Geralt's sword, silent monsters and the unseen performance. Inside Maciej Kwiatkowski's world of mocap

It might seem that the ceiling in the Gamepressure.com editorial office is suspended quite high, but only until a man who wields a sword shows up there.

Of course, there were no casualties in the lamps - Maciej Kwiatkowski, from Alpha 7, who visited us is a real pro and knows exactly how much space he takes up, making swift cuts and turns. After all, he is the Witcher - the person who created the fighting styles known from CD Projekt RED games, and also the actor who performed the movements of Geralt transferred to the game through motion capture sessions.

Motion capture is a fascinating topic. This is one of those professions that didn't even exist half a century ago and has developed alongside the advancement of technology; now it is inextricably linked with video games as a medium. And also... with the theater, as Maciej Kwiatkowski made me aware, telling me about the secrets of his art. How much training does it take to move like a witcher? When does an animator have to step in? How do technical limitations determine how a character in a game can move? Can the performer portray the movement of any character, even those differing in gender, age, size? What is more important when deg character movement - realism or effectiveness, and can AI help with this? How do you convert a book description into movement that will look great on the screen? And finally, what Geralt of Rivia and Darth Vader have in common?

But let's start at the beginning.

Anna Garas: What is the difference between a stuntman and someone who does motion capture?

Maciej Kwiatkowski: We had a problem with naming ourselves before because we started out as stunt performers, and in the credits, we appeared as stuntmen. Over time, optimizing the production process requires the performer to adapt. It doesn't matter how we look, so the more roles you can play, the better for you and the client, who doesn't have to hire extra actors. Over time, you expand your range of skills beyond pure stunt work, you get into acting, then into playing monsters - this is practically nowhere else, except in the theater. In Hollywood, performers who play non-human characters are a very niche specialization, but in games, there is a high demand for them because there are many monstrous opponents. By force, I was learning all of this; I was improving, but I also discovered that I have a knack for it. And suddenly it turns out that I'm listed in the end credits as a stuntman, while I play the main character - but I'm not just an actor.

A.G.: It reminds me of performance art.

M.K.: Yes, although it has already been adopted in our context, I hear it in the film industry. We're working more and more with the Western market, so we're standardizing our vocabulary to make communication easier. When it comes to stunt performers, we really like this distinction because there are so many different areas, roles, and specializations within the stunt industry. There are stunt performers who simply do stunts - falling out of windows, acting out fight scenes, driving cars - but there is also a huge group on the other side of the camera, safety, rigging effects, securing the entire crew at heights, or preparing cars. Performer and performance are the words that best reflect how internally diverse mocap is. Because there is body performance, there is full performance capture, when the subject of animation is not only the body, but also the face.

A.G.: Speaking of the face, one of the clips posted on your site stuck in my memory; the performer there was acting very expressively, like an actress in a silent movie. Does face mocap require such overplaying?

M.K.: This video clip that you just mentioned needs context - this character was supposed to be exaggerated, almost crazy, that was the intention of this expression. However, performance capture is quite specific, and it's something we teach when we bring in new people, especially if they are actors or folks involved in the film industry. They have their vision and their habits related to the film medium, and sometimes they find it hard to switch to mocap. Of course, when we're shooting cutscenes, we have the cameras roughly set up for the scenes, there are storyboards for that, so the actor knows which camera to play to. But once they were totally lost in empty space, without scenery, without costumes, without any context.

In of the strength of facial expressions and gestures, mocap is much closer to theater than to film. In movies, the camera is usually closer to the actor, especially in dramatic and dialogue scenes. In the theater, on the other hand, we perform to an audience that is far away. We don't just play to one camera and we don't fake anything with perspective or editing, so theater is much more expressive - and this also relates to mocap, because in the end, in games with a third-person camera, we see those characters from a distance, so the expression has to be very broad. Also, not all movements are necessarily emotional, the idle poses must be exaggerated. A good example is breathing - taking deep breaths and exhaling so that they can be seen from 10 meters away requires intense hyperventilation. If the character is to wear armor and we want it to be visible that this armor moves as the character breathes, we should further enhance this effect with the movement of the shoulders. Animation simply likes a broader gesture. There's also a technical aspect to consider. In simple , it's easier for the animator to narrow down the movements rather than to make them bigger.

A.G.: I think there's a parallel between you as Geralt and the most iconic movie character of all time, Darth Vader. He was played by two different people, and his charisma came not only from his voice but also from the stage presence of the two-meter-tall David Prowse. In the end, Geralt's voice differs depending on the audio version, and his lines vary depending on the translator's skills, but his body movements remain the same, and that's largely you.

M.K.: You brought up a very cool comparison... Darth Vader was probably played by three actors.

A.G.: True, the third one is in this short scene, when we see him without a mask.

M.K.: This is a perfect comparison to mocap. In games, one character can be made up of elements from up to five different people. Sometimes one person provides the face, while another is responsible for the expressions. But in the end, I can technically play most of them. Including female characters, although playing a woman by a man, and vice versa, is an interesting topic in itself. It's challenging, but doable. But, of course, you play any character, except for the really big or really small ones. It's hard to fake body size; that's just physics. The heavier a person is, the more effort they have to put into moving their mass, and a falling child has much less inertia than an adult. All of this is visible in motion capture.

A.G.: Does this mean that there are child mocap performers?

M.K.: Of course - not far off, the actor playing Atreus in the new God of War. By hiring children for mocap, you can get really cool material, but this is often avoided, mainly for production reasons. The work of children is strictly regulated by health and safety regulations. A child can be on set for a shorter time than an adult, must have a guardian, and any activities that are slightly more risky are obviously not permitted. And besides, a child is a child, right? It can be moody, not want to cooperate, and get tired much earlier than an adult. Working with children is much more demanding, so often instead of them, small actresses are hired. For instance, the motion capture animations for little Aloy in Horizon are done by Luisa Guerreiro, who is a motion capture performer and a film actress. Luisa is a person of very short stature, and she is very agile. She was hired for all the action scenes where the several-year-old Aloy performs somersaults, fights, jumps over obstacles, etc., and it worked great. But a good child actor is irreplaceable.

A.G.: Your work is not just fantasy and swords, right? I heard that you got some advice from GROM for a project.

M.K.: Yes, during the episode of the Crossfire series from Secret Level, which we were making with Platige Image for Amazon. This is tied to a bigger story - this project is actually an anthology, with each episode being created by a different team. However, it's all overseen by the main director, Tim Miller, and Blur Studio, who ensure quality and consistency throughout. One of the main aspects of our episode was realism - photorealistic graphics, but also tactical realism. And the client had Navy Seals consultants on their side, commandos with combat experience. The director of our episode, Damian Nenow from Platige Image, is a ionate enthusiast of this subject. He perfectly understands this aesthetic and does not tolerate fakes. He told us that he worked closely with consultants from the Navy SEALs. These are the guys who, when the boys asked how a situation they were directing would really look like, would just play recordings from their combat cameras, real footage from the battlefield. In short these are the people who can't be fooled, so we really wanted to present something at the highest level. On top of everything, there were rumors that the client was coming from a position of thinking, "This is some company from Poland; maybe they don't necessarily know what they're doing." This only fueled our ambition even more. Fortunately, I had previously worked with Adam Rycman - a GROM veteran - on promotional materials for Alpha 7 and the series Go Ahead, Brother." So I could have gotten him involved in our production. It is very satisfying that we were able to do it the way they do it in Hollywood, that is, by engaging real veterans with combat experience, not just theorists. We don't want to do things like in the 80s anymore because it was like the Wild West back then. Shooting from the hip, putting a gun to your head, or waving a weapon at people just because the director thought it would look cool. Now we want to stick as close to reality as possible, with the exception of single subtle nuances, because real things simply make sense.

A.G.: Looking from this perspective, I respect even more the effort put into creating the Witcher's fighting styles, because theoretically they have no right to work, right? Because they should be used to fight against opponents who do not exist, performed by people whose capabilities significantly exceed those of ordinary humans. So how to do them?

M.K.: There are a few aspects to this. In video games and movies, and especially in stylized movement, the primary guideline is aesthetics, so the movement is intended mainly to look pleasant for the audience and provoke a reaction. The Witcher is an epic character, extremely proficient, so his way of moving and fighting is primarily meant to be very spectacular. Various purists often criticize such styling, arguing that all these spins and pirouettes do more harm than good and are pointless. Here, however, I will say that everyone must go through the path that we have gone through in our work - because if we strip the combat of all these flashy moves, coolness and things that are mainly meant to evoke emotions, if we leave only those movements that are optimized, utilitarian to the extreme, only those that would allow you to win in a duel - then it just looks bad. Firstly, this is boring, secondly almost invisible, because the determinant of effective combat movement is that it will be invisible, very brief and unnoticed by the opponent. So what is there to see, where can you get fun from?

As a teenager, I asked myself how to convey the Witcher's move, when my friends and I read The Witcher and after hours we created the first Witcher choreographies - with real swords, because as of knightly brotherhoods we had access to steel weapons. Later, when we professionally approached this topic with CD Projekt RED in 2005, the same question came back - how to interpret Sapkowski's prose, where on the one hand there were pirouettes and various oriental elements, and the Witcher moved in a very spectacular way, but on the other hand his slashes were so fast that they seemed invisible. How to show this? Leave our character motionless and just show that the opponent is falling? These are the aspects that need to be considered and people who undermine them simply do not understand them. If they saw what they want, they would realize they don't like it.

A.G.: So this requires a process of a more creative translation of these movements.

M.K.: Yes. What we and a multitude of people before us have come up with is the process of storytelling; the extra flair that we add, which some might consider unnecessary, work for most people because they simply enjoy watching it. The second aspect, when it comes to portraying a supernaturally fast character, is the animation itself. First, there are us, the performers, who record this interpretation of movement as best as our bodies allow us. Of course, this is preceded by training, I optimize this movement for a very long time, I learn it, I train my body; these are thousands of days, tens of thousands of hours of practice - but still, the animator later speeds up, makes cuts and optimizes this movement, and thanks to this, the witcher is closer to the image from the books. But of course, in games there are other characters that move even faster; it's all a matter of aesthetics and conventions - and mechanics above all, because for example fighting games have movement cut down to single frames.

A.G.: You mentioned in our conversation before the interview that your role as a performer was to design Geralt's movement in such a way that every positioning of his body testified that he is the most competent person around. I wonder how you would approach playing a young Geralt from the Crossroads of Ravens? Would he be a bit clumsy?

M.K.: Here, gameplay would stand in our way. This is a challenge that I encountered during my first experience with motion capture in the first Witcher. We had spectacular, cinematic ideas for attacks, but the developers said "no, no, our system says otherwise. You cannot make such an attack, because our tree of delivering blows is built in a specific way." Gameplay takes precedence and that's that. We would like the character to stumble occasionally, but we simply didn't have room for that many animations - although now the amount of data in the game has increased and we can afford, for example, to have the same attack have two slightly different animations, played alternately, or for the character to perform some specific movement while running. All of this is just extra data.

Of course, we would like to have more freedom here, because the more a character has animations, the more it feels alive, real and gives more immersion. I worked on Quantum Break, where a lot of emphasis was placed on the fact that our character is not a superhero, but just a normal guy, and the whole world around him is falling apart and burning, so this chaos was very nicely transferred to his movement - he was stumbling, bumping into elements of the environment, all of which added drama. This was an example where we could afford this and it was great, but this wouldn't be advisable for every game, right? The character becomes more human, has more character - especially in times when everything has already been done and perfect characters have been created so many times that they are no longer attractive. Now every uniqueness counts and it is more attention-grabbing when the character is less perfect, has its own flaws, when its character is reflected in the fighting style. So it would be perfect to play Geralt with this youthfulness. Moreover, one would even wish for his fighting style to evolve: to start as a novice and end as a professional witcher.

A.G.: Maybe over time he would learn the coolest attacks?

M.K.: Yes - but here again technicalities come into play, because it at least doubles the amount of already massive animation sets. Now, the developers of every game strive to make the gameplay provide more and more diverse experiences, so we have basic move sets, basic attack sets, then special attacks, contextual attacks, then some QTEs, finishers, attacks on specific characters, on specific enemies. Even specific states - separate animations for a drunk character, moving in water, carrying heavy armor or without armor at all. When each set has, for example, 200 animations, then each subsequent one is another 200 animations, and the production process will not go on indefinitely - this is the reality we face together with the developers.

A.G.: Is it because of these production limitations that playable characters with motor impairments are so rare?

M.K.: Playing a character with impaired motor skills would probably involve a greater gameplay challenge; in my opinion, it would be a very tough nut to crack, because the protagonist is always key, and one of the ruthless rules is their responsiveness to the player's commands. All responsiveness limitations are annoying - that's why movement animations often have their beginnings cut off so that the character starts running right away, without a run-up, as soon as you touch the analog stick. The attacks are as responsive as possible, and all those wide swings belong to the high risk - high reward group, because they carry the risk that during them you expose yourself to the enemy. Would playing as such a character be associated with a higher degree of difficulty?

But all these limitations disappear when it comes to friendly or hostile NPCs - the developers like to go wild here, they like imperfections and quirks in movement, whether they are zombies or some monsters, which lack arms, legs, or have one leg shorter or even sport additional limbs. Such a quirky, outlandish and inhuman movement arouses greater emotions. And to make such a move, we use various aids - if one of the limbs is to be larger, we attach weights to it, because it will affect my center of gravity and the motor skills of the entire body. Or we will add extensions to our arms - here we can cite the example of Planet of the Apes and those characteristic stilts designed specifically for performers, because as humans we have much less developed upper limbs than monkeys, so walking on all fours is very uncomfortable for us and looks bad. But extend your arms a little bit and it's already much better. There are companies specializing in the production of such gadgets for performers, such as stilts in various sizes and configurations, with additional ts. There are even tails that are attached in such a way that they move realistically while walking.

A.G.: Sounds like a cosplayer's dream.

M.K.: For cosplay, it's great, for mocap less so, because in the end it all adds weight. I once tested such stilts - they work for walking, but God forbid you try to make turns, any dynamic movements like jumps - you can forget about it. The body mass ratio turns this super agile monster into a guy on super heavy stilts. [laughter]

A.G.: So, are they at least useful for walking gracefully?

M.K.: Once you get moving, you're like a locomotive, you go by momentum, your motor skills look good, but to record animations for the entire gameplay - no chance. In the end, an animator will do a better job at animating these legs.

A.G.: Speaking of deg the movement of non-human, monstrous creatures - I an interview with director Hayao Miyazaki, who commented on an AI-generated animation of a zombie moving around using its head. It looked very creepy, and Miyazaki said that this way of moving reveals a fundamental lack of empathy towards living beings, that it repels him on some fundamental level. I wonder, do you think such an application of AI could work?

M.K.: AI is already being used to create basic human movement. Of course, the results are a bit creepy for now, the "uncanny valley" is very apparent, but I expect that the development of this technology will now be much faster, because it will be very easy to earn money from it. From what I've read, Mr. Miyazaki criticized the fact that AI affects such a distinctly human domain as creativity - like the aforementioned unnatural movement. For me, this is an opportunity - for AI to generate something that a human could not create. When we make games featuring very creepy characters, we try to move away from our human limitations as much as possible to create something that will send shivers down your spine. Sometimes we even resort to editing tricks, because we can only depart from our human nature up to a certain point. We won't leave our bodies, we won't break our bones to perform some inhuman movement, so from a certain point, we are dependent on animation tricks. So, I believe that this is the area where AI can surprise us, because we as humans will ultimately not get rid of this humanity, although I think that we still did a good job. Me and Madeline Page-Ulmer had the opportunity to work on Silent Hill 2; I played the main character and male characters, and she played all the female characters, of which there are plenty, including those iconic nurses. We had a great reference point in the form of the original Silent Hill, but we still had to play out a lot of this nightmare - and from what I hear, it turned out cool. Creepy enough.

A.G.: For me, it was definitely creepy enough [laughter].

M.K.: Fortunately or unfortunately, developers always have to push further, test our sensitivity - and develop themselves.

A.G.: Right, like everyone else. Great, thanks for the talk!