Experiments with Stable Diffusion XL and custom LoRA models — using AI generation as the visual backbone for virtual productions and still editorials.

Custom LoRA training carries personalization: characters, lighting languages, and material palettes encoded into the model. SDXL handles the base generation. The output sits at the intersection of photography, 3D, and synthesis — finished editorial frames that can stand alongside camera or render.

Brief

Personal R&D — SDXL for virtual production & editorial

Role

LoRA training · Workflow development · Art direction · Lookdev

Stack

SDXL · Stable Diffusion · LoRA · ComfyUI · Automatic1111

Output

Visual library for virtual productions & still editorials