← All work

The Stag King

AI film · character-consistent video generation · Foxelli Group

A 43-shot AI micro-drama in vertical 9:16, produced as a test submission. The hard part wasn't generating footage — it was making the same two characters survive 43 shots without their faces, clothes or scars quietly changing.

The episode

The work

Brief
Produce a complete short-form AI drama for vertical viewing — 43 shots, coherent story, consistent cast, deliverable as a finished episode.
Characters
Lyra — a hunter. Thin scar on the right jaw, worn leather jacket, hunting bow.
Vane — appears 39, ashy silver hair, iron ring, transforms into stag form.
Environment
Scandinavian old-growth forest, late autumn turning to early winter. A cold grey-white palette held deliberately across every shot so the cuts read as one location and one day.
The problem
Generative models drift. Ask for the same character 43 times and the scar migrates, the jacket changes weave, the hair shifts tone. Prompt-level description alone doesn't hold a face together across a sequence.
The approach
Two-step generation for consistency-critical details. Lock a base portrait first, then apply an isolated, targeted edit pass for the detail that matters — rather than regenerating the whole frame and hoping the scar lands in the same place. The identity stays fixed; only the intended change moves.
Also solved
Vane's stag-form transformation across shots, and holding the cold palette consistent through scenes generated days apart.
In progress
Image-to-video generation that preserves source composition and character positioning, so a locked frame animates without re-rolling the staging.
Stack
  • Nano Banana Pro
  • Kling 3.0
  • ElevenLabs
Outcome
Delivered as a complete 43-shot episode with a consistent cast — the consistency method now carries across to other character-driven AI film work.

Stills

This is one of a few projects shown here — a small sample of what I've built. Happy to walk through the rest, or go deeper on the consistency method.

Contact

Currently exploring AI filmmaking and automation roles. Happy to walk through any of these systems in detail — architecture, trade-offs, and what broke along the way.

Résumé (PDF) ← All work