Generative AI Explorations
- Year
- 2025 – Ongoing
- Type
- Personal Explorations
- Tools
An ongoing collection of personal experiments with generative AI video.
I began experimenting with generative AI video in early 2025, using Runway's early models to make short films and animations. This collection brings those early tests together with more recent experiments with Higgsfield's latest models. I enjoy discovering new AI tools and trends, learning through small experiments and adapting my workflow along the way.
2026 · Higgsfield Genjutsu
An ongoing Genjutsu test for product commercials. Filmed at home, the left clip sets the composition and movement with simple props. Genjutsu swaps them for a puppy and product in a studio, letting the scene be directed before generating the final visuals.
A character replacement test using my own cat, exploring how Genjutsu transfers movement from the original clip to a different character. It was quite fun to create this series.
2025 · Runway Gen-4
A short thriller made with Runway Gen-4, exploring how a sequence of AI-generated shots can build suspense. The model available at the time generated silent footage, so I added sound effects separately using ElevenLabs and other AI audio tools.
An action trailer featuring a giant cat in a post-apocalyptic city, made to test camera movement and dynamic shots. The image models available at the time had limitations in visual consistency and movement, and the footage was generated without audio.
Early character animation tests with Runway Gen-4. Controlling the characters’ movements with this version of the model was not easy, and it took many attempts to achieve the intended actions.
An educational video experiment presenting a historical scene in a vlog-style format. This was part of the educational video work featured in Client Work.