
implementation. The texture maps, however, are quite intricate and
were created in Photoshop using actual photos of bins and dumpsters.
One special material is the ShadowMatte material used on the "ground"
object which makes the ground invisible except when a shadow is cast
upon it. By carefully aligning the ground geometry with the ground
in the photos it was possible to have the bins casting shadows over
gutters and up driveways.
The Composites
---------------------------------------------------------------------
Composites were used for three purposes: creating texture maps,
layering CG animation over live photos and editing the footage from
the twelve cameras together.
For a number of bin textures stickers, decals and paint were put over
the top of the scratched plastics to make a variety of different "bin
skins" using only a few image files with different combinations.
The "Camera0" etc COP networks are where the various cameras are
rendered and compositied over their matching background images.
Sometimes a "brightness" COP was needed to match the illumination
in the photos and the CG.
The "video_edit" COP network is where the outputs from the camera
composites are edited together using crossfades and sequence SOPs
into the final animation. The editing, like the animation, was
influenced by the timing of the music.
The "Output" COP network is where the "video footage" and the credits
are put together. This is the network that is rendered as the
complete animation.
The Outputs
---------------------------------------------------------------------
The important outputs in this animation are the twelve camera
renderings, one for each camera in the scene, that get used in the
compositing COP networks, and the "Output" output, which is, of
course, the COP rendering which will output the entire animation,
credits and all, from beginning to end. "Override default Res" must
be checked for this output and the desried resolution chosen.
|