Stitch Head (2025) is an animated feature film based on Guy Bass’s beloved children’s novel. A co-production between Gringo Films (Germany), Fabrique d’Images (Luxembourg), and Traumhaus, with animation by Assemblage Entertainment in India, the project brought together artists across continents, formats, and pipelines.

We spoke with Oliver Finkelde (VFX Supervisor), who guided us through the production — alongside Stéphane Lecocq (Production Design), David Nasser (Animation Director), Nico Rehberg (Lighting), Juliane Walther (Line Producer), and Viola Lütten (Producer).

DP: Oliver, how did you get involved with Stitch Head and how would you describe the film to our readers?

Oliver Finkelde: While vacationing in Zanzibar, I was contacted by Juliane Walther, the line producer at Gringo Films, about Stitch Head. The studio was looking for an experienced VFX Supervisor with a strong background in AAA-quality feature films. Having worked on many major productions in the U.S., I was immediately drawn to the project — it felt unique, ambitious, and genuinely fun and the team was incredibly welcoming. 

The story of Stitch Head is deeply touching and profoundly moving — filled with hope, love, and friendship. Its themes resonate strongly with me as a father. In addition the director’s vision and creative ambition were enormous — in my eyes, a real game changer for Germany. 

DP: Can you give us a sense of scope? How long did the production run, how many people and studios were involved, and what did the international collaboration look like?

Oliver Finkelde: I will hand this one off to Juliane as she’s been part of the project since the beginning.

Juliane Walther: The actual image production, starting from the script, started in 2020 and took around five years. Before that came the earlier stages – financing, development, and scriptwriting, which added several more years to the overall journey.

If I’m not mistaken, over the course of production around 800 people contributed to the film in different phases. The size of the core team varied depending on the stage of production – from about 10 people in quieter periods up to 20 during peak times.

The team was spread across Germany, France, Luxembourg, Belgium, England, India, and the United States, with eleven studios in Europe and India involved in the image production alone. Beyond that, additional creative and technical partners from these countries played key roles in bringing the project to life.

DP: From your perspective as VFX Supervisor, what makes Stitch Head interesting in terms of how it was produced?

Oliver Finkelde: When producing in Europe, proficiency, productivity, and agility are key to ensuring that every cent ends up on the screen. The core team was intentionally kept small, which allowed us to make decisions quickly and enable all departments to work as efficiently and creatively as possible.

Production Design

DP: Oliver, production design sets the tone for everything. Can you introduce Stéphane, who led the design? What made his work crucial?

Oliver Finkelde: It was my first time working with Stephané, and I couldn’t have been happier. Stephané has an incredible eye and a remarkable sense for creating worlds that truly serve the story, adding depth, texture, and tangibility, and allowing the audience to become immersed from the very first moment.

DP: Stéphane, how did you approach the look and feel of Stitch Head? What references guided you in balancing a gothic atmosphere with a children-friendly story?

Stéphane Lecocq: The look and feel was a mix between the old school Hammer movies from the fifties like the bride of Frankenstein, and the more actual 3d cartoony movies. It was quite challenging to find the sweet spot between those 2, in order to make the movie an experience for the whole family I had to mix the dark tones of the castle, balanced with a friendly and cosy atmosphere with all the monsters. It’s still a comedy. Our director, Steve Hudson, is an encyclopedia when we talk about movie refs. So we tried as much as possible to incorporate them in the movie.

DP: Can you walk us through your workflow ? What tools and methods you used, and how you translated sketches into production-ready assets?

Stéphane Lecocq: My creative process always begins with sketches on photoshop, followed by a 3d model on Blender, which is my favorite 3D tool. I add some lights in the scene and then voilá. We can do so much now with those programs to speed up the process. Then I send my rough models to the teams (storyboard, set modeling etc…). It’s easier for them with all the references and the different camera views.

DP: When your designs moved downstream, what did you want people to take away from them?

Stéphane Lecocq: I was very picky with the “style “of the environments, where I tried to avoid straight lines, with every furniture and props a bit broken and bent..

DP: Which character or environment in Stitch Head are you most proud of?

Stéphane Lecocq: Definitely the lab, with all the moving parts, with all the lights coming from the different devices, plus the FX parts. It was challenging to make this set work. But the result is great. The caravans also! It was really fun to design the charabanc.

DP: Looking at today’s technology: if you were to start Stitch Head now, would you consider a game engine like Unreal for design and visualization?

Stéphane Lecocq:I think nowadays, Unreal Engine can be a really powerful tool to speed up different process, like creating sets, skies and moods. The library of props is so huge. But for that, you also need a correct pipeline, and with so many studios working on a movie, it’s not an easy thing to set up. The future, with the rise of AI is not so peaceful. The way we handle production will change, and is already changing, at fast pace. I just hope that we will still continue to draw, and create designs by hand, and not only prompt with AI.

DP: Was there anything about this production design that posed unusual technical challenges?

Oliver Finkelde: What I love about animation is that anything is possible and nothing is ever unusual. Still the creative leadership worked closely together to ensure that Stitch Head’s world, the environments, characters, and overall design, translated beautifully into 3D. The goal was to capture the essence of Stephané’s artwork while keeping everything as manageable and production-friendly as possible.

With the color script always in mind as the guiding reference, a great deal of time was spent on planning and carefully managing complexity and levels of detail. Understanding the color script, layout, and working hand-in-hand with the director and production designer to know where to focus resources was absolutely key.

In terms of technical support Vishnu Ram at Assemblage played a key role in making sure shot production including deliveries to lighting were consistent and efficient.

Animation

DP: Animation gives these designs life. Oliver, can you tell us about David and the animation team, especially India?

Oliver Finkelde: As an animation veteran, David was the perfect fit to direct and mentor the animation team over at Assemblage in India. In the Animation department especially, being present during key stages such as team building and character exploration was crucial. Having the ability to divide time between Europe and India proved vital to the success of the production.

David Nasser (Website I LinkedIn I IMDB) is a director and animation director with a strong foundation in visual storytelling, honed at the California Institute of the Arts (CalArts), where he specialized in animation. Over the course of his career, he has held senior and lead animator roles at top studios across Europe, the UK, and the United States.

David’s credits span a wide range of acclaimed animated features and series, including Arcane, Despicable Me, Hotel Transylvania, Rio 2, The Chronicles of Narnia, and many others. He served as animation director on the Oscar-nominated film I Lost My Body, as well as Zombillenium, The Tiger’s Apprentice, and the recently completed Stitch Head. Currently, David is directing animation on the upcoming feature High in the Clouds, and is also co-directing the original animated film ADAM, which is in development.

DP: David, what was your first step in setting an animation style and making sure the animators followed it half around the world?
David Nasser: The first thing I look for are boundaries and limitations that help define the animation style. Most importantly, it starts with character design—it sets the tone and complexity of the animation. Some designs offer more range and freedom, while others demand a tighter, more controlled approach. When I see the characters, they immediately come to life in my mind, and I follow my gut.

Next, I gather everything else: the script, tone (comedy, drama), concept art, animatic, the director’s vision -and of course, budget and resources. That’s how I begin shaping the animation style… in theory.

In practice, there’s a lot more to it. You have to assess complexity, production time, and costs early on to see whether the vision is actually achievable. Dreaming big is great, but we also need to deliver on time -and consistently.

Beyond the artistic vision, my role as Animation Director connects many departments and people. I rely heavily on production support, supervisors, leads, and every single artist. Especially on international productions, clear communication is essential. To ensure consistency and quality – no matter where artists are based – we need a well-established workflow, strong motivation, and ideally, personal connection. Meeting people in person, when possible, makes a big difference. Just as important is creating a respectful environment where people can enjoy the process, feel valued, and grow through the experience.

Back to the style itself: this project, being a comedy-drama-creature tale, required a nuanced approach. Emotional moments called for subtle, more naturalistic animation, while the comedic scenes leaned into snappy timing and cartoony exaggeration. The monsters, who often appeared in groups, had to be more limited and stylized to manage complexity and technical constraints. The challenge was balancing these different styles while keeping everything in the same universe.

Together with the director, we curated reference materials, both animated and live-action, to help communicate the tone. From there, we began test animations, usually with a small team of 6–8 artists and a supervisor. These early tests helped define character performance and build a pose library, which helped to keep characters on model and supported consistency as new artists joined.

I build new teams for every project – sometimes up to 70 artists and more. It’s the hardest part, but also the most rewarding.

To keep the animation style consistent and stay connected to the team, I rely heavily on live reviews. Every artist presents their shot and receives direct, real-time feedback. I prefer to keep things simple and avoid too many layers between artists and decision-makers. It keeps communication clear and creativity flowing.

Motivation and team awareness are key. Artists should know what their colleagues are working on – it’s inspiring and helps align everyone creatively. Our director, Steve Hudson, was a huge help here. With his background in acting and voice work – and being the scriptwriter – he brought a contagious energy. He was an engaging presence and always available, which really lifted the team. The performances that came out of that were just fantastic.

DP: How did you coordinate such a large international animation crew, and what processes ensured consistency?

David Nasser: I definitely wasn’t alone. Coordination is a team effort that includes production coordinators, a production manager, and line producer. Together, we manage timelines, workflows, and resources. As Juliane already mentioned, having clear structures and a transparent workflow is essential. Both long-term goals and weekly targets should be visible to the entire team. I believe in giving artists ownership -it makes the whole team stronger and more reliable.

Animation supervisors and leads play a huge role. They help manage the team and ensure the creative vision is understood and executed correctly.

Steve and I would typically brief a new sequence each week in live meetings with the supervisors and teams. These sessions were where we shared references, direction, and intentions. Ideally, the supervisor had already assigned the shots, so the artists came prepared and could ask questions early on.

Artists would then pitch their ideas, usually with video references. It’s a quick and effective way for us to get a sense of where a shot is going and give initial feedback. Once that’s aligned with the director’s intent, they move into blocking. That’s when we really refine the performance, strengthening the character, emotion, storytelling, or comedy. After that, the review process shifts more to me. I guide the team through the final notes and help bring shots across the finish line. We do daily live reviews.

Beyond shot work, I try to hold short classes or ask artists to share tips and tricks. It’s a great way to boost morale, share knowledge, and keep things fresh. Celebrating wins, whether through screenings or positive feedback, also helps keep the energy up.

DP: Can you share a particularly complex sequence and how the team solved it?

David Nasser: There were quite a few complex sequences, especially the musical numbers. Some went surprisingly smoothly, with the team exceeding our expectations.

But one of the toughest was Stitch Head’s escape from the castle, when Freakfinder tries to take him away in the hot air balloon. That sequence had everything: drama, comedy, emotion, and physical action. Timing, continuity, and performance were all critical. Plus, it was technically demanding.

I’m rarely worried if we’ll finish something, there’s always a solution. The real challenge is when time and budget run tight. I try to shield the team from that stress as much as I can, while working with production and supervisors to find ways to make up for lost time, ideally without compromising quality. That’s when a team’s resilience is truly tested.

This sequence was one of those moments. But the team at Assemblage – from the supervisors to every single artist – was outstanding. Thinking back on it still gives me goosebumps. It was an incredible collaboration, and sad to see it come to an end.

DP: In your view, what’s the “heartbeat” of Stitch Head, the animation choice that best carries the story?

David Nasser: That’s a tough one. The film has so many characters and perspectives, each with their own emotional arc. Stitch Head goes through a “coming-of-age” journey. He realizes that what he’s been searching for was always right in front of him, but he had to explore the world to see it.

Creature – my personal favorite – remains pure, innocent, and full of love. He doesn’t change much, but he experiences the world in his own way and I just love how he is reacting to it.

Two moments always stay with me. First, when Stitch Head tells Creature his life story. It’s the first time Creature feels sadness. It’s sweet, funny, and deeply touching. Second, the scene by the river, when Stitch Head decides to go into the village with Arabella, leaving his “bestest-best friend” behind. Both are subtle and emotional – my favorite kind of moments.

From a more dynamic perspective, I’d also highlight the musical number “Make ’Em Scream.” It’s bold, theatrical, and one of my favorites and really captures the playful side of the film while serving the story. Overall, Stitch Head was a wonderful experience. Great story, great team. I truly hope we get to do this again.

Handing off to Lighting

DP: Lighting defines the mood. Oliver, how did you work with Nico to balance gothic darkness with a family-friendly look?

Oliver Finkelde: Stephané had previously worked with Studio Rakete, which was a huge advantage in building trust and ensuring seamless communication. Walking the Rakete team through the color script while also focusing on the emotional arc of the story and of Stitch Head himself was vital to making sure key shots landed as close as possible in the first round. The studio brings a great combination of artistic vision and technical expertise, which made their contribution indispensable, especially in complex sequences involving FX and crowds, such as the laboratory and circus tent.

When it came to the look, particularly in the night sequences, we were very intentional about guiding the viewer’s eye through careful control of saturation, values, and contrast striking a balance between family-friendly visuals and that subtle sense of “being on the edge of your seat.” During color grading, this became especially crucial: we wanted to preserve the film’s soul and visual integrity, whether viewed in a theater or later at home.

DP: Nico, what was your lighting concept for Stitch Head? Which moods and palettes guided you?

Nico Rehberg: The look and feel for each sequence was guided by a very detailed mood board done by Stéphane and his team. We had at least one painted frame for each sequence. Combined with a briefing this gave a good direction for the desired final picture. Thus our job was mainly to translate this into 3d lighting and compositing while keeping everything on time and in budget. I know this does not sound very artistic, but in the end this is Steve’s and Stéphane‘s vision, crafted in close collaboration with Oliver, we are trying to bring to screen, not mine. I just fill in the gaps and give advice.

DP: Which tools and render engines did you use, and why?

Nico Rehberg: Our choice of tools was rather old school. Everything was centered around Maya, with Arnold as the renderer and Yeti for the hair and fur generation. Compositing was done in Nuke and Royal Render managed our farm. We have been using this setup for many projects now and really like how rock solid and mature all tools are. Technical render issues are nearly non-existent and render times are very predictable.

Generally I try to avoid overengineering our pipeline. Our workflow is simple and direct. We light shot by shot with simple tools to handle all the tedious and repetitive task. Even through all the input data to lighting is cached, we still enable the artists to use every tool available in Maya. Moving things around, tweaking shaders, fixing broken geometry or floating characters, adjusting FX – in our productions lighting is often more than just putting lights in a scene and setting up layers.

DP: What was the most challenging sequence to light, and how did you solve it?

Nico Rehberg: The most complex light setup was definitely the one in the laboratory. A lot of blinking lamps, lightning arcs, coils, glowing bulbs, luminous liquids and the big gun itself. The lighting was required to be very dynamic and directable. From Stéphane we had a detailed breakdown of the different stages the laboratory goes through during the reawakening sequences.

Since I try to avoid to animate lights in 3D as much as possible one artist, Phillipp Wibisono, broke this all down into a lot of layers and light AOVs which we then reassembled and animated in Nuke. This allowed much faster turnaround times while orchestrating these sequences. But it still meant a lot of layers and hundreds of lights, which sent the render times through the roof.

On top of that the lab features in the 2 longest shots in the movie. One with 1290 frames and the other 1436 frames (60 seconds!). We only had room for one try in rendering. There was no time for a re-render. The second shot was actually the last shot we rendered, occupying the farm for more than a week, right before delivery.

DP: Achieving visual consistency across teams and departments is notoriously hard. What steps did you take to make sure the look stayed cohesive all the way through?

Nico Rehberg: I think a good art direction is key here. A detailed mood board and precise briefings counter the fact that shots are often produced in a seemingly random order, but still have to fit together later. We had Stéphane, Steve, and Oliver on board until the very end, so we did reviews on the key shots in lighting and all the compositings with them. Furthermore I did a lot of the lighting setups myself and had the luxury to work with a small team of artist over a longer period of time instead of rushing the production.

My goal in lighting is to create a good basis for our compositing team around our Supervisor Tim Liebe. Often consistent lighting output is more desirable than perfectly lit shots that don’t fit together or leave no room for comp. And since we handled all the lighting and compositing here at Rakete we could often just talk about things. 

DP: Which scene are you most proud of, the one where light really tells the story?

Nico Rehberg: I don’t think there is a one shot to name. Light is but one part of the experience. There are very beautiful moments that are lit with only 2 directional lights and a dome, while other shots with tons of lights and lots of cheats to make them readable fly by in 14 frames.
But I’m very proud of the overall movie. I think we managed to really support the story, bathe the world of Stitch Head in fitting light and deliver high quality images. And all that within our time frame and limited budget.

DP: With comp workflows evolving fast, from OCIO color pipelines to more open interchange Formats. How did those tools shape your work on Stitch Head, and how do you see them changing lighting/comp production in the future?

Nico Rehberg: A lot of these newer big standards require changes and investments throughout the whole production pipeline of a project. And since our productions are usually split across multiple studios it is not easy to change. We have to find the least common denominator and agree on workflow very early. For example Stitch Head‘s toolchain was fixed around 2020. These new standards open a lot of possibilities, but they are not turnkey solutions. They increase the needed technical expertise at a studio a lot. We have to decide if they really fill a hole in our workflow or solve an existing problem.

OCIO is by now very well integrated in most programs and the basic workflows are more or less standardized. But USD on the other hand requires a vast amount of planning, restructuring and development. It is also still evolving rather fast. And although it seems to be inevitable in the future we haven’t really embraced it yet.
I am much more looking forward to the small brilliant things popping up that will make our lives easier. Like Cryptomatte, which was a game changer. Or Arnolds denoising. It solved a lot of our render problems.

DP: Oliver, how did lighting and compositing feed back into your supervision role, especially across distributed teams?

Oliver Finkelde: After Stephané walked the Lighting and Compositing Supervisor through the color script, we reviewed the key shots in context meaning together with their connecting sequences during the first rounds. This approach ensured visual continuity and strong support for the story arc before moving on to the rest of the sequence. Our goal was to have shots approved within roughly three review rounds. During lighting dailies, we also discussed which refine tasks made sense to hand off to Compositing, allowing us to keep shots moving efficiently and make the most productive use of the render farm.

Handing off to the Line Producer

DP: Behind all this creativity is production reality. Oliver, can you introduce Juliane, who kept the budget and schedule on track?

Oliver Finkelde: Sequences are never truly finished at some point, they simply have to go. Juliane was instrumental in keeping us focused on what mattered most in terms of schedule and budget. For me, maintaining a close collaboration with the line producer is always essential, and I couldn’t have asked for a better partner when it came to balancing art, schedule, and budget.

DP: Juliane, how did you structure the production schedule across Europe and India?

Juliane Walther: Over the course of five years of production, we had to coordinate a complex network of partners and independent studios – eleven of them dedicated solely to creating the final image. Different cultural approaches to work and communication, multiple time zones, and eventually the COVID-19 pandemic turned the production into both a logistical and a human challenge. The balance between creative ambition and production reality had to be constantly maintained: building structure without stifling creativity, while keeping workflows, technical standards, and overall quality consistent throughout the entire production.

That balance relied on clear organization and a good deal of sensitivity. We had to find time slots where teams in Germany, France, Luxembourg, Belgium, and India could actually talk to each other., sometimes even shifting working hours to make personal dialogue possible. Communication was about much more than exchanging information: it meant building trust, reading between the lines, and understanding the different dynamics of each team. Regular calls, from daily check-ins and reviews to several weekly production meetings kept the collaboration alive. And sometimes, it simply took getting on a train or a plane to be there in person and strengthen the sense of connection that video calls can’t fully replace.

For such a large network of contributors to work smoothly, clear structures and binding workflows were essential. From the planning stage onward, tasks were divided to keep dependencies as low as possible, allowing each step to be completed and handed over cleanly to the next department – reducing friction, overlap, and rework. The planning itself was continuously fine-tuned and adapted to the realities of production to stay flexible to new demands. Defined communication channels and established procedures ensured that questions and issues didn’t linger but were resolved quickly. Quality checks in the studios were closely monitored to maintain a consistent visual and technical level across the entire film.

For planning and tracking, we needed a tool accessible to all partners – of course we used ShotGrid. It became the central hub of production, containing all relevant data on assets, shots, and milestones. Everyone could see the current status in real time, track progress, and identify potential bottlenecks early. It kept everyone connected across all studios, time zones, and departments and helped maintain a shared understanding of where the project stood at any given moment. Beyond ShotGrid, we relied on structured spreadsheets, milestone plans, visual diagrams of the production status, several tracking sheets, and detailed production notes – all to ensure transparency, clarity, and traceability in everyday work.

In the end, all of this was above all teamwork – built on trust, openness, and a shared sense of purpose. Across borders and time zones, transparency was key so that everyone knew where the project stood and how their individual contribution fit into the bigger picture. Five years is a long time – teams, technology, and processes evolve, which makes it all the more important to keep motivation and cohesion alive. And sometimes, when all plans reached their limits, it simply came down to willpower, because otherwise, a project like this just wouldn’t get made (smiles).

DP: What was the toughest challenge in managing this co-production?

Juliane Walther: One of the toughest challenges was balancing creative ambition with production reality. When vision and production responsibility come from the same person, expectations can shift quickly – and it takes structure, diplomacy, and constant communication to keep everyone aligned. Translating artistic goals into achievable production steps without losing the creative essence became an ongoing exercise.

Adding to that complexity was the constellation of studios involved. Some of them had collaborated successfully on previous projects, which helped – there was already a shared understanding and rhythm. But for this film, we brought in several new partners, both large and small. Each studio came with its own expertise, experience level, and production culture. Bringing everyone onto the same page – creatively, technically, and in terms of pipeline – required patience, coordination, and a lot of mutual learning.

The film itself also came with its own set of complex creative and technical challenges. Just to name a few: our beloved OWGGAGOFFAKKOOKKK, or “Oggi,” with his tangle of dreadlocks; Ermintrude beneath glass and water, accompanied by her fish and moving air bubbles; the crowd sequences; the burning and collapsing circus tent; the beautifully abstract dance sequence that proved especially challenging for animation and the laboratory, which became quite a challenge for the lighting department. Each of these moments pushed us to explore new solutions, workflows, and creative approaches – and to find ways for all studios to contribute seamlessly to the same cinematic world.

What made it possible was the dedication of a truly great team – one that stayed motivated, kept looking for solutions, never gave up, and always tried to move forward together.

If I could, I would name many more team members who did an incredible job bringing this film on the screen. But talking about managing a big thank-you goes to our production team at Gringo Films. At peak times our production team included up to eight production managers and coordinators: Alexander Erkens, Larena Schwarzenberger, Dario Ramm, Dorothea Mersmann, Belen Heydt, Sandra Fendauer, Ester Vicente Ballester, and Marco Muñoz. Each was responsible for a specific department and brought valuable expertise to the table, which allowed me to rely on them completely and made for a professional and trusting collaboration.

I’d also like to mention my esteemed colleague Alexander Erkens, who together with our Animation Director David Nasser and Larena Schwarzenberger oversaw the animation department for more than 14 months. Their dedication, leadership, and teamwork were essential in keeping one of the largest departments running as smoothly as possible.

DP: How do you balance the creative team’s needs with financial constraints?

Juliane Walther: Balancing creative ambition with financial realities is a constant negotiation. From my point of view transparency is the key: everyone, from the creative team to the production partners, needs to understand and see where the limits are and why certain decisions have to be made. Once the “why” is clear, compromises are easier to accept and priorities become clearer. To make sure as much of the budget as possible ends up on screen, you have to choose your battles wisely.

I always try to start from the creative perspective: what is the core of the idea, what emotional or visual effect are we trying to achieve? Once that’s clear, we look for ways to make it happen efficiently – through clever use of existing assets, optimized workflows, or simply choosing the right moment to say, “Let’s render that once we’re sure.”

For example, we wanted the circus crowd to feel big and alive, but we also had to be smart about it. So we worked with a small set of base models and created as many variations as possible by changing accessories, clothing, and skin tones to make it look rich and diverse while keeping things manageable. Early pre-production tests and renderings quickly proved that this approach can work.

Of course, not every wish can be fulfilled, that’s part of the process. Already at the storyboard stage, some ideas had to be let go. There’s a limit to how many challenges a team can carry at once. In theory, everything is possible, but everything takes time, money, and motivation. In those moments, I prioritized maintaining consistent quality throughout the entire film.

With open communication, trust, and a shared understanding of the film’s priorities, limitations can turn into collaboration. That’s when the budget stops being a constraint and becomes the framework in which creativity truly thrives.

DP: If you had to explain your role to a film student, how would Stitch Head be your case study?

Juliane Walther: If I had to explain line producing to a film student, I’d say it’s about building bridges between creativity and logistics, between vision and reality, between all the people who make a film come alive. On Stitch Head, that meant coordinating an international team, keeping communication flowing across time zones and cultures, and turning artistic ideas into something that could actually be produced within given financial conditions and a fixed deadline.

A big part of the job is organization, yes, but even more it’s about understanding people and helping them work together toward the same goal. You’re constantly listening, translating needs, solving problems, and keeping an eye on both the big picture and the small details that make it work. The best moments for me are to find solutions with the team and when a plan comes together – when all the coordination and chaos somehow turn into a scene that works beautifully on screen.

Lighting & Comp

DP: What was your system to keep an overview of all the shots, especially in lighting and comp, where details change constantly?

Viola Lütten: Definitely ShotGrid, as the main source of information on the status of each shot. Luckily all other partners relied on Shotgrid as strongly as us, so the databank was constantly in a fresh and up-to-date state. The production period for LRC was in the end stretched over 17 months.

In the beginning Juliane had drawn up this huge plan, where it was determined what sq will be ready for us in which week. That looked good in theory, but in reality there were so many unexpected challenges and hurdles, that we at Studio Rakete kept our focus only on the foreseeable future, like the next six weeks or so. That made it easier to keep an overview of the current and the possible future workload. 

DP: What were the biggest challenges in terms of ingesting upstream and staying on schedule?

Viola Lütten: It is always a challenge to estimate the incoming workload, when you are so dependent on others to deliver. I kept a close look on the incoming frames per week and the needed average for the farm to keep it humming on a constant basis. We have been working with the same core team in Lighting and Compositing at Studio Rakete for quite some time, so I know how much shot input we will probably need for each individual artist per week to keep them busy and happy.

I may have communicated a slightly higher needed overall input to our partners, just to be disappointed as expected and get exactly what was necessary. I absolutely hate to lay off or postpone artists or ask for overtime – so the planning was very conservative.

We worked for a long time on this production with small teams. In the last three months we almost doubled the size of the Compositing team for the final sprint. No postponing or overtime involved. It also helped that Studio Rakete is an established animation studio with twenty years of experience and a strong administrative and technical team, so it was not necessary to build up new structures, but everyone could concentrate on the work.

DP: Which shot is your personal favorite, the one where you felt all the pieces came together?

Viola Lütten: Each shot that is approved becomes my personal favorite (laughs).

DP: Lighting and comp often sit at the end of the pipeline, where every upstream mistake shows up. How did you build trust and communication with upstream so the problems didn’t just pile up on your desk?

Viola Lütten: Isn’t it part of my job description, that problems will pile up on my desk? Just kidding. As mentioned previously the production had this huge sq planning for the whole production period, which was a great starting point for the weekly needed output and input of shots and frames. As in every production things didn’t quite go as planned.

The good thing was, that it did not result in blame games, but a rather great collaboration between all partners to find the hidden sq gems that can be moved up in priority and fill looming gaps. Studio Rakete had also worked with Fabrique d’Images on several feature projects in the past – so there was a lot of trust and respect on both sides. So not all partners were new.

Nevertheless my production team, Carrie Schilz and Marcel Tie, and I kept a close eye on the notes in ShotGrid to anticipate possible frustration flare-ups from our artists and sent a preventive “I see your pain, but please answer nicely”-text to them, before they reply to a particularly unnerving note in ShotGrid. It also helps that many artists still work on site at Studio Rakete and that there is a couch in my office: it is very comfy and everyone is welcome to sit down and share their troubles. Even the cleaning staff makes use of it.

DP: To all of you: what’s your favorite shot or moment in the film?

Oliver Finkelde: The hot air balloon sequence. It’s visually absolutely stunning. 

Juliane Walther: Among several proof-of-concept shots, we specifically chose the sequence where the hot air balloon rises over the wall. It served not only to test technical challenges, but also to explore the emotional impact and overall mood of the film and also to create an early “wow” image that would inspire both the team and our partners.

From the very beginning, we aimed to include proof-of-concept shots as early as possible, to identify potential issues before the start of shot production and to refine workflows for the later stages so that the actual shot production would run as smoothly as possible. Of course, problems always come up ;)
During Pre-Production we had a dedicated quality assurance team at Daywalker Studios for this purpose, ensuring that the ongoing production remained unaffected. In parallel, it also allowed us to test the pipeline early on and make necessary adjustments.

Juliane Walther: Also the Shot when Stitch Head is on the roof of the glass dome. It’s raining and he tries to catch the lightning with the umbrella – I already loved this shot when I saw it in the moodboard. 

Viola Lütten: I especially liked the shot at the end, where the crocodile-gorilla-monster breaks through the glass window in slo-mo. The shot has 1.436#, so when that one got its approval we had suddenly 1,1% more of the whole movie done.

Stéphane Lecocq: The lab sequences, and the village set, which looks really amazing.

Nico Rehberg: I revel in the long shot flying up the mountain into the lab at the beginning and how it turned out. It looks so seamless now, but is made of 3 different shots, multiple light setups that blend into each other, sets that don’t really fit together, tons of layers, masks, comp, spit and duct-tape. We could only render it once in 3D, and when a comp version was sent to the farm all the other artists knew the farm would be blocked for the rest of the day. 

DP: What’s one lesson you’ll carry into your next project?

Oliver Finkelde: Review and celebrate the beautiful work of departments more on the big screen. It’s so inspiring to see all the love and details that often get lost on the smaller screen. 

Nico Rehberg: Don’t be afraid. There were some rather bold sequences in the animatic. More challenging than we usually like. But Steve pushed them through and with some careful planning they turned out to be less complicated for us than I thought. In the end they present some of the most fun parts of the movie and it would have been a shame to not have them in there.

Juliane Walther:To move into 3D even earlier during the storyboard phase. It’s easy to get misled by the simplicity of a 2D storyboard, things can look perfectly fine on paper but reveal major challenges once you start translating them into space, movement, and lighting. And testing and rendering early as possible. Even on a smaller scale, it helps to spot problems before they grow and saves a lot of stress later in production. In short: the sooner you test, the smoother it can be finished. 

DP: If you had to give one piece of advice to newcomers who want to work in your role, what would it be?

Oliver Finkelde: Nature is the best teacher. Observe, listen and keep trying. Failing is part of the journey.

Nico Rehberg: Make a realistic plan before you start and stick to it. Be consistent in your output. Don’t relax or try to be too perfect just because the deadline is still far away. If you can not stick to your planned output in the first month, the next month will be worse and at some point you will start to sacrifice quality for speed. Crunch time can be avoided!

Viola Lütten: Please do your job properly. Don‘t half-ass it. Poor planning is the most avoidable cost driver.

Juliane Walther: Communication means a lot and there’s no such thing as the perfect plan. Plans change. People change. Productions evolve. What really matters is how well you communicate, adapt, and keep everyone moving in the same direction. In production you can’t just follow a schedule; you have to listen, translate, and connect people so the project can keep flowing – even when things shift. In the end, it’s less about controlling chaos and more about guiding it.

Stéphane Lecocq: To have precise designs, storyboard and moodboard, with attention to every details, don’t be afraid to have multiple angles for one ref, the more the teams will have, the easiest it will be for them.

DP: What’s next for you as individuals and as studios?

Oliver Finkelde: Looking ahead, I’m focused on wrapping up a VFX show at RiseFX while continuing to explore new rigging and animation approaches in Houdini. With Stitch Head about to be released, I want to express my gratitude to Sonja and Steve from Gringo Films, and to Juliane, for bringing me on board this incredible journey, their bold, out-of-the-box vision and approach for the project. Thank you to the whole Stitch Head team, you are amazing and of course, my deepest thanks go to RiseFX for their unwavering support.

Stéphane Lecocq: Art directing two other productions, still at fabrique d’images.

Viola Lütten: Studio Rakete is very happy to work on the sequel of „The Amazing Maurice“, which is currently in production.

Nico Rehberg: After the movie is before the next movie. Or actually right in the production of the next movie. So I will stick around at Rakete and make sure the next movies look just as appealing as Stitch Head. 

Juliane Walther: I’m taking a short family break and giving new ideas a little time to grow in the background. The next project is never too far away.

The post Stitch Head! first appeared on DIGITAL PRODUCTION and was written by Bela Beier.

by Florian Genal

In the 2019 cinema film “Fritzi – eine Wendewundergeschichte” (“Fritzi – a miracle story of change”), viewers can experience an emotional journey through time. The film tells the story of the peaceful revolution in Leipzig in 1989 from the perspective of a 12-year-old girl called Fritzi. Five years later, the series “Fritzi und Sophie – Grenzenlose Freundschaft” (later F&S) is produced under the direction of Ralf Kukula, Matthias Bruhn and Thomas Meyer-Hermann and a team full of passion. In the eight 22-minute episodes, the TV series explores the themes of everyday life in the GDR and tells the story of Fritzi’s friend Sophie, who dares to make the dramatic escape to the West with her mother in order to start a new life in West Germany.

Pipeline overview

The cinema film was an excellent reference for the set models, animators and lighting artists. We wanted to capture the look and atmosphere of the cinema film, but at the same time create our own look that probably didn’t exist before. For the body animations we used motion capture and created special 2D toon shaders. Blender’s Grease Pencil Outliner, which was integrated into Blender just in time for the start of production, was used for the outlines. Everything was produced at 50 frames per second. There are a lot of large sets in the series, which were realised in 3D with great attention to detail. Internally, we treated the sets like an important character, as the realistic backgrounds are a very important part of the emotional story. 170 characters, including twelve main characters, plus demonstrators, animals and vehicles were a huge challenge for everyone involved. And that’s why we thought: let the individual departments have their say!

Storyboard/Animatics

Hi, I’m Alex Tiedtke and I was responsible for the storyboards for Fritzi & Sophie. I admit it: I’ve been around since the Stone Age. When I did my first jobs at the end of the 90s, we were still using pen and paper. Then soon came the graphics tablets, then programmes like Toonboom Storyboard with a timeline, sound and the ability to move different layers past the camera at different speeds for an exciting parallax effect. For the most part, however, this has remained the case to this day. Even in 2024, storyboards are usually quick digital sketches of characters in the most expressive poses and expressions possible against a flat drawing plane with hints of background.

And then came Fritzi & Sophie

The production team approached me with the idea of using Blender in the storyboard. At the time, I had absolutely no idea what was in store for me – but as a good freelancer, my answer was of course: “Sure – great idea!”

I started with a small test scene. There was already a blocking set for it: Sophie’s room with balcony and tree house tree. Install Blender, watch a few tutorials, how hard can it be? Well… let me put it this way: for someone who has never opened 3D software before, the beginning is, to put it mildly, a challenge. But: I was also quickly a little enamoured. The ability to move freely around a set with a virtual camera? Fantastic! In a conventional board, if things go well, I have a nice drawing as a reference from one or maybe two angles. In a 3D set, I can whizz around freely and literally discover the most beautiful shot.

And the best thing? Everything is just right: Perspectives, sizes and focal lengths are effortlessly correct. For me as a draughtsman, the Grease Pencil was the obvious choice for the figures and therefore my first approach. However, it quickly became clear that there were a few advantages of board work in 3D space that remained unutilised. If I wanted to change a camera angle significantly, my drawing would no longer work because the perspective would be distorted.

When my old friend and Blender Bunny Robert Schlunze joined the team, he suggested using animatable proxy characters instead. A simple Mixamo rig, a few quick and dirty Blendshapes and an eye rig tacked on for the facial expressions – completely sufficient for what we had in mind. The modelling department quickly delivered the necessary assets and we staged our 180 minutes in one blender file per location and sequence with sound, key poses, expressions, camera settings and movements. Where something was missing, quick kitbashing helped. If the door handle was huge, a tree in the way or a street 40 metres wide – even we noobs could remove, scale or move objects in the simple low-poly sets. The blocking sets adapted to the production were then sent back to the relevant departments as a “bulletproof” basis for development.

Along the way, we quickly realised other advantages. First of all, of course, the new freedom to free the camera in selected situations from the usual 2D right-left-up-down-in-out. A steady cam with the typical slight swaying – just one modifier away. Panning and moving, flying along, rotating around characters, attaching the camera to running characters or moving cars… everything works if it is needed and makes sense.

We were able to save every successful pose in a library and often even transfer it to other characters. A few lights simulate moods or times of day or help the composition with light and shadow. Crowds of people, which play a major role in a story set against the backdrop of the events of the peaceful revolution in 1989, can be realised in a dreamlike way through the magic of instances and particle systems. Continuity almost takes care of itself. Once an asset has been positioned, it remains in the same place in all subsequent settings. Complex sets, such as the entire St Nicholas Church full of people with its perspective-challenging pointed arch vaults? No problem.

I felt like a child in a sweet shop… So many new possibilities, so few limits. The tedious suddenly became easy. I had a lot of fun – even explaining our little trail through the Blender jungle to the new boarders. Since Fritzi & Sophie, I’ve been able to use this way of working in other German animated film productions. Let’s see… maybe this new storyboard production method will continue to develop in future 3D productions. I would be delighted!

Pipeline

Hi, my name is Elias Schwarze. As Technical Director, I was able to develop the pipeline and support the project in many technical aspects. I was also one of the operators during the performance capture. This project was a challenge in many ways. A TV series consisting of almost 180 minutes of 3D animation, the development of a new pipeline, performance capture, and the switch from 2D to 3D.

Fritzi & Sophie meant all this and much more. Switching tools and software was not enough, we also had to rethink and learn new ways of working. Automation was a much bigger issue for the pipeline than before, as was the development of better tools for our artists that were specifically tailored to production. Together with my colleagues Robert Lehmann and Michael Schieber, we developed a series of add-ons and plugins that we continually optimised and improved with growing insight and feedback directly from our artists. We wanted to leave as many dull and repetitive tasks as possible to the scripts so that our artists could concentrate even more on their actual creative work.

We also redesigned and modernised the backend for this production. The aim was not only to reduce unnecessary friction and costs, but also to become less dependent on third-party providers and at the same time to have more security and control over our own data. We had our own central server, which involved more administrative effort than rented cloud solutions, but also offered us more freedom and flexibility as we had full control over software and configuration. In future, we want to make our infrastructure even more flexible and efficient with the help of VPS (Virtual Private Servers).

The three cornerstones of the backend were running on our server: Mattermost, Kitsu and Perforce Helix Core. Mattermost is an open-source alternative to Slack, which greatly reduced our email-heavy conversations and enabled the quick, informal exchange of information within the team. Since we host Mattermost ourselves, we know at all times that all files exchanged always remain in our hands and on our server. Then came Kitsu, an open-source production tracker and an alternative to Autodesk Shotgrid and ftrack. Kitsu has developed enormously in recent years, from which we have also been able to benefit.

The third cornerstone was Helix Core, the version control system from Perforce. We needed a system that could manage, back up and version all our files. After all, we had to avoid losses in the event of errors, and no file could be edited by several artists at the same time. We found the solution in the games industry, which had already solved these problems many years ago. Since every version of every file is saved in Perforce, it was always ensured that there were never any major losses if mistakes were made. At the same time, it was essential to keep our production data accessible and up-to-date at three different production sites. We were also able to provide our freelancers and artists working from home with all the project data they needed at all times.

Performance Capture

The use of performance capture was also new for us. I supported the shoot technically as an operator. Thanks to the Xsens Link system we chose, we were very flexible in our choice of filming locations, as this system can also be used on the move. This proved to be helpful during the course of production, as we sometimes didn’t have enough space in our locations to show particularly long walkways. In these cases, we were able to simply go out and capture outside. With the help of a laptop or iPad, the technology could also be operated on the move.

For both our director and our actors, most of whom have a background in theatre, the work took some getting used to. But after the first shooting block, we had already developed a strong routine that allowed us to work through our tight shooting schedule with great efficiency. We moved more and more from shot-based work to planned sequences. On the one hand, this was an economic advantage, on the other hand, this way of working helped the actors as they were able to develop a better “flow” in planned sequences and were not constantly interrupted, as would have been the case with the shot-based workflow.

Although this meant that we had to adjust some scenes to the new length after the performance capture if the actor had played faster or slower than the animatic, it also meant that the actors were able to bring in more of their own ideas and also their own handwriting, which is actually more typical of live-action film. Overall, we went into this production with a lot of unknown variables. But now that we have successfully completed Fritzi & Sophie, we will continue to optimise and develop our pipeline so that we can realise further projects here in Central Germany more efficiently and with higher quality in the future.

Set Design

My name is Paul Siedler and I was responsible for the sets and lighting at Fritzi & Sophie as a supervisor. With over 70 different sets, the Fritzi & Sophie series includes a very wide range of environments that had to be created. These range from small indoor scenes such as flats and tents to large cities and landscapes, some of which are very different.

The two main categories into which we divided the sets are East and West Germany. These had to differ visually in order to show the scarcity in the East on the one hand and the abundance in the West on the other. The look of the East was largely determined by the cinema film Fritzi – Eine Wendewundergeschichte. However, due to the sheer number of sets compared to the cinema film, we had to choose a somewhat simpler form of representation in many places in order to keep to the production time frame. In contrast, the western sets, of which there were none in the cinema film, were new. The biggest contrast here was in the depiction of the cities, with large, sprawling buildings and crowded shops.

The biggest challenge was that we used the real-time renderer “EEVEE” for the entire production. Even though real-time rendering on its own is no longer an excessive problem in many areas these days, the combination with the outlines caused us quite a lot of difficulties. Especially with large sets such as cities and landscapes, we had to adapt workflows several times so that they could still be rendered with outlines. In the end, we treated each asset like a computer game and built it as low-poly as possible, as the performance of the outlines depends mainly on the amount of polygons they are applied to.

The trees made a particularly large number of changes during development. It quickly became apparent that large landscape scenes with lots of trees and outlines were impossible to render because Blender simply crashed. The solution for large quantities of trees was an add-on called “Instant Impostors”, which was able to project our trees onto a simple geometry in three dimensions. For the buildings, we developed a simple modular system that uses the Tissue tool to apply complex building parts to a simple basic framework. This framework could be customised as required, allowing variations for buildings to be created quickly.

In terms of lighting, we also had a very wide range. This meant that in addition to the large number of different sets, there were also different lighting situations in which a set could be located, from sunrise, day, to sunset and night. For the right performance boost when rendering, we also used little tricks from time to time, for example by rendering parts of a city as images and distributing them over several simple surfaces in the background.

Layout

My name is Gideon Volle and I was layout supervisor at Fritzi & Sophie. Welcome to the wonderful world of layout! This is where the artistic specifications of the 3D storyboard and the digital assets are used to create the basis for the animation. In our case, we focussed particularly on the use of mocap data and the transformation of the 3D characters. A crucial step in the layout was the continuation of the 3D storyboard scenes already created in Blender.

The placeholders for sets and props were replaced by the final assets, and even the characters were given their final rigs. But “final” is relative, especially in the world of animation where everything is constantly in motion. Our assets were constantly updated during the layout process, which required regular updates via our server tool “Perforce”. This occasionally led to surprising results – from deformed characters to altered dimensions of the environment. Close co-operation with other departments helped us to overcome these challenges. In addition, we imported thousands of mocap data capturing every step and gesture of our characters. This data had to be carefully transferred to our characters’ rigs, which was largely automated thanks to scripts. Choosing the best take was an art in itself, and although the director had already made preliminary decisions, we often had to combine different shots to achieve the desired action.

The mocap data served as the basis for the movements of our characters in the layout. Even though adjustments were still needed in the animation, they provided a solid starting point. The original mocap data was then packaged into an NLA layer where we went through it, filtered out the best take and adjusted it to match the timing of the storyboard. This process was crucial to achieve the right dynamics and expressiveness in the scenes. The sequences were timed according to the storyboard, with particularly complex scenes with many characters and long walkways presenting a real challenge. The camera positions were defined and animated in accordance with the storyboard and in consultation with the director in order to achieve the desired dramaturgy. At the end of the layout process, a fairly detailed version of the sequence was created that contained all the elements – characters, movements and camera angles. From here, the animation could take its course and the magic of the animated film could unfold. The layout may seem like a puzzle where every piece has to be carefully placed, but in the end it becomes the basis for an animated masterpiece.

Of course, surprises are not uncommon in the production process. A Fritzi with 20-metre legs can make you smile, but ultimately it is this variety of challenges and improvisations that make every project unique. We are inspired by them, learn from our mistakes and grow together with our characters and stories.

Motion capture

My name is Patrick Knott. Full-time generalist, whether 3D, compositing or IT. At Fritzi & Sophie, I was responsible for many areas from start to finish: development of the animation & cleanup pipeline, prop modelling, rigging of props & vehicles, further development of the character rigs and compositing.

At the beginning of the production, it quickly became clear that in order to stay within budget and realise the animation output in the shortest possible time frame, we decided to use the motion capture process. All the studios involved already had a lot of experience in all forms of animation, be it 2D pose-to-pose, lay animation or 3D animation. But motion capture was completely new territory for most of them. The decision quickly gave rise to a host of new questions: Which motion capture system should be considered? How good and clean is motion capture anyway? What can be captured? Complex body movements, steps on a solid surface, interaction between several actors, the finest finger movements, even facial expressions and mouth movements?

Where are the limits? Can the real-captured movements be combined with our desired 2D look at all or does this mean that any cartoon charm that an animator would normally conjure up in the movements is lost? We did a lot of tests and tried out various systems. Rokoko, OptiTrack, Captury, X-Sense. The initial results were sobering. The movements often turned out to be too imprecise and unclean. Shaking characters, legs hovering above the ground, drifting, penetration and clipping during actions between several characters or in interaction with props.

Whether multi-camera-based, using depth sensors, markerless or with tracking dots or even tracking devices with in-built accelerometers, gyroscope sensors, etc. in various price ranges. Sometimes there were enormous problems with drift, sometimes with the precision of finger movements. It quickly became clear that there is no such thing as the “perfect system”. Simply “capture” and the animation is finished, unfortunately it wasn’t that simple. Even if you capture with the most elaborate setup under ideal shooting conditions. Unfortunately, a lot of the problems described above can only be fixed afterwards: CleanUp is the magic word here. However, this doesn’t have much to do with magic, but with a lot of painstaking detail work.

Animation

My name is Florian Genal and I was the animation supervisor for Fritzi & Sophie, among other things. After the layout department had worked its way through the large amount of mocap data for the actors, put everything in the right position and timed it, it was the animation department’s turn. The animations were divided into different areas and then edited and animated in sequence. First came the body animations, most of which were corrections to the mo-cap data.

As the characters have different proportions to our human actors due to the “Fritzification”, i.e. the design of our production series, there were always intersections, e.g. of arms and hands. But the animators also had to help out with contacts between the characters, such as hugs or interactions with objects. The motion capture caused problems, especially with smooth movements. To compensate for this, however, we got very nice, subtle movements that would not normally be animated in this way. For example, a small head movement with a mischievous smile. Our actors delivered a great performance.

We used the mo-cap data exclusively for the body animations, except for the fingers. After many attempts and a lot of time invested, we decided against the mo-cap fingers as the correction effort was too great. A hand simply has too many bones. The then new Asset Browser in Blender was just what we needed. It had replaced the previous Poselibrary, came with a few additional extras and was a bit of a game changer. We created hand poses, which then formed the basis for the finger animations. We did the same with the facial animations and lipsyncs. We created pose libraries for all the main and supporting characters.

It was very helpful that we were able to use the video recordings from the mo-cap shoot as references for the facial animations. One of the main characters is a dog called Sputnik. We animated Sputnik and all the other animals in Fritzi & Sophie by hand. In the final step, we added additional animations, such as vehicles, but also additional characters to make the backgrounds more vivid.

Body CleanUp in Blender with the help of the NLA

The question quickly arose as to how the CleanUp, i.e. fixing and improving the motion capture data, could be technically implemented and how we could make the process as simple and efficient as possible for our CleanUp animators. We came up with the idea of placing a kind of “correction layer” over the mo-cap data in order to be able to make corrective movements here. For example, lowering an arm slightly at a certain point or rotating the legs slightly higher at a certain point if, for example, the feet clip through the floor.

The obvious solution would be to make such corrections directly in the Mo-Cap data. However, this usually turns out to be far too time-consuming, as keys are created in the mo-cap data for every bone, every character rig on every single frame for every translation and rotation axis! Although it is possible to use tools available in Blender such as “Proportional Editing”, it was clear that working in the original mo-cap data had to be reduced to the absolute minimum for an efficient cleanup. For example, the smoothing of shaky and jerky movements in the mo-cap curves. Also the phased removal of mo-cap keys, e.g. to incorporate stagnation phases between movements. Or temporal shortening or lengthening of movements via graph scaling.

For all other corrections, we utilised Blender’s built-in Nonlinear Animation (NLA) system. This makes it possible to create animations on several levels. Just as you can layer audio tracks in a DAW (audio editing programme), you can layer several tracks with animation data (keys) in Blender and combine them with each other.

The Blender file from the layout was sent to the cleanup team. In a first step, we smoothed the movements directly in the Mo-Cap data and, if necessary, made temporal corrections. We then used the mo-cap data to create an NLA layer in which we made all our correction movements. Everything was possible here, from counter-animating incorrect mo-cap movements to incorporating completely new movements that had been lost or even forgotten in the mo-cap. This involved a lot of detailed work and corrections, sometimes frame by frame. However, it also contained the same building blocks and work that occur in every classic animation. It was therefore crucial for us that our cleanup team consisted exclusively of members with animation experience.

Mo-Cap cleanup rig instead of the NLA

Unfortunately, it quickly became apparent that the NLA system also had some limitations and problems. This gave us an ingenious idea: what if you didn’t have to separate the animation using a layer system, but could simply use additional bones in the rig that were specially designed for the cleanup? You would simply duplicate each bone and have the option of applying the cleanup to it. This way, both information, mo-cap and correction, would be available at all times. In addition, you can always see the effect of both levels at the same time: a “mo-cap cleanup rig”! In the months that followed, this theory was further developed. One major problem, for example, was how to couple two bones with different values (mo-cap and correction) so that both bones always “represented” both values and did not simply “drift apart” and take over the local axis of each other without causing an endless-loop disaster. Using a specially devised technique, it was possible to transfer information from bone A to bone B (and vice versa) without getting it back to bone A again. Such a value feedback could be prevented by a special use of “drivers”, which is why I also liked to call this technique the “driver diode”. Ultimately a kind of sleight of hand. However, at the time, this method was an important breakthrough for the realisation of our cleanup rig.

Expansion of the mo-cap cleanup rig

Over time, we were able to develop a few more techniques and Patrick Knott expanded our rig. These include the option of displaying only the raw Mo-Cap data at any time in addition to the cleaned data, so that a before-and-after comparison can be made at any time.

A further control level was also established within the arms, on which the correction could be made using tracking poles (similar to an IK rig control), as mo-cap data is unfortunately always only FK. In fact, attempts were even made to transfer the Mo-Cap data to the IK arms or even to completely merge IK and FK kinematics within the rig architecture, but in the end this was more like reaching for the stars.

Another interesting function was to be able to pause or freeze mo-cap movements for individual bones as well as the entire rig and to be able to fade back into the mo-cap animation at a desired point in time. Or to be able to amplify or attenuate the mo-cap data. By the end of the development period, our rigs had many new functions that were specially tailored to the needs of motion capture editing. To keep these functions as tangible and organised as possible for the animator, I built a custom UI consisting of bones into the rig. This actually turned out to be particularly useful and innovative. In fact, this was born purely out of necessity, as all our programmers were busy elsewhere at the time. For me as a rigger, this was the easiest way to accommodate the functionalities in a timely manner. Just try it out and do it. That’s what this production taught me.

Crowd simulation

We are Stefan Kreller and Martin Schmidt from Raumkapsel in Kassel and were responsible for realising the crowd simulations for Fritzi & Sophie. This mainly involved the Monday demonstrations, which take place towards the end of the series. This included many shots in which crowd characters can be seen directly behind the main characters. Therefore, great importance was attached to the possibility of manual adjustments to animations and the positions of individual models in order to be able to respond to specific directorial requirements in individual shots. For this reason, we developed a procedural crowd pipeline in Blender in which all time-consuming steps could be automated using Python scripts.

Workflow

For the basic crowd simulation, we used the Blender add-on CrowdSim3D(crowdsim3d.com). For the crowd, 21 individual character models were available, each with several animations from motion capture data. The models were converted into Alembic caches, as Blender was able to process them relatively smoothly in large numbers.

After simulating the rough crowd movement with a proxy model, the Alembic caches were automatically imported, randomly distributed to the simulated crowd agents and the original materials were linked. In a further step, the speed of the walk cycles was calculated and adapted to the forward movement of the crowd. To ensure the necessary variation within the crowd, the models were assigned random colours from predefined colour palettes for hair, tops and trousers during import. To enable subsequent manual adjustment of the colours without having to change the linked materials, the colours were saved in custom properties on each character, which the material could access via an attribute node. This meant that corrections could be applied directly in the final file, independently of the linked materials. Collection instances of larger groups of people were used to fill the backgrounds of particularly large crowd scenes, sometimes with over 10,000 characters, without affecting performance too much.

Challenges

An unforeseen problem was that Blender on Windows systems only allows about 500 Alembic caches at the same time. This limit, imposed by the standard C library, could only be circumvented during rendering by using a self-compiled version of Blender with a higher limit. The biggest challenge, however, was to keep the entire crowd workflow procedural so that changes to models, materials and animations were possible at any time. In this way, we realised a total of around 10 minutes of crowd sequences.

Line Art

My name is Tsuyo Grohé and I was part of the line art team at Fritzi & Sophie. Line art is an essential part of classic cartoon looks and therefore important to achieve certain non-photorealistic rendering (NPR) styles in 3D. As a rule, line art consists of the outline, which visually separates objects from the background, and additional lines within the object surface.

Line art in Blender?

In Blender, you can use the Grease Pencil Tool to draw lines by hand for a hybrid 2.5D solution. Alternatively, the software can also create line art automatically. There are various techniques for this, all of which have their advantages and disadvantages. Here are three of the most common methods: Inverted Hull is the simplest way to create an outline for objects. This approach works by providing an enlarged duplicate of the mesh (hence the word “hull”) with a suitable material and inverted normals. The Solidify modifier can be used for this in Blender.

Inverted Hull offers comparatively the fastest performance and is visible in real time in the viewport. However, it usually looks poorer in quality than other types of line art, especially in close-ups, and allows little or no control over details such as the lines within an object. Freestyle is Blender’s own NPR engine for creating line art. Because Freestyle works with post-processing, the line art is only visible in the rendering result. With the right settings, however, it can deliver a visually higher quality result than the inverted hull.

In addition to the outer contour, Freestyle also supports lines that are defined by a specific crease angle, i.e. the angle between two neighbouring faces, as well as edges or faces marked on the mesh. When creating the lines, all objects within a view layer are loaded and therefore a relatively large amount of memory is required for rendering. With Blender version 2.93, a new line art modifier for Grease Pencil was added, which can also generate outlines. Some of its options are based on those of the Freestyle technique, but the biggest innovation is that Grease Pencil Line Art can be viewed directly in the viewport and offers more flexible options. Unlike Freestyle, Grease Pencil is vector-based instead of pixel-based.

The TV series Fritzi & Sophie is one of the first major productions in Germany to use the line art modifier. In order to achieve a line style in 3D that comes as close as possible to the template of the 2D animated cinema film Fritzi – eine Wendewundergeschichte (2019), our team developed its own add-ons for our workflow. This allowed us to create the line art object with the required modifiers with just a few clicks each time.

In the next step, we sorted all objects in the scene into different collections in order to assign properties for the line art. This included whether an object should have line art and what it should look like. We separated the line art of the characters from that of the environment to enable precise editing in compositing afterwards. The lines were coloured in post-processing and given a texture that makes them look as if they were drawn with chalk.

It is also crucial that the lines become thinner the further away they are from the camera. We set the line thickness of the environments to one value per shot, but the setup for the characters is controlled by drivers that adapt to the movement of the characters through the room. We had three different line widths for each. For example, the outlines around the characters’ eyes are thinner than those of the body and the outlines of the teeth and tongue are even thinner.

Like probably every major production, this one was not without its challenges. For the line art, these were mostly unwanted lines that appeared where we didn’t want them. Or vice versa: lines that were not visible even though they were supposed to be there. This was caused by all kinds of factors: for example, a line automatically appeared on a Sharp Edge every time. Or the line art settings of an object were set incorrectly, or a hidden object was blocking the line, and so on. In addition, the lines sometimes started to flicker. In most cases, the flickering was caused by the booleans used to animate the characters’ mouths.

Even if an object is too far away from the World Origin, the lines start to flicker. Or to be more precise: the entire mesh starts to flicker and the lines flicker with it. For example, there are a few scenes in the series in which long car journeys can be seen. And because we initially animated entire sequences in one go in Blender instead of individual shots, the car was sometimes many kilometres away from the source. This results in rounding errors and the modifiers can no longer keep up. This can be easily recreated in Blender by moving a mesh with a subdivision modifier on it a few thousand kilometres away.

Incidentally, this phenomenon not only affects Blender, but also other 3D programmes that work with a coordinate system of this type. In a few cases, we had to move the entire set to fix the problems. In addition, there were some serious performance bottlenecks – mainly caused by too many polygons, materials and particle systems. Once the scene reaches a certain level of complexity, the frame rate in the playback drops sharply, even causing the programme to crash. The whole thing is further slowed down if line art is also calculated. This is because the lines for each individual vertex within the camera view are calculated per frame. Before the line art can be displayed in real time, it must be baked.

This means that it is converted into a mesh and then no longer adapts to changes in the animation or camera. However, it can now be treated like a mesh, i.e. it can be rendered on a different view layer than the rest of the scene. To improve performance, we have decided to dispense with line art for objects that are either never directly visible in the frame or are far away from the camera. So that we could still use lines on the trees in large forest scenes, we used the inverted hull method in some places. By default, it is not possible in Blender to render an inverted hull separately from the underlying mesh. Only with the help of a script were we able to control the visibility of the solidify modifier per view layer.

Grease Pencil Line Art is currently the most flexible method of creating line art in Blender. Precisely because the modifier allows many possible options, it tends to be difficult for beginners to understand. However, as a comparatively new tool, it is constantly being developed further. For example, the Grease Pencil will be revised for upcoming Blender versions and the Line Art Modifier will become its own Geometry Node in the future.

Compositing

My name is Jan Mildner and I was compositing supervisor and responsible for the final look of Fritzi & Sophie. Rendering was done in single frames with Blender Eevee in OpenEXR Multilayer format. Passes were AmbOcc, Beauty, Plant-Alpha, Z-Pass, Sky, Char-Vectorlight, Char-Diffuselight, Volume, Spectacle-Alpha, Char-Normals, Environment Outlines, Char-Outlines, FX (Rain/Fire) and Cryptomatte. The comp was made in Fusion (Studio) by Blackmagic.

This is where the beauty pass with the sky and the outlines for the characters and backgrounds were put together. These were given the desired look. The character outlines were monochrome and were given a structure similar to that of a hand-drawn chalk engraving. The outlines of the backgrounds took on the colours of the objects, so they are in the final image. Neither was possible in Blender and could therefore only be done in the comp.

As our 2D shader meant that the characters were not affected by the set lighting and were always displayed as if it were a summer day by default, all the necessary shot adjustments had to be made in the comp. Additional highlights and rimlights could be added to the characters via the character normals if required. Grading the night shots proved to be very time-consuming because the colours in the entire shot were often not blue enough and skin tones moved strongly towards red as soon as they were darkened. Of course, some depth of field also came into play, which was also added entirely in the comp. Headlights, torches and streetlights (i.e. all the highlights) were also not visible on the characters and had to be recreated in the comp.

Frequent problems included clipping errors, for example when a shirt blinks through a jacket, or when the booleans of the faces repeatedly showed unwanted deflections. Transparent objects such as spectacle lenses and windows were also a challenge. On the one hand, they were very noisy and the Z-pass, which was used for grading and depth of field, perceived transparent objects as opaque, meaning it could not look “behind” them. The partial transparency of the grasses made clean grading difficult.

The most important tool in the comp in this project were the cryptomats, where individual elements in the image could be easily selected via Gizmo and used as a mat. You could either select entire objects and characters or just individual parts such as hair. Especially with moving images, it is essential not to have to mask certain areas manually using a polyline or similar in order to colour, fix or change them precisely. For the basis of the final grading, all shots were exported in single images as EXR single layer with 16-bit float linear (without LUT).

Conclusion

Fritzi & Sophie was an interesting and extraordinary project for me – and I really enjoyed working with the team – so once again a big thank you to all my colleagues, and I’m looking forward to the next project with you!

Florian Genal is a freelance 3D artist/generalist. After training in business administration, he studied character animation at Animaton Mentor and has worked on numerous projects. www.FlojoART.com

---

• Director: Ralf Kukula, Matthias Bruhn, Thomas Meyer-Hermann
• Producer: Ralf Kukula, Richard Lutterbeck, Thomas Meyer-Hermann
• Production management/company management: Christoph Kukula
• Production management: Christian Asmussen
• Co-producers: ARD, MDR, WDR, SWR
• Screenplay: Beate Völcker
• Dramaturg: Martin Muser
• Art Director & 3D Storyboard Supervisor: Alex Tiedtke
• Technical Director: Elias Schwarze
• CG Supervisor & Blender Mentor: Christoph Werner
• Character Design: Olaf Kamin
• 3D Modeler Character Lead: Fynn Große-Bley
• Rig Supervisor: Sophie Fink
• Prop Supervisor: Mikola Debik
• Vehicle Supervisor: Patrick Knott
• Set Design Supervisor: Sven Höffer
• Set & Lighting Supervisor: Paul Siedler
• Layout Supervisor: Gideon Volle
• Animation Supervisor: Florian Genal
• Crowd Animation Supervisor Space Capsule Animation: Stefan Kreller
• Outlining Supervisor: Tsuyo Grohé
• Render Consultancy Virtual Republic: Michael Klein, Steffen Duenner
• Rendering Supervisor: Josie Steinmetz
• Compositing Supervisor: Jan Mildner
• Editing: Francie Liebschner
• Editing & Grading: Stefan Urlaß
• Sound Design & Mixing: Florian Marquardt
• Music: André Dziezuk
• Production Coordination: Bianca Just, Sabine Kraft and Moana Klein
• Editorial management: Anke Lindemann
• funded by: BKM/FFA/MDM, Film- und Medienstiftung NRW, MFG Baden-Württemberg, Additional Funding Saxony
• Production studios: Balance Film GmbH, TrickStudio Lutterbeck GmbH, Studio FILM BILDER GmbH

Thanks to the entire Fritzi team!

The post Fritzi & Sophie first appeared on DIGITAL PRODUCTION and was written by Bela Beier.