DEEP SIX

Article -

by Paul D. Lewis

  Workflow Approaches  

 

When I first started doing work in 3D, no one talked about workflow. That was because there was only one workflow:

MODEL - ANIMATE - COLOR & LIGHT

This made sense at the time - but that was a long time ago. Today, this approach is still widely used in both Film and Games - it works well for neither.

A not-to-far-out example:

There once was a young and talented animator. He thought of a wonderful animation to do. He planned out the story, got other people excited and involved and they began work. Storyboards were drawn - scenes were thought up - characters were sketched in beautiful detail. Soon, each person took a character to work on and the race to create the best model began. The models became more and more beautiful - and complex. Their equipment reeled. 'More memory!' they shouted. But there was no budget. It was decided that these models were complete - so they moved on.

Each person began to setup Their character for animation - but things went badly from the beginning - joint placement was slow and binding was worse. The time to do anything increased - but they pressed on. 'We must work harder!' they stated to each other. As each character was rigged, it was then animated. Motion went quickly at first - but playback with the bound skin was far from real-time and the animation suffered.

Several people dropped out - the few who continued to work were delighted with the completion of Animation. Shading began quickly - but lighting was more tedious than before.... The first render test took 10 times longer than expected - and had significant artifacts and problems. This was more than anyone could take - the project stopped. 'When hardware is faster....' was all anyone said.

I've heard: 'That's just the way it is' and 'You need to throw more horsepower at that' more times than I can remember. There's a better way - and it works!

ANIMATE - COLOR & LIGHT - MODEL

Where are the bulk of the tools in a 3D package?

Right - in modeling.

Where are the least tools?

Right again - in animation.

Better to start where you are weak and inflexible and then move toward your strengths as things become tight. The basic scenario:

Start with the skeleton - go though all of your moves (or shots) and block them out. Then create the crudest place-holder geometry (primitives), parent it to the skeleton and place around the environment. If you can sell your character motion using ugly geometry - you have good motion. Because the scene is so simple - you can play it back at full speed several times and try combinations that give you the most bang for the buck.

Next is to light everything. Work to get the MOOD you want with lighting. Try colored light - heck, try two! Because the scene is still so simple - lighting changes and complex setups are easy to test out. Now is the time to make exclusion lists and focus specific lights on specific things in your scene. Get shadows all worked out - really.

Once the lighting is near FINAL (no, I'm not kidding) - it's time for shading/texturing. The trick here is to use the MINIMUM to sell visually what you're doing. Texturing place-holder geometry takes discipline - and is not for the squeamish. Stay the course. Complete as much texturing and atmospheric effects as is practical without final geometry.

Yes, it's time for geometry. Again, the focus is on doing the MINIMUM to get it to work. Adding resolution to a model with a texture already on it makes the modeling work more focused. Also, UV's are much more manageable if you already have the layout - and you're just sub-dividing it.

Volah! You completed the whole project!

Because the skeleton gets tried out under fire from ALL of the motion - it's virtually bug free FAST. This means programming can have the dimensions almost immediately - and they won't change. Happy programmers are friendly programmers.

Game scenario: The WHOLE game is ugly, but almost completely playable immediately.

Film scenario: The entire show can be rendered from the beginning - FAST - and sent to edit. More iterations, better looking - better work.

Lighting is the MOST neglected thing in 3D today - good light is still rare. This is a way to shine above the rest (pun intended).

Shading / Texturing and 3D paint are fun - but are time holes. Keeping it simple up front keeps your eyes on the goal - not just on the nice bump you tweaked for the last hour.

Radical changes in character structure are lethal to a production. Radical changes in character geometry are painful, but are possible in the final hour.

Remember to focus on what buys you the most:

Perception of quality:

Animation = 60%

Color & Light = 30%

Geometry = 10%

Don't get caught spending 50% of your time budget on something that only gives you 10% (MAX) of your visual impact. A beautiful model by itself is like a picture of Michelangelo's David. If you have a great model that moves badly - you don't get talked to. If you have an ugly model that moves well - you get told: 'This is great - when are you going to fix the model?'

When you have both - you get paid to do it again. ;-)

 

© 1996-2007 Deep Six L.L.C.
All rights reserved. Reproduction in whole or in part without permission is prohibited.