Legible:A Visual System for Big Ideas
A brand and AI visual language for a new explainer series
Currently in progress

01The challenge
Legible is a new YouTube series about the systems behind the biggest news: supply chokepoints, private power, the hidden parts inside a phone.
Every episode needs visuals for big, abstract ideas. They have to look crafted, stay consistent from episode to episode and be fast to make.
02The insight
Meaning first. Style last.
The AI kept failing when it had to guess what an image was supposed to say. So every idea gets worked out as a plain white model before any style touches it.
03The system
Four metaphors
Four simple metaphors do most of the work. The metaphor stays the same across episodes. The story inside it changes.




04Built to scale
A rough wireframe comes first, so the AI renders the layout instead of guessing where things go.
Every prompt fills the same six slots: Frame, Subject, Camera, Style, Lighting and Constraints. Anyone on the team can pick it up and write one.
The subject, seed and lighting stay locked while the style changes, so one idea can become 3D collage, glass or abstract spheres.






05Where it’s headed
A script-to-screen pipeline
Script, shot list, stills and motion, rough cut, then a Premiere file. The look lives in one style file. Swap that file, and the same system makes a different show.

06My role
- Role
- Created the brand identity, then built the visual language, prompting system and AI pipeline.
- Through
- The Big Bad Good
- Status
- In progress
- Tools
- Z-Image
- Qwen Image Edit
- Kling
- Seedance
- Premiere
- Next.js
- Work
- Brand identity
- Visual language
- Prompting rules
- Script-to-screen pipeline