Contribution
- Created the visual identity from scratch, systematically testing prompt structures across Midjourney v5–v7, Stable Diffusion, and Flux to establish a consistent character language and palette.
- Built a repeatable content pipeline linking stills, motion edits, and music — with structured quality checks at each stage to maintain style coherence across 1,000+ published works.
- Diagnosed output failures across model architectures (artifact detection, style drift, anatomical inconsistency) and documented effective prompt patterns for reuse.