I direct and build generative film work: storyboards, environments, characters and motion, using ComfyUI pipelines I design myself alongside Kling, Midjourney and ElevenLabs. Behind it is fifteen years of film production, so the work is judged on framing, light and pace, not on whether a model produced it.
Bespoke ComfyUI workflows on open image and video models, taking a shot from 3D block-out to finished frames.
Character and style LoRAs trained across model families, so a face stays itself across a full performance.
Nodes written where the stock ones fall short, including a relight stage that separates lighting from surface detail.
Training and inference on RunPod and similar cloud GPU platforms.
A live performance carried onto a stylised, painterly character in continuous takes.
Frame-by-frame boards and animatics an animation team can read like a traditional board, with redlines turned around in a day.
ToolsComfyUI · Flux · SDXL · Qwen · Wan Video · LTX Video · Kling · Midjourney · ElevenLabs · IC-Light · RunPod · Python
Before the generative work, fifteen years of film production: brand films and TVCs for adidas, Pepsi, Nike, Visa and Gillette, including a Cannes Lions Bronze, and four animated features plus two seasons of an animated series with DNEG and ReDefine.
See my production work→