She–Hulk: Attorney at Law is the latest and eighth television series in the Disney+ Marvel Cinematic Universe. Tatiana Maslany stars as Jen Walters, a lawyer specializing in cases involving superhumans & villains, and who herself is the green superhero: She-Hulk. Filming began in mid-April 2021 in Atlanta, Georgia, and lasted until mid-August. One of the primary visual effects companies on the series was Wētā FX, in New Zealand. For a sizable number of the actual She-Hulk shots, Wētā FX produced feature film-level character animation converting 5’4″ Tatiana Maslany into the towering green 6’7” She-Hulk. While She-Hulk’s visual effects needed to match the brilliant work done for Hulk in the Marvel cinema films, the script required Tatiana Maslany and Wētā FX to deliver something very different from what her male cousin had exhibited on the screen.

(Note: while this clip above shows the fun Tatiana Maslany had on the Law office set, Wētā FX did not primarily do the shots of She-Hulk in the law office.)

As seen in the humorous clip above, much of She-Hulk’s performance involved shooting Tatiana Maslany on set in a MoCap suit, but given the vast height difference, many of her performances were filmed with the actress walking on a special raised platform or wearing a false plaster face suspended a foot over her head. The actress also wore a head-mounted (HMC) vertical stereo camera rig to capture her facial performance.

Guy Williams was the visual effects supervisor and Sidney Kombo-Kintombo was the animation director at Wētā FX. We sat down with them in New Zealand to discuss the complexities of animation, re-targeting, and producing complex and subtle acting performances that matched Tatiana Maslany’s original performances. The two leads both independently praised the dedication and commitment of the show’s lead actress. The entire team worked to not only technically deliver but to also keep the emotional response they saw in the original footage of Maslany.

The Process

The first task for the Wētā team was to build an actress puppet. This is a matching version of the actress that shows the team if the data processing from the HMC is producing a faithful reproduction of Maslany before that data is used to re-target to the She-Hulk digital character. Maslany was scanned in the Medusa rig and a standard feature film Wētā facial rig was built. This complex and powerful actor puppet rig has been used for all humanoid characters at the company since Gemini Man and is the result of years of detailed work to develop an exact topology with a certain mesh structure. Maslany did not need to be scanned in a Light Stage as this is only needed for skin texture and pore detail and no viewer will ever see this digital puppet, nor did She-Hulk need scans of Maslany’s skin, since She-Hulks skin is quite different to Maslany.

She-Hulk’s face used the same underlying high-end standard Wētā FACS-based face rig which allows the topology of the actress to be the same as She-Hulk’s face, but this is only needed when the character of Jen transforms on screen into She-Hulk. Without matching topology, technically skin or aspects of Jen’s face may not line up seamlessly and grow into She-Hulk’s face. However, while their faces could be solved this way, there is no such perfect match from Maslany’s Jen’s ‘very short frizzy perm’ hairstyle to the dramatically-styled hair of She-Hulk. Furthermore, for She-Hulk’s hair to move appropriately, it was always simulated, thus it moved and bounded correctly based on She-Hulk’s head and body movements. To make the two hairstyles match visually as much as possible, both the actor puppet and the final She-Hulk had the “same number of hairs between the two grooms and the hairs are in the same place, so that they just grow along their length,” explained Williams. “We took one groom and then re-groomed it to the other style, but the simulation aspect made matching them visually very complex… we didn’t groom She-Hulk’s final hair to be the final result. We groom what we call a ‘finger in a light socket’ (zero gravity) version, and then let it fall in the simulation, as it is the friction between the hairs that contributes to the amount of movement her final hair has.”‘

While She-Hulk and Jen share the same topology, their skin properties and of course color are very different. Initially, the studio was planning for a magazine cover-like skin property for She-Hulk which is slightly stylized as most ‘air-brushed’ magazine cover shots look, but this worked against the realism of the digital character and a more realistic version was developed, but still one that did not sweat or have the normal level of human skin oil or even eye moisture. Wētā also did the Hulk in the early scene at the Hulk’s beach hut location so the team got to balance their green skin tones against the more familiar colors of Smart-Hulk.

William’s texture team leveraged the very human disparities we all have in our facial skin. The skin around the eyes tends to be yellower than the forehead. The skin around the cheeks is a little bit pinker than the eyes. The skin on the nose is typically pinker or bluer than the skin around it. “You have to get all that in there. And on the original (feature film) Smart Hulk, ILM did a beautiful job of that,” Williams points out. “Hulk has a lot of flesh tones in his green. He has a lot of reds in his green. It takes a very complex render to be able to achieve that without making the skin just appear muddy brown. I thought ILM did a beautiful job,” Having matched the ILM solution for Hulk, Wētā built on this work to give She-Hulk her own feminine version, influenced not only by gender but by the different nature of Jen’s response to Bruce Banner’s blood, and thus She-Hulk’s unique persona. She-Hulk was finally rendered using the main Wētā spectral renderer: Manuka.

The re-targeting of She-Hulk’s face was difficult but there were other complex re-targeting issues. In a signature shot when Jen arrives at the Wedding, Wētā had to match Maslany’s sassy walk but with She-Hulk having much longer legs and thus a much larger stride. For some sequences, as limbs are not only longer but in a different length ratio, this means referencing another much naturally taller performer on the MoCap stage and then blending that walk pacing into Maslany’s performance. “Jen takes 10 steps to go three meters and She-Hulk only needs to take seven steps,” comments Williams. “So there was a lot of work to try to figure out how to keep the stride and the performance, in the end, there is no perfect solution, it just comes down to the skill of our animators.

 

The Wētā Character Animators

“We never in animation let anything get out of our department saying, ‘Oh, it’s gonna look better in lighting’. That’s not the conversation. We need to make sure that our emotions are right. All the rest later is a bonus,” -Sidney Kombo-Kintombo

Sidney Kombo-Kintombo is a senior animation supervisor at Wētā FX. Sidney was one of Wētā FX’s Animation Supervisors on Avengers: Infinity War (2018) where he pivotally contributed to Thanos’s nuanced facial animation in the fast-paced third-act battle sequences. Sidney first joined Wētā FX as Senior Animator on The Jungle Book (2016), quickly stepping up to Animation Supervisor on War for the Planet of the Apes (2017), most notably he was key in creating the emotional depth of Koba’s character. With She-Hulk his team needed to deliver a very human performance, that was re-targetted but not a caricature. “The Hulk is a caricature. So when the audience looks at him, they forgive a lot of things that may not be accurate, because you don’t know how his face would be in real life, but on She-Hulk, you can’t do that,” he explains. “This is because all of She-Hulk’s facial features are the ones that we see every day… So, if something we do is wrong, people will catch it even if they don’t understand what the exact issue is. So having the chance to do She-Hulk and seeing how she came across – I was really proud.”

In terms of priorities for the animation team, Sidney Kombo-Kintombo focus was initially on She-Hulk’s eyes. While he points out that on many pitches, and test projects creatives and studio execs want to see a character talking, a more important test is the nuanced acting that comes from a character listening to what is being said and emotionally reacting. “When She-Hulk is talking, she can go through a lot of different types of emotion. I don’t start there. I try to find the shots where Tatiana is nuanced and not speaking that much, – like when she is intensely looking at somebody or analyzing them.” Not that mouth animation is less important, but rather he sets a very high animation bar for the team.

Sidney Kombo-Kintombo animation team works with an accurate 3D Gazebo render, but not the final render quality that is seen from Manuka. For example, the skin shaders or the character’s eyes have no moisture. The Gazebo preview is detailed but still rough compared to the final Manuka renders, but Sidney Kombo-Kintombo requires that every shot works when it leaves his team and no one is allowed to say “Oh, it’s gonna look better in lighting.” He requires all the shots to fully communicate the sub-text of a shot, and the emotional story of the character fully before the extra realism is added that comes from Wētā world-class compositors and Manuka’s final ray-traced rendering.

The Gazebo render is what the team calls a ‘light render’ meaning the character is placed in the shot with the approximate right lighting of that environment. If the animation team need a highlight in the character’s eyes the team can get that as the process is very concurrent and agile with the other teams. But still, the light is not interacting with the character’s skin yet and while her skin is textured, it is not yet final. “We try to disassociate the emotion from the reaction to the emotion,” he explains. “If you can read her emotion — if you can read that she’s nervous — everything that happens on the surface of body or skin, at that point, is not my concern. What I’m trying to sell you as my client is, ‘Hey, I captured her franticness, or I capture her being nervous. If you say, ‘Oh, can she be sweating a little bit here?’ — that’s not my job. Guy (Williams) team is going to give you that, of course. But, we focus on the fact that the expression is right and the intensity of the look is right. That though you ‘feel’ her fingers are touching something, – that you feel that she’s nervous. If that feeling is right, everything else is a bonus.”

 

Sidney Kombo-Kintombo actually cites his favorite shot from Infinity War, when Thanos is defeated and he declares ‘the hardest choice requires the strongest will’. And as he’s saying that he’s looking down and you can feel that his eyes tearing up.” While there is no closeup of Thanos’ eyes at that point, there is a tiny glint of light to suggest his eyes are getting moist, but it is so small the audience may well miss it.  But for Sidney Kombo-Kintombo, the audience can feel it even if they can’t see it, since the animation is so strongly built on the powerful performance of actor Josh Brolin. “We see from the performance that he has lost so much and we try to make sure on our side of the business that we don’t lose that. I never lose the intention of the actors. They come on stage and they just act it out so well.”

One might think the partial final render quality is a problem for animation but the opposite is true. “To be honest, “I fight for that to still be the case,” he states. “I don’t want to deal with say the timing of sweat forming.  It’s constraints on the movement that are relevant to me. In a sense, what we are fighting to do is convey emotion. So I fight for the animation team to be able to control when a muscle flexes, the muscle flex is artistic. But the change of skin properties at the physiologic level – we don’t want that. I don’t need that to convert emotion”.

We asked Sidney Kombo-Kintombo if he saw a role in the work Wētā has done in machine learning and if this applies to the animation team? “We have loads of brilliant tech here, but I’m on the side of those who think that the efficiency of the expression is more important that the details of the expression. When I see an actor’s performance, I don’t look at it in terms of what her skin is doing. I’m looking at it in the sense of the detail of the different expressions and transitions in expressions. And I try to make sure that we capture that.”

While modern HMCs and face puppets transfer a lot of data from the actor’s face, the data set is still very sparse compared to the subtly of a human face. For this reason, the animator’s skill is very much central to correctly emotionally telling the story, and allowing the acting choices to be accurately weighed and represented after re-targeting. ” Yes the technology can indicate a smile, but maybe She-Hulk has to hold that smile a frame of two longer to match the same sense or emotion we get from Tatiana’s original performance”.  There was a small percentage of shots where the animation team needed to make She-Hulk do something that was not filmed that way on set with the actress, but “90% of the shots it is just Tatiana.” The production worked hard to make sure the actress was physically in the right position so her eyelines were correct and her co-stars could properly act to her and not a ‘tennis ball on a stick’.

While focused on She-Hulk shots away from the Law Offices, Wētā FX also did Daredevil’ digital double at the parking garage (and nearly all of the actual parking garage set), the majority of the Demon fight in episode 3, and later in the series, the character animation work of Abomination.

All Images are © 2022 MARVEL.