WeAdapt Studios

In the darkroom. This engine is still being developed and tested; details can change.

WS-001 · FR 01 · Engine Room

Brand Look Engine

Ten mood images in, one written style preset out that every shoot reads.

Get the free presetRun the full engine

A · Free · keyword BRANDBOOK

The preset and the shotlist

The preset this engine reads and the shotlist template it fills. Enough to run one shoot by hand and see if it fits your brand.

Fig. 1 · The board

A Input slot

5-15 mood or reference images (own shoots, saved references) + brand URL

B Engine

  1. 01style-extractor
  2. 02image-brand-knowledge
  3. 03brand-research

route gemini · fallback gemini

C Contact sheet

one style preset file (light, lens, palette, set, mood, never-list) + a brand folder in the image library

Data sheet

Input
5-15 mood or reference images (own shoots, saved references) + brand URL
Output
one style preset file (light, lens, palette, set, mood, never-list) + a brand folder in the image library
Models
gemini-vision
Route
gemini (fallback gemini)
Command
/studio-brain-setup
Cost per run
a few Gemini vision calls, no image generation
Maturity
Working: used in production here

Fig. 2 · Sample output

Placeholder frames · real output replaces them before launch

These are drawn placeholders, not engine output. Real frames from this engine replace them when it is released.

What it does

It looks at the images a brand already likes and writes down what they have in common: the light, the lens, the colors, the set and what never appears. That description becomes a preset file every other engine reads, so the next shoot starts from the brand’s look instead of from a blank prompt.

What you see in the post

The board: the input slot fills with ten mood images as a loose pile on the light table. A loupe moves over them and pulls out exposure data in mono (KEY · SOFT LEFT, LENS · 50MM, BACKDROP · WARM PAPER). The lines stack into a recipe card. Last frame: the preset file next to one new frame shot with it.

How it runs

  1. /studio-brain-setup starts the Brain setup (planned command); or directly /brand-research <url> for palette and voice.
  2. /image-brand-knowledge creates the brand folder in the image library with the curated mood images.
  3. /style-extractor analyzes the mood images with Gemini vision and writes core/studio/styles/<type>_<name>.md.
  4. uv run --project core/studio python core/studio/tools/check_style.py validates the preset.

What you get (the giveaway)

  • free: an example preset with every field explained + a mood-image checklist (../weadapt-studios-products/product/studios/brain-studio/free/)
  • studio: the full Studio Brain (skills, style templates, commands) for Claude Code and Codex

Honest limits

It describes a look; it does not invent one. With fewer than five coherent references the preset gets vague. The analysis runs on Gemini, so it needs GOOGLE_AI_API_KEY. Working: the skills exist and are used in this repo.

Same studio

More from the Studio Brain