Physical vs Motivated lighting
One of the most common discussion in lighting pipelines is the bias technology should have towards either
Physical or Motivated driven lighting. The challenge is much more complex than it may sound to an untrained ear.
Both solutions have critical advantages. As well as disadvantages.
Based on my experience,
motivated lighting is a key factor in both full CG and live-action shots.
Even
though an audience or a production may not always relate directly to it, effective driving
lighting cues are always
a key factor in narration, composition and emotional response, independently on the medium.
As the "Suspension of Disbelief" rule prooves,
the human mind is biased to respond to abstract cues to a level beyond physical accuracy. Willing or not,
this applies to lighting as well.
Tools have to be built to be strong enough in maintainin realistic continuity and credibility... but with a motivated flair.
---------------------------------------------------------------------------------------------------------------------------------------------------------------
In the article
State of the Art from
Cinefex issue #100,
which sported a fantastic interview to some of the leading minds in
the visual effects world, great care was put in highlighting that one
of the biggest challenges in lighting cycles was still the ability to
give TDs a complete set of efficient ligting tools.
CineFex: As far as digital technology has come,
what limitations does it still have? What tools do you wish you had in
that toolbox?
JOHN KNOLL: Nothing is ever fast enough.
Our render times have remained fairly constant since The Abyss!
They're always at the pain threshold. If you look at how long it took
to render one frame on The Abyss and what it takes to
render a typical frame on the new Star Wars, they are
fairly comparable. Of course, that's because we're doing far more
complicated renders - denser, wiith more sophisticated shading models.
If we had tried to render that back in 1989, it would have taken a month
to do what we're doing in an hour now. But people are always trying to
figure out a way to make the process faster. Faster is always better.
JOHN DYKSTRA: It's still really tough to do
random stuff in the computer. The computer doesn't like to do random
stuff; it likes to do organized stuff. So that's a limitation - even
when it's done really well. I was astonished at the work in Pearl
Harbor. I thought the explosions, fire and smoke were
phenomenal in that movie, some of the best-looking procedural stuff I've
seen. But it is still really hard to make that kind of
random stuff - water, smoke, clouds, fire. Those things are a huge
challenge for digital imaging.
ALEX FUNKE: There's also no good
digital lighting program. Software designers have never sat down with
real cinematographers and said, "OK, let's see how you do this and let's
design a program that actually works the way the cinematographer
lights." We're way behind on that.
ERIK NASH: Given the complex nature of CG
lighting and how non-intuitive and non-interactive it is, I don't know
how the artists do what they do. Artists who really understand how
digital lighting works do remarkable things with tools that don't mimic
real-world lighting tools in any way, shape or form! It's liberating,
in that you can do things you could only dream of doing in the physical
world; but the non-interactivity of it, to me, would be very
frustrating. That's one aspect of it that still seems incredibly arcane
- how slow it is to see what it is you're really doing when you're
lighting a CG object.
RICHARD HOLLANDER: That's what we need - a way
to realistically light environments and get results that don't take
40,000 years to render. It seems to me that if I was to put a synthetic
creature sitting next to you in this room, since the lighting is not
changing for all of our shots, no matter where my camera is, there
shouldn't have to be too much human intervention on the lighting of that
creature. We are striving for that. We want to understand this room
from a light-sourcing POV - and we want that to happen fast. That's the
next layer - to light 90 percent semi automatically, then tweak the
rest.
CineFex: It seems ilke lighting, since it is
physics-based, ought to be something that could be proceduralized fairly
easily.
ERIK NASH: Yes, it can all be quantified.
The problem is just the complexity of the calculations, the number of
calculations necessary to mimic what happens in the real world. All of
that is so staggering, to be able to run that simulation in anything
approaching real-time is just currently beyond the realm of possibility.
But, as computer power grows, interactivity will improve in leaps and
bounds.
HENRY LaBOUNTA: The past couple of years at
Siggraph, it seems like more than half the presentations are about
real-time issues. The stuff Nvidia is doing with Gelato is brilliant.
It's a whole new package they've come out with that is basically
intended to do RenderMan-style rendering in real time.
PHIL TIPPETT: That would be ideal - to be
able to go in and manipulate a thing with all of the lights on it, in
real time. When I was doing stop-motion animation, I found that
lighting had everything to do with how you moved the character. And
when you're dealing with a fireframe thing, you don't have that. You're
just dealing with movement in the abstract until you see the thing
rendered out.
RANDY COOK: That's the biggest disadvantage in
computer animation, versus stop-motion or some of the older techniques -
you don't see what the light is doing as you are animating. On LOTR
we tried to get our lighting setups as soon as we could and incorporate
them into our work. We also suggested lighting positions, so that our
character's face would 'find the light' at the right moment. But most
animators today don't get to do that. They're not allowed access to
that part of the the toolbox, which is unfortunate.
ROB COLEMAN: I think we're going to have that
real-time rendering in the next 3 to 5 years. It may not have every
level of specular highlights, or subsurface scattering or all those
beautiful things that they do to make realistic skin these days; but it
would certainly be good enough for me to judge the animation at a level
beyond what we have now.
---------------------------------------------------------------------------------------------------------------------------------------------------------------