Photographing without a Camera.
Most people see 3D and AI as competing tools. They are not — they solve different problems within the same workflow.
The Curriculum
01 — Think
Introduction
The conceptual foundation
Why deterministic and stochastic tools solve different problems — and why that distinction changes everything.
02 — Build
3D & Blender
The deterministic world
A complete rendering environment and the precise compositional foundation for the AI workflow. Understanding one means mastering both.
03 — Synthesise
AI Image Generation
The stochastic world
How diffusion models work, what encoders do, and how to prompt with precision rather than luck.
04 — Connect
RAY-L
Bringing both worlds together
RAY-L bridges Blender and ComfyUI via ControlNet Canny — full compositional control, full AI creative range.
05 — Show
Case Studies
The method applied to real projects
Real projects, documented from first Blender sketch to finished image — with every decision visible.
06 — Complete
Image Design & Post-Production
Colour grading, image looks, consistent series
Colour grading, image looks, and the decisions that turn a good single image into a consistent series.
Case Study
The RAY-L Workflow in Practice
Case Study F1
Highland Cottage
A case study that deliberately works with minimal Blender geometry — and shows just how much the RAY-L workflow can do with it. How precisely composition and image structure can be controlled while giving the AI maximum creative latitude.
Read Case Study →Position
AI image generation has a cost — in energy, in resources, in dependencies. Anyone using this technology professionally makes decisions every day: which models, which infrastructure, how many render passes. This site doesn't try to talk those questions away. It tries to show a workflow that takes them seriously.
→ Full statementWhat's new
Jun 2026Prompts for Flux
Two encoders, two fields, two different languages. Once you understand the difference between CLIP and T5 — keyword lists on one side, full sentences on the other — prompting Flux stops being guesswork.
Prompts for SDXL
CLIP doesn't read sentences — it reads concepts. Order is weighting, weighting is control, and the negative prompt is an active instrument, not a safety net. The SDXL-specific things worth knowing.
Prompt architecture: the foundation
Before SDXL or Flux specifics: the two encoder logics, the 8 categories every image idea can be broken into, and the four reasons a prompt fails. The overview that makes the rest make sense.