// brand dna lock
YOUR AESTHETIC, AS DATA.
Generative tools have creative amnesia — every prompt starts from zero and the brand drifts. We cure it structurally: your visual DNA is captured once as a 7-layer preset, constrains every generation, and is measured on every image — with self-correction when outputs drift.
The seven locked layers
Environment
setting, architecture, natural vs built
Light
direction, quality, time of day, temperature
Colour Palette
dominant + accent colours, contrast
Props
visible elements, brand context
Pose
energy, formality, body language
Composition
framing, negative space, focal cues
Mood
the atmosphere tags that bind it together
Aggregate ≥ 0.85
the drift threshold every image is scored against
Locked once, applied everywhere
Your brand's 7-layer MIRE preset is captured from reference imagery and persisted with your brand profile. Every downstream generation — catalog, campaign, film frames, even music direction — threads the same DNA into its prompt.
Measured on every generation
A vision model scores each generated image against the locked profile: seven dimensions, 0.0–1.0 each, plus an aggregate. Not a vibe check — a rubric, run automatically, recorded per image.
Drift self-corrects
When the aggregate falls below the 0.85 threshold, the validator writes a concrete correction fragment — and the next iteration generates with it, steering back toward the lock inside the existing correction budget.
Auditable by your team
Every score lands in an audit table your studio can read. When you tell your brand board the campaign stayed on-DNA, you can show the numbers per image, per dimension.
See your own scores in the dashboard, or lock a DNA for your brand today.
VoiceCosmos — Design Director
Tap to speak