There is so much buzz around artificial intelligence in animation today that it seems the industry is on the verge of tectonic shifts. Some proclaim a revolution and the death of the profession, others shrug skeptically and continue drawing frame by frame. The truth, as usual, lies somewhere in the middle. In studios producing complex projects for television channels and major brands, AI has long become a working tool—not replacing people, but integrating into processes. The question is how exactly to build this symbiosis so that technologies enhance creativity rather than suppress it. Here is how it works in our studio and among our peers.
The temptation to fully delegate production to neural networks is understandable: studios want to speed up, reduce costs, and eliminate the headache of staffing. But in practice, a complete rejection of human involvement in animation produces results that are difficult to call art. And a complete rejection of technology in the modern world means losing competitive advantages. Therefore, studios working on serious projects take a third path: finding points where neural networks are genuinely useful and integrating them into human workflows. Clients often think that implementing AI is like pressing a "make it beautiful" button. In reality, it is a complex process of restructuring production workflows. Neural networks do not replace work stages; they redistribute the workload within them. Some things accelerate, some become cheaper, but the final quality still depends on people. Because only a human understands what exactly the client needs, what the brand's essence is, and what emotion each scene should evoke. A neural network can suggest a thousand options, but only a professional can choose the right one—and refine it manually if necessary. In our studio, we use AI tools at various stages but never delegate key creative decisions to them. For example, neural networks excel at generating backgrounds and environments for secondary scenes. This saves hours of artists' work, allowing them to focus on main characters and key frames. Or, at the pre-production stage, AI helps quickly explore visual ideas: styles, color palettes, compositions. The team receives dozens of references in a couple of minutes instead of searching for them manually. This makes it possible to reach an agreement with the client on the direction faster and avoid lengthy approval processes.
The greatest potential of AI is at the stage when the project is just taking shape. Here, neural networks act as powerful accelerators of thought processes. A director or artist can input a text description of a scene and receive a visualization that previously would have required an illustrator's work. Of course, this is a draft that will later be reworked, but it provides a sense of direction. And this is much faster than describing ideas in words and waiting for them to be drawn.
Previously, to assemble a mood board for a project, artists spent days and sometimes weeks. They needed to find dozens of images online that matched the style and mood, process them, and systematize them. Now, a neural network does this in minutesbased on a text query. The team can explore dozens of directions in a single day and choose the one that best aligns with the project's objectives. This is especially important for commercial orders where approval speed is often critical. In projects for major brands, we frequently use AI to create character variations. Clients may find it difficult to envision how a character will look in motion or from different angles. A neural network can generate numerous options in a short time, allowing the client to choose the closest match. Then the artist renders it in detail, with all the nuances that AI could not account for. This reduces the number of iterations and makes the approval process more comfortable for all parties.
Storyboard creation is one of the most labor-intensive stages of pre-production, especially for complex projects with many scenes. Neural networks cannot replace a professional storyboard artist, but they can offer draft compositional options. The director takes these sketches, adjusts them, and adds their own logic and vision. As a result, the final storyboard emerges faster, and creative exploration is not constrained by time limits. In practice, it looks like this: we give AI a scene description, and it generates several compositional variants. The director selects one as a foundation, refines it, adds angles important for dramaturgy, and builds the editing rhythm. The result is a living working material that reflects the author's intent but was created with technological support. This does not make the process mechanical—on the contrary, it frees up time for deeper work on the story.
At the production stage, once all styles and characters are approved, the most voluminous part of the work begins. Here there are hundreds of frames, thousands of details, and each requires attention. It is in production that neural networks can deliver maximum time savings if we correctly define what can be trusted to them.
An animation project contains a huge number of backgrounds: streets, rooms, nature, interiors. Not all of them require unique artistic rendering. For crowd scenes where background elements are not in the viewer's focus, neural networks work perfectly. The artist sets the style and parameters, AI generates dozens of background variants, from which suitable ones are selected and manually refined. This frees up artists' timeso they can focus on complex shots—close-ups, expressive facial expressions, and intricate compositions. Of course, in key scenes, backgrounds are drawn exclusively by hand, because every detail matters there. But for secondary episodes, automation is an excellent solution. A good example is projects with many repetitive elements: forests, cityscapes, furnished rooms. AI can generate a basic structure, and the artist adds unique details and brings it to the required quality. This approach maintains a high artistic level without overloading the team.
Technologies can already generate in-between frames between key character poses. This is routine work that in classical animation took a lot of time. Neural networks can calculate smooth motion, especially for simple actions: walking, turning, waving. But the final quality still requires an animator's oversight, because AI does not always correctly convey physics or the nuances of character movement. In our studio, we use AI interpolation for rough passes to quickly assess the overall dynamics of a scene. The animator sketches key poses, the neural network fills the gaps, and the director can view a draft version and make decisions: keep something, rework something, or completely re-animate something manually. This saves time in the early stages and allows for more experimentation with motion.
The final stage of production is polishing. Here, the animation is already complete, but it needs to be improved, cleaned up, and made more expressive. In post-production, neural networks can also be useful, with the caveat that their intervention must not damage what has been done by humans.
AI tools excel at removing artifacts, increasing sharpness, and improving color reproduction. This is especially important for projects that will be broadcast on large screens or in high resolution. Neural networks can automatically process frames, remove minor flaws that the artist might have missed, and equalize the color gamut across the entire series. But the final decision always rests with the human, because AI might process a frame too aggressively and strip it of the desired atmosphere. We also use AI to create final renders for projects containing complex special effects: glow, particles, fog. Neural networks can generate these elements faster than traditional methods. But again—this is not a replacement for compositing, but an acceleration of specific operations. Compositing, the assembly of all layers into the final image, is still done manually because it is a creative process where taste and a sense of proportion are essential.
| What We Leave to People | What We Delegate to Neural Networks |
|---|---|
| Character creation and personality development | Reference and mood board generation |
| Script and dialogue writing | Draft storyboards and compositions |
| Key animation and expressive facial expressions | Background scenes and environments |
| Final editing and compositing | In-between motion frames |
| Developing a unique brand style | Retouching and quality enhancement |
No matter how advanced neural networks are, they do not understand dramaturgy and do not feel the audience. Therefore, in our studio, we have established a strict control system: AI may propose, but only a human approves. Every scene generated or processed by a neural network undergoes a director's review. If the quality or emotional precision is unsatisfactory—the scene is reworked manually. This is especially important for projects where the client needs animation with a long-term effect. For example, animated series for brands that should become part of the company's image do not tolerate compromises. In such projects, we use AI only for auxiliary tasks, while everything related to characters' images, their interaction, and dramaturgy is done exclusively by hand. Artificial intelligence helps us get through routine stages faster but does not interfere with the creative core. Clients often ask: "Can we save on production by using neural networks more?" We answer honestly: yes, but at the cost of quality. If your goal is to "make a video that will be watched once and forgotten," AI might be a solution. If you want to create content that will work for a brand for years—a professional team is indispensable. And we confirm this with projects in our portfolio where animation delivered millions of views and tens of thousands of subscribers.
The final formula that works in modern studios is this: neural networks accelerate what does not require creative solutions and offer options where a human needs an initial push. Everything related to emotions, characters, and uniqueness remains with people. This is not a compromise—it is a conscious distribution of roles where technology works for the team, not the other way around. Ultimately, the process looks like this: we start with a human idea. Based on it, we assemble mood boards and references, using AI for acceleration. Artists develop the style and characters—this is manual work that defines the project's uniqueness. At the animation stage, neural networks help with drafts and backgrounds, but all key scenes are hand-drawn. Post-production again uses AI for technical refinement, but final editing and color correction are done by people. This approach delivers results that we see in our work: animation that is memorable, evokes emotions, and serves the business objectives of clients. In this scheme, neural networks are not a replacement for artists but additional hands that take on routine tasks and allow the team to focus on what matters most. And this, in our opinion, is how the animation industry should develop. If you are considering how best to organize production for your project—whether with AI or using classical methods—we are ready to provide a consultation and show with specific examples how technologies can enhance the result. The main thing is that a living story always remains at the center, with everything else serving as tools for its realization.
Send request
