Mikros Animation on how they brought a new stylized look to Teenage Mutant Ninja Turtles: Mutant Mayhem.
Jeff Rowe’s CG-animated Teenage Mutant Ninja Turtles: Mutant Mayhem very quickly made an impact upon audiences for its stylized and illustrative look and feel.
It also one very deliberately leveraged off the main characters being teenagers, both in terms of the interaction between the turtles and even the ‘style’ of animation.
In part one of befores & afters’ coverage of the film, we talk to Mikros Animation visual effects supervisor Matthieu Rouxel about achieving the desired look and feel.
Then, in part two, we turn to Cinesite, which also worked on the feature.
Illustrative, but realistic
Working with director Jeff Rowe, co-director Kyler Spears and production designer
Yashar Kassai, Mikros’ Rouxel helped early on to establish Mutant Mayhem with a mix of a stylized and illustrated 2D look and realistic cinematography with a handheld camera language. This included a photoreal approach for the lighting and a grounded performance of the characters.
“The 2D aspect of the picture was intended to give the idea of teenagers drawing their own movie in a paper class book,” says Rouxel, whose team would match 2D references that featured pencil textures, looseness and color shifts, from the creative team. “This meant we needed to bring imperfections on perspectives, proportions, contours and avoid as much as we could a digital feeling to the final render. You need for that to be cleverly approached with a mix of 2D elements–DMP and FX–and CG elements, that is, for those elements to be cohesive and fight against the nature of 3D rendering in many ways.”
Realistic cinematography was also key to connecting more emotionally with the characters, as Rouxel notes. “One of the early challenges we spotted regarding that is the complexity of having a very dynamic and erratic camera locked after the animation of the characters combined with an animation in step and the multiplicity of 2D features added to the pictures such as FX in a challenging production timeline.
We had the chance to have DOP Kent Seki the layout supervisor Christin Marineau delivering brilliantly this camera language, but also being very communicative and collaborative with the teams.”
How that look was achieved
Mikros Animation was one of several studios that completed a test shot for the film. It featured about 40 seconds of animation of Donatello talking. “The global brief was simple in a way,” recalls Rouxel. “We had to match with CG as close as we could to the 2D reference provided. Jeff Rowe was saying that if the picture looks like a 2D painting when it is not moving, that means we are good. Even if the camera was static, it was important for me as a supervisor for the purpose of the test, to build everything in CG–with surfaced elements presented on turntables–including the background, and not ‘cheating’ with use of matte painting and 2D draw-overs to stylize the picture, even if they would contribute a lot to the look of the show later on.”
This test helped inform Mikros Animation of the need to make specific developments in different sides of rendering for the project: stylized shaders, curves and edges.
In terms of stylized shaders, Rouxel says “we explored many options for the treatment of the character and built a shader prototype using the toonshader of Arnold render engine as a starting base. The prototype was good enough for the test but we knew we would need development for the scale of the movie. The goal was for us to have the look of a 2D paint out of lighting render, without having to apply a post process already. Jeff was very specific about getting way from a Photoshop filter feeling.”
“We figured out rapidly regarding the sophistication of the look which was not a simple cell shading effect and had to be different from the baked lighting look we can have sometimes, for example, with Arcane, as gorgeous as it is,” continues Rouxel, “and that we would need to be able to tonemap and add graphic features per light and/or group of lights, making it look as if they were painted one after another with different strokes. We had seven channels of light available for that, which was barely enough regarding the complexity of the lighting we often had.”
At Mikros, lead developer shading/lighting/rendering Marcel Reinhard developed a shader with which artists could isolate lights in the shading of the characters and objects and give them specific treatments. These included scribbles, cross hatching, stepping and color transitions. “We also built a shader and a workflow in Katana dispatching the work between the surfacing team, with presets of stylizations, and the lighting team, who were responsible for modification of colors and position of the stepping,” outlines Rouxel.
“Even if stylized, the lighting style had to be quite realistic. No unjustified rim lights or overexposure of the characters to make them more visible artificially was allowed. That’s part of what makes it successful and cohesive with the animation.”
Surfacing supervisor Anne-Claire Leroux and her character surfacing team utilized the Mikros toonshader to layer in textures. “They could isolate patterns and hatches to make them appear dynamically with the lighting and avoid that baked texture feeling,” says Rouxel.
“Environment surfacing supervisor Guillaume Chevet and his environment surfacing team also did great to avoid using any classical CG tilable maps or bump maps,” adds Rouxel. “These would have ruined the illustrative look by bringing realistic texture or high frequency and repetitive detail. It required very good painting and creative skills from the team to approach any material the way an illustrator would do it.”
Meanwhile, a curves approach was used to add pencil strokes on top of the render. “They were both present already on the models,” explains Rouxel, “and could also be added by animators or environment artists as expressions lines or speedlines. “They really became an important part of the look, the characters seemed naked without them.”
These curves were achieved via 3D ribbons in Maya and then rendered through Katana. Rouxel advises that Mikros needed to build a shader for the curves to maintain their width and texture whatever distance they were to camera. “Without this, they were just becoming invisible after 30 meters away from the camera. The CG perspective reduced everything naturally, but your pencil doesn’t become thin when you draw elements in the far distance.”
Finally, edges come natively from the toonshader, with Mikros able to tweak their visibility and thickness. The edges helped break the sharp and clean edges of CG objects, and were rendered separately for compositing.
The extra steps to give ‘Mutant Mayhem’ that extra push
While Mutant Mayhem is a 3D animated film, this animation, too, has a fantastical and stylized feel at times. But, says Rouxel, that didn’t come from a particular departure away from traditional 3D animation. “When we started to imagine what would be the film at the beginning, we didn’t really expect such a realistic approach to the performance of the characters. There is not an exaggerated use of stretch and squash or crazy deformations of faces. That is part of what makes it so unique to me. It is not just about style. “
So, where might that extra stylized look and feel, including in the animation, come from? Rouxel observes that modeling was one way that could imbue something different into the look. “We wanted to imply teenage imperfection everywhere, and it started with modeling. For the characters, we had to deal with a lot of asymmetries. The modeling involved approving the characters on pose with the asymmetry, then putting back symmetry into a T-pose for the rigging team, then the asymmetry being reapplied after that.”
“You can easily imagine how complex the process became,” adds Rouxel. “The modelers also had to really integrate the modeling of the curves, or pencil lines, as part of their jobs. It was fully integrated into the design, even though we also had tools for compositing to add these when finalizing the picture.”
Of course, the teenage turtles are not the only characters in the film. For a furry character like Splinter, Mikros also worked to bring a stylized look and feel to his groom. “We wanted,” details Rouxel, “to have richness of the silhouette brought by procedural hair but at the same time lose the high frequency detail and try to reproduce what an illustrator would do for Splinter, which is to put detail only where it is necessary.”
Hair supervisor Olivier Pierre oversaw the approach to crafting a specific shape language to the hair. “For the render, we adapted the shader for the hair to be able override the normal values of the hair and replace it with simplified normal values so that the light behaved more like on a smooth surface on which we applied sharp stepping and patterning,” says Rouxel.
Mixing 2D and CG
Aside from the characters, city environments and backgrounds also came into special treatment. New York City is somewhat of a character in the film, itself, and played an important part of the storytelling. Many views were achieved with digital matte painting (DMP) overseen by DMP animation supervisor Arnaud Philippe Giraud.
“The DMP gave early the tone and the target for the backgrounds,” discusses Rouxel. “Making it cohesive with the CG buildings was a big thing and it was crucial to have a good global approach of the mix. CG supervisor Olivier Mitonneau did a great job overviewing this part, too. The stylization already built in modeling and shading helped to make the split invisible most of the time. The 2D FX were another big challenge as they were stepped but needed to look natural and not strobe with a handheld camera.”
The DMP team worked mainly with Photoshop and Blender–“Blender was used for its Grease Pencil features allowing us to add graphic pencil lines with a keep alive look,” says Rouxel. “A lot of the animated city lights in the background were obtained with this feature. The layers of the DMP were also delivered in separated channels in EXR files so that compositing could have the same kind of control that they would have with CG elements.”
Compositors would then bring all this together and often add even more graphic elements such as textured volumetrics, scribbled motion blur, stylized defocus and scribbled flares. “They were working with a render already including quite some stylization but that was the opportunity to go even further by adding more features,” notes Rouxel. “One of the challenges for this department was that most of the soft classical blur effects like defocus, motion and glows had to be banished as they would bring back too much of a digital feeling.”
For defocus instead, then, Mikros treated depth by simplifying the textures in depths, making the pencil construction lines bolder and by creating scribble-like bokehs for highlights. Motion blur was achieved more by generating scribbles on top of the picture in compositing to replace the classic CG vector blur look. A flare tool enabled the drawing of 2D shapes–many of the shapes had been provided by 2D FX team, but compositors were also able to draw their own.
The most challenging shots: the long ones
Asked which shot or shots were the toughest in Mutant Mayhem to achieve, Rouxel identifies “the many long shots we had with a handheld camera following the action. The director really enjoyed these kinds of shots because they were very immersive and gave a lot of space for the performance of the characters.”
What made the shots even more of a challenge was to ensure that so many angles and movements could be brought to life and maintain a consistent style. “It can be challenging first from a lighting point of view,” admits Rouxel.
“The tone mapping of the light is adjusted to make it work nicely with the position and the shape of the character. When we see the character in many angles in one shot, It can require several animated tweaks to keep it appealing all along. Then for 2D elements like FX, volumetrics and DMP, these need to properly placed or tracked in 3D space with a certain amount of layers to have a good sense of depth.”
Stay tuned for part 2 with Cinesite…