An AI spot that reached TV and cinema screens

Where it ran

We’re used to stories like this living online — but this one went further. The spot launched across social platforms, aired on major CIS TV channels (STS, Domashny, Channel Five, Che), and ran in cinemas — before feature films, inside ad blocks. Most importantly, the media window was fixed: if we didn’t deliver the final master on time, the spot simply wouldn’t run.

Task and constraint

Task: produce a social advertising spot in a matter of days to hit a fixed placement window.

Constraint: there was no time for a classic CGI pipeline and a large team. A standard route — assets → animation → render → revisions → re‑render — usually takes weeks. Here, we had days.

Approach: speed without losing control

We built the production around generative tools without sacrificing control over:

• message — what we’re actually saying;
• style — how it looks within brand rules;
• tone — the emotion and correctness of social communication.

Core principle: short iterations and constant sync. Not “generate and hope,” but rapid versions, tight alignment, and deliberate refinement until the result could confidently hold up on TV and cinema screens.

Process

• 1. Idea and boundaries — quick alignment on goals, constraints, and tone: what matters, what’s unacceptable, what must be preserved.
• 2. Story and storyboard — narrative structure and visual logic: pacing, meaning beats, key frames, transitions.
• 3. Key character style — brand‑consistent main characters and style baseline: proportions, materials, overall look.
• 4. World‑building — environments and camera angles that support the message, not just “nice‑looking frames”.
• 5. Anchor frames — first and last frame of each scene to reduce randomness and keep directorial intent intact.
• 6. Animation & finish — animation from anchor frames, upscaling for clarity, final direction: sound, rhythm, pauses, edit, and VO sync.

Tool stack

• Nano Banana — characters, environments, key frames (style lock).
• OpenArt — rapid staging and camera angle variants (composition search).
• Topaz — upscaling and image cleanup for broadcast / big screens.
• Kling — scene animation generated from prepared key frames.
• Higgsfield — pipeline hub: generation workflow, version control, stage‑by‑stage sync.

Credits

• Client: Bondarchuk Family Foundation.
• Script, producing, editing, project management: Kotikov Production.
• AI generation (characters, environments, video scenes, upscaling, animation): Addicted.
• Voice‑over: Fyodor Bondarchuk.

Why this case matters

The unusual part isn’t just speed — it’s that a generative AI spot made it into real‑world distribution. Not only digital, but onto platforms that are typically more conservative: television and cinemas.

Special thanks to the Bondarchuk Family Foundation for the trust. Generative AI is still a young field, and choosing to ship a spot like this to real broadcast channels takes openness and courage.

Need to ship fast — at production‑level quality?

If you also have a fixed launch window and no time for a traditional production pipeline, we can help build a solution around your constraints — from concept and visual language to a final master ready for large placements.

Reach out: my@addicted.design