Saturday, February 3, 2018

Cloudy With a Chance of Meatballs 2 Test #1

Cloudy 2 test from 2012.

It's best to focus on answering the big questions when you spend the time to do an animated piece. "Can this character carry our film?", "Does the design match the emotional tone of the film?", "How do we do this thing no one has done before?"

We had done previous tests with Barry and the Pickle before we made this test. Everyone was happy with the cute strawberry and the goofy pickle. The big question for the film was, "Will the predatory food be too scary / scary enough?" Finding that line was tricky, not just for this test but also for the film. From this and a subsequent test we discovered that it's the character's reactions to the threat that increase or decrease the scare. As long as we make Barry cute and Pickle goofy, we can get away with a lot.

This came together fairly quickly, four weeks in total, largely due to the fact that Barry and the Pickle were textured and rigged from a previous test. Omar Smith modeled all the characters and rigged the Cheespider, James Battersby textured the Cheespider, Chad Stewart animated and I rigged and textured Barry and Pickle and lit.

Emoji Movie Animation Test #1

A good example of a typical voice test. This was early 2016.

After we hit diminishing returns on a character sculpt, we try to have them perform, similar to a live action film. If you think this is your star, go ahead and have them act out a scene with some dialog. Lots of issues are spotted with a test like this, both good and bad.

SPA does not have a single design style that carries over to every film. That's a great way to avoid stagnation, but for development that means we have to custom craft every rig. For Gene, I spent the most time on the face rig ( he's 75% face ). Even though he looks simple, a lot of r&d goes into it.

Each iteration of a rig is not 100% complete. I get a working version in my and the animator's hands quickly. We assess, suggest changes/additions/etc. Then I go back for round after round until we're satisfied that the rig can perform well enough for this task. It's a very similar process to agile development, just on a very small scale with the stakeholders and customer rep are the same person. Development in a small team requires flexibility and objectivity.

Three artists worked on this over three weeks. The character design went through a lot of iterations before we landed on what you see above, and that's not counted in the three weeks.  Those iterations were months of work. Omar Smith sculpted Gene, Troy Saliba animated and I rigged, textured and lit.

Friday, February 2, 2018

Smurfs Animation Test #1

Better known as "The Cupcake Test" circa 2009. Our first hybrid test.

You'll notice that this Smurf is closer to the original Peyo designs and the 2017 film compared to the hybrid films in 2011 and 2013. As confusing as it sounds, this test came first and green-lit the live action films even though the look for those films is different.

We typically start with original material if any exists. That's the case for the Smurfs as it was for Popeye. A faithful representation is a great comparison tool because we can refer back to it rather than imagine how it might look in 3d.

This doesn't exactly nail the Peyo look, but time and budget sometimes win over asthetics. We picked up from this point and refined it further for the 2017 film, just as we would have if the decision was to go this design route for the first two films.

The dog was a PRO. The handlers would put a drink coaster on the ground and run through the performance once or twice, then pull the coaster for the take. If we needed another take, they told him to go to his mark, which he did almost to the exact spot. For position adjustments, they would say "move right/left/back/etc" and he would take one tiny paw step to his right/left/back/etc. They used a treat on a stick to get him to appear as if he was eating the cupcake, which we later replaced.

The whole test was done in about 20 weeks, including the live action shoot. Six SPA 3d artists and four SPI animators worked on the test. Omar Smith modeled Smurfy, James Battersby helped with the live action shoot, textured and lit, Ernie Rinard textured and lit, Sungwook Su textured and did fx. Chad Stewart, Spencer Cook and Kenn MacDonald animated, unfortunately I don't recall the names of the other two. I supervised the live action and cg, modeled, rigged, animated, textured and lit.

Cloudy With a Chance of Meatballs Animation Test #2

This piece is from 2006 or 2007 when Phil Lord and Chris Miller were on board to direct.

It was the most ambitious project we had attempted. Flint and Sam were in active design revisions, the design of the world only existed in Justin Thompson's head and two new directors were retooling the project. This test green-lit the film and led to further development where we created the look that you see in the feature film.

For the feature, the design of Flint and Sam changed, the look was refined and the animation improved. All of these things are expected in development. Production has hundreds of very talented artists and a long process ahead to create the highest quality possible.  Early development leans toward speed and low cost at the expense of quality. Quality is improved over time. The largest jump, both in cost and quality, is when production begins.

What's surprising to a lot of people is that 3d visual development does not make the assets for the film, nor should it be required to do so. If we get a design approval, then production only has to make the asset once. There's huge cost savings in having a large team with a very specific, unchanging goal. It's best to think of what we do as a concept or prototype similar to how an auto manufacturer creates a concept car with a team of designers before the production version is made.

There were six 3d artists who worked for about 20 weeks on this test. Takao Noguchi and Omar Smith modeled characters, Ernie Rinard and James Battersby modeled, textured, lit and did fx and Sungwook Su did fx. I supervised, modeled, rigged, textured, animated and lit. Justin Thompson was the production designer.

Surf's Up Animation Test #1

I think we did this in 2004 or 2005. This was the first large scale piece ( we call them sizzle pieces ) that the 3D Visual Development group had assembled.

We try to answer as many questions as possible if we'll be spending lots of time on a test. For this piece, no one had done a crashing ocean wave before, there were lots of rolling swells done at that time, but nothing that had a tube and crashed. We were also experimenting with sports-like cinematography.

The wave rig was a prototype that we handed to Imageworks, who in turn built a rig that suited their needs for the film. With only two of us animating on a handful of shots, we can make a very specific rig rather than having to make something very robust for scores of animators working on hundreds of shots. We're free to try things you'd never dare try to support in production.

It was suggested to us that we simulate the waves. As in: make a giant body of water, make a sloping ocean bottom, put 1000 fans and blow them over the surface thereby creating waves. We knew immediately that would be the most expensive, time consuming and inflexible method possible, so we set out to rig the wave as if it was just another character.

Since all of us have a production background we could put ourselves in the future animator's shoes. "I bet the director will ask to change the shape on the top of the wave." -or- "It would be great if we could see the wake and trail as we animated." -or- "The animator and director should see texture flowing over the wave to get a sense of how fast or slow the wave is moving." We could cobble together a lot of ideas very late in the process, sometimes days before delivery.

The directors brought the test and their outline to the executives. They showed the test first and the execs green-lit the film in that meeting.

There were five artists working for about 16 weeks ( including the wave rig prototype ) for the entire piece. Takao Noguchi modeled and textured Cody, Shawn McInerney animated and helped with the wave prototype, Ernie Rinard and James Battersby textured and lit the waves. I supervised, rigged Cody, made the wave rig and animated. The boards were done by Richie Chavez and Marcelo Vignali.

Thursday, February 1, 2018

Cloudy With a Chance of Meatballs Animation Test #1

An early development piece done around 2004 or 2005. This was before Phil Lord and Chris Miller were on board as directors, so that's why Steve ( at this time his name was Bonkers ) looks so different.

This was the second or third large piece we tackled as a group. Even though both the directors and design ultimately changed, we still view this as a success. It's important to keep in mind that development is an evolution. We put together lots of sketches, some 3d models and a fully animated test. Then we stand back to see what's working and what needs to be improved. Sometimes that means starting over.

From a cost standpoint, it's better to retool an idea as early as possible with a small crew than to go into production with a very large crew and then decide to scrap everything. Failing early saves time and effort in the long run.

There were six artists working for about 12 weeks in total ( model to final render ) for the entire piece.  Takao Noguchi modeled Bonkers and the Taco Truck, Shawn McInerney animated, Sungwook Su did the fx, Ernie Rinard did the spaghetti tornado and James Batersby modeled, textured and lit. I supervised, animated, modeled, textured and lit. I forget if Shawn rigged Bonkers or if I did.

**Side note* The dancing Drac at the end was done by Chad Stewart and myself. Along with development, we do fun little animated pieces for marketing. Chad animated and I did the rest.