Blog

Animation Stopped Being a Question of Scale

October 12, 2026 · #arteki #ai #animation

Dmitry Saksonov in the audience

Contents

In late 2024, a few dozen people at Arteki Studio took an animated feature from an idea to its first screening in about three months. It later reached a wide theatrical release. For conventional computer animation, that schedule and crew size are unusual. How did we get a feature into cinemas on that timetable?

The scale of a feature

Feature animation absorbs an enormous number of working days. DreamWorks Animation practitioners describe the conventional computer-animated feature as a project that takes several years and hundreds of specialized artists across departments. 3D artists build character models and environments. Animators work on the movement of characters and objects within a shot. Each correction means more work for the specialist responsible for that part of the film.

That production model puts a feature within reach of a small number of studios that can fund hundreds of specialists for years. It creates a high barrier across the industry. At Arteki, we tested a different production process on a feature.

How we set up Arteki

We built Arteki Studio for AI-assisted animation from day one. All the film’s visuals were generated. Artists selected and refined the images for Belovezhskaya Pushcha. People wrote the screenplay and recorded the voices. With that setup, a small crew could take on a feature.

What reached the theaters

Belovezhskaya Pushcha runs about 55 minutes, above the Academy’s length threshold for an animated feature. Its closed premiere took place on December 26, 2024. On January 30, 2025, it opened in more than 80 theaters. During the opening weekend, 10,654 people saw it. By March 4, 2025, during its fifth week in theaters, the studio reported 42,822 viewers across the country and first place among animated films in the capital’s theatrical market.

For me, the test was simple: any moviegoer could buy a ticket and judge the film. That made it a finished product.

Dmitry Saksonov giving an interview to journalists

The decisions behind a generated frame

Getting a suitable frame sometimes took around 20 generations. The artists spotted problems and tried again. Even a good frame had to fit the shots around it.

AI took on part of the labor of creating the visuals, allowing the team to do much more in the same amount of time. The screenplay determined what each scene needed to show. Artists set the requirements for each frame and selected the versions that went into the film. Authorship and responsibility for the result stayed with the people making it.

Where the first film fell short

I see the weak spots in the finished film. The plot is uneven in places. Character details change between shots, and some scenes are too static. We found a way to produce the volume of visuals a feature needs. The next step was to work on visual consistency and movement. The release figures show audience interest; the film’s quality has to be judged on screen.

Since then, we have developed our own production pipeline. Our newer work has more movement and far fewer visible artifacts. We now have full control over the generated result.

What matters to me is that we started making animated films with AI early. The first film took the team through a complete AI-assisted production cycle and showed us what needed further work. We used those lessons to develop our process and continue to refine it with each project.

House Spirits, the studio’s second animated feature, is in the final stage of production using that pipeline. An animation project for JGGL is also in production; both the client and the studio are part of ATOM.

The wall behind the film

The profession itself has a productivity limit. In a working day, an animator can complete a certain amount of work on movement within a scene. A 3D artist has a similar limit: the time it takes to build models and environments. Across a feature, that workload adds up to years of work for a large team.

At Arteki, AI changed the work itself. Artists described the shot they wanted and chose from the generated versions. A few dozen people handled work that conventional production divides among separate departments for modeling, environments, animation, and lighting. The boundaries between those specialties blurred, allowing a small team to bring a feature to the screen. Creative decisions stayed with the artists. I see a similar wall in every profession we work with. I’ll explore that further in the next article.