|
|
 |
 |
|
 |
 |
|
 Anonymous is Roland Emmerich's
Shakespearean project set to be released in September of
2011.
The team at Uncharted Territory
produced some state-of-the-art vfx for this 'non-destruction'
film, a distinct change from Roland Emmerich's usual genre.
We spoke with Pieter Van Houte who was given the
daunting task of creating the water simulations on the film.
And let us tell you...there is a lot of water surrounding
London.
Here's what he had to say about working on
Anonymous.
|
|
|
Q: The entire post production pipeline at
Uncharted Territory was based on Fusion. Where do you
see the main benefits of that? Can you describe a
typical scenario that shows how Fusion saved the
day? | |
|
Well the main benefit for me is that I am a long time
Fusion user. Fusion pipeline => work for Pieter. :-)
It's just a great all-rounder. I never feel stuck in
it. More often than not there are several ways to tackle a
problem, and in the known compositing world, Fusion allows for
all of those. If you encounter a problem which requires
specific solutions not provided for as standard, there might
either be a plug-in for it, or there are ways to easily design
your own tools, even if you lack any programming skills. As
you will read further down...
At Uncharted Territory,
they share that philosophy. They allow for a bit of a 'rebel'
approach to processes which are elsewhere often tackled using
rigid, closed pipelines. Under Rony Soussan's supervision, we
were welcomed - within reason of course - to experiment,
explore and share off-beat solutions to common and not so
common challenges. That way, together, we were constantly
pushing the limits of compositing further and further, which
was very satisfying indeed!
|
|
|
Q: A lot of customized tools
and macros where developed. What was your favourite of
those? How does it work and how was it used? How did it
simplify or improve everyday's work? | |
|
I am going to have to blow my own trumpet here I'm
afraid...in Anonymous there are a large number of shots
featuring water. Since the film is set in London, that made
sense. All of the water was to be CG rendered though, and
water is still one of the trickiest effects to get right.
Everyone knows what water looks like, even intuitively, so
when you get that wrong, it immediately shows. So at a certain
point, the CG department could no longer keep up with the
sheer amount of shots needed. It took too long to set up the
water, test render it, change the animation a bit, re-render,
change the camera a bit, re-render, anyway, you get the gist.
So one day I sat down with one of the 3D artists
(Robert Freitag - a wonderful guy who was known as possibly
the only person in VFX never to have seen Terminator 2 -
that's fixed now) and took a look at how the water was
generated. He showed me a process which is called the
Tessendorf method. I looked at the patterns it was based on
and thought 'hang on - I could possibly generate this in
Fusion...'
And that's what I did. I took a number of
Fast Noises, animated them, linked them all up using
expressions and making use of Fusion's ROIDS system built a
little 3D system which allowed me to test different settings
really quickly - near real time in some cases, and once I was
done applied those settings to a larger water surface with a
mouse click. It worked really, really well. The icing on the
cake was from eyeon's Robert Zeltsch, who added a little trick
using bump mapping and a bit of math which allowed me to
change even the sharpness of the waves, on top of their
overall shape. Everything controlled through a custom set of
sliders giving the user a nice, neat interface for quick
results.
So all of a sudden there we were, with
everything in one scene: the water, the matching 3D cameras,
and the green screen footage. We even got the reflections in
the water right. No going back and forth, in and out of the
software, and as such we kept to the deadline.
What
was great is that I was able to keep it simple. Took me less
than half an hour to explain the whole tool to other artists,
who then did stuff with it which I hadn't even thought of!
Another personal favourite on the show was a Macro
made by Gregory Chalenko. Simply said, it mimics 'onion
skinning' in the 3D view. So you can see the positions of
objects you are animating over time, at the same time as their
positions in previous (and next) frames. A feature not present
in a standard install of Fusion. There is a sequence in the
film depicting an early form of tennis. The sequence was shot
without the tennis ball, as the game was obviously scripted
and you'd find it difficult to find fine actors who are at the
same time stellar tennis players. The ball needed to be made
and animated in CG. Again, I never had to leave Fusion for any
of that. It must be said that I find Fusion's spline editor to
be one of the best available in any application. It's fast,
unbloated but at the same time very powerful. It's actually a
great tool to animate in. When I saw that tennis sequence I
knew I wanted it. I have been trained as a traditional
animator and somehow it was terrifically pleasing to be able
to go back to animating a bouncing ball. One of the interns on
the film, Caroline Weidenhiller, assisted me in setting up the
shots, using Fusion tools to rebuild the set in 3D so we could
do shadows, then I animated and rendered them all in Fusion.
Robert Zeltsch, 'Mister Icing On Cake', did a cool setup here
as well, faking very convincing soft shadows in 3D space,
without using shadows at all.
Being able to do all
this custom stuff obviously simplifies things immensely. I
think what really makes the difference is how easy it is to
share these things between artists. You can literally copy
something from an existing setup you think is useful for
someone else, drop it in a corner of your interface, and
instantly it shows up on your colleague's screen. Copy and
paste it into something else and you're good to go. Here's
hoping eyeon continues developing stuff like this even
further, with more visual feedback for how tools work and
interact, how they connect to each other, so the method itself
becomes self-explanatory.
|
|
| Q: There were a tremendous
number of CG water shots in Anonymous. How did you do
it? | |
|
The most important part
of the setup are the 4 creator nodes on the left. Those
together create the noise pattern of which you see a fragment
in the left viewer. With water it's all in the motion. Even
more than in the look, which is largely (if not completely)
determined by its surroundings and environment. Get the motion
of the water wrong and it won't look like water. It will look
like mud, or cloth, or something which simply feels fake.
Getting that motion right was the most time consuming
part of the setup, which at first consisted simply of feeding
the resulting noise pattern into a single displaced image
plane. I studied noises generated by specialized water
simulation software, but also looked at real water, or live
action footage shot as reference for the film. I eventually
found that dividing up the water in four different levels of
waves (from large to small) gave the most satisfying results,
as long as the ratio of movement between them made sense. For
example, small waves running faster than large ones made it
look like a thin sheet of water running over a surface instead
of water which is deep (but perhaps that could one day be used
for a different setup).
At a certain point I had
something that indeed had the feel of water. I then
backtracked through what I had done, and singled out the
controls which I thought were determinative for creating
different styles of water. Once I had those, I added a number
of sliders to one of the tools in my comp (in this case the
SetDomain node which selects the area of the water plane I
want to work on) which allow for speedy changes without
leaving that one single node. You can see those on the right
hand side. It also means you can save different kinds of water
settings out of one single node, for quick access later.
So there I had it; a base setup which I could easily
pass on to someone else to play with. The next step was
dealing with working on film and large bodies of water, so the
3D mesh risked to become quite large (=slow) in certain cases.
Using Fusion's Domain of Definition system and a few
expressions, I branched off a smaller section of the noise
into a much smaller mesh which you can then zoom into. That
way you didn't need all that geometry all the time, which was
pushed back to the very last step: once the movement was right
for your scene, branch off the large image plane and you will
have your large surface ready to render and integrate into
your scene!
|
|
| Q: There also was a
collaboration between uncharted and eyeon. How did you
like the feedback and support supplied by eyeon software
on the project? What special tasks where
solved? | |
|
Well there were two eyeon people with us in the studio.
Robert Zeltsch, who was there for the whole project and
developed some really nifty tools to do with 3D and deep
compositing, or anything else we might want (or think of just
for fun). Further down the line, we had the company of Stephen
Horwat, who for some reason I always imagined as being short,
dark, and skinny. For some reason I had a clear image of him
in my head before I ever met him, which doesn't often happen.
Turns out I was quite wrong. But I digress. It's great to have
specialists and actual developers by your side. Even for
something as simple as filing a bug report. Some of the stuff
which had been reported before, or was deemed irreproducible,
was explained in minutes and sometimes fixed before lunch
time!
I guess it also works the other way around.
Sometimes Fusion, because of its complexity, can be
frustrating to use. You sometimes get results you might not
expect, especially if you're working in floating point, or
dealing with Domain of Definition and so on, basically highly
technical stuff. So you see something which you don't get, you
think it's a bug, and then there's someone there to show you
why things are the way they are, and why they do make sense.
As such you gain more insight into something you think you are
already pretty familiar with.
With that exchange in
place, really cool new tools got developed, like one creating
volumetric fog within Fusion, alongside artists actually
working with them. For example, Robert would develop the
concept and the plugin for doing volume fog, and then we could
sit with him and rethink the interface from an artist's
perspective. To clarify; Robert would have features listed in
the interface in the order they would make sense in the code.
However, when an artist thinks 'I want to create fog', perhaps
the order of doing things, or thinking about them, is
different. So after a few times going back and forth, we ended
up with a tool which had the exact same features, but was ten
times as easy to use.
While you're doing all this, all
that really happens is that wish lists get even longer. It's
almost always a case of 'oh, if we could only take this
another step further'. I think it's great for us to have the
opportunity to ponder on these matters while having the
feeling our pondering might one day transform into a greater
set of tools. I, for one, am eager to find out what the next
steps will be!
|
|
|
|
|
|
|
|
|
 |
 |
|
 | |