Freelancers using Google FlowFreelancers using Google FlowThe objective in this work was to create a series of premium assets for Balto, featuring five specific dog breeds (Pointer, Corgi, Poodle, Westie, Beagle) interacting with a 3D branded element. The project required a seamless blend of AI-generated characters and high-end motion design.
THE CHALLENGE: Raw AI video generations often suffer from significant technical flaws: blurred fur textures, inconsistent lighting, and "shimmering" artifacts. For a premium brand like Balto, these "AI artifacts" create a cheap, amateurish feel that dilutes brand authority. Prompt engineering alone is incapable of producing professional and stable assets.
THE STRATEGY: A Hybrid Digital Pipeline. As the Creative Lead, I developed a proprietary hybrid workflow to bridge the gap between raw AI output and cinematic quality:
- Biological Realism Pass: I executed a heavy post-production layer focused on "fur fidelity" and ocular clarity. By refining the noise-to-texture ratio, I removed the plastic "AI look," ensuring the dogs felt organic and physically present.
- Geometric Fidelity & Interaction: The interaction with the 3D coin required precise physics.
- Iterative Refinement: Every frame was audited for "shimmering" and warped geometry (common failures in AI) and corrected through a multi-pass upscaling and denoising process.
THE CONCLUSION: This project serves as a benchmark for the future of AI in high-end advertising. It proves that AI is a tool, not a replacement for Creative Direction. To achieve this level of quality, a professional eye for VFX and post-production is mandatory. Without this specialized oversight, the result is merely a "cool experiment." With it, it becomes a brand asset of elite quality.
Production Note: The coin in this video served as a spatial and temporal anchor, requiring precise synchronization to ensure realistic physics and interaction.
I was responsible for the high-fidelity AI generation, biological texturing, and character performance. The final high-resolution 3D asset rendering and compositing were managed by the client’s internal team post-delivery. This project was an exercise in designing trust — where the visual language, layout rhythm, and micro-copy all had to work together to make an AI-powered medical tool feel safe, credible, and effortless.
I handled the full design process: research, wireframing, visual system, animation direction, and high-fidelity delivery in Figma.
The dark cinematic aesthetic wasn't decorative — it was strategic, positioned to differentiate the brand in a space dominated by sterile, clinical UI patterns. WORLD CUP '26 — Fully Simulated Cinematic Sports Film
Every frame of this film is generated. No broadcast footage, no 3D software, no stock — every shot, every player, every stadium was created from written direction alone.
One pass travels through a century of football: from the golden grass of the modern game back through eras of muddy pitches, leather balls, and packed terraces — each decade rendered with period-correct film grain, kits, stadium architecture, and light.
What's simulated: Everything. The camera movement — ground-level tracking shots, chase cams, dynamic reframes — is directed through prompt language, not keyframed. The crowd, the physics, the motion blur on the grass, the mud spray on the slide tackles: all generated. Even the soundtrack is AI-composed (Suno), built to drive the edit's rhythm.
The pipeline: Generated through a chained-sequence direction system that holds lighting, atmosphere, and spatial logic across the full 1:12 runtime — directing, not generating isolated clips. Upscaled, then color graded, stitched, and finished in DaVinci Resolve.
Why it matters: Most viewers can't tell this isn't real broadcast footage. That's the point — and the proof that AI cinematic production is ready for commercial work that holds up to scrutiny.
Direction, edit, and grade: Andras Roland Partamas An experimental AI-driven short film exploring the fragile space between presence and absence, reality and perception.
Created at Katman Studio (https://linktr.ee/katmanstudio?utm_source=linktree_profile_share<sid=fce648e2-321e-41f6-8fdd-2f5e57a45a53), In Between merges cinematic storytelling with emerging AI workflows, pushing the boundaries of visual narrative and post-production.
The film has been officially selected, nominated, and recognized across multiple international film festivals, including:
– Award Winner, Best Editing — Portugal Indie Film Festival
– Finalist — Queen City Culture Film Festival
– Finalist — Red Movie Awards
– Official Selections — Cannes Film Festival, Korea AI Cinema Festival, New York City Independent Film Festival, Mashup Cinéma AI
This project reflects a hybrid creative process where human intuition and artificial intelligence intersect — not as a replacement, but as an extension of cinematic language.
