01Product

Dynamic Intelligence
Prompting

The only product on the site. One tool, operated by you: upload a product photograph, make a handful of choices, and generate consistent brand images that stay locked to your product and your aesthetic.

02Overview

DIP is not a prompt box. It is a controlled prompt engine for eyewear brands.

We do not make your images. We give you a tool that is prepared, trained, tested and structured so the AI image model returns an accurate result every time — not a lottery ticket.

What DIP actually does

You make the creative decisions through structured dropdowns — story, model, director style, aesthetic, camera, scene, output. DIP translates those choices into precise, stress-tested instructions the AI image model understands. The product stays locked. The result stays consistent.

Who DIP is for

DIP is built for eyewear brand teams, not prompt engineers. You bring the creative direction. DIP brings the standards, the guard rails and the tested prompt structure that make the AI image model behave.

03The Binding Contract

The Lock Sheet, explained

The Lock Sheet is a structured spec that captures everything about your product and your desired aesthetic. It is not a prompt. It is a machine-readable contract that ensures every generated image stays true to your brand and your product.

What it captures

  • Product geometry mesh (sub-pixel accuracy)
  • Surface material properties and light response
  • Brand identity marks and placement rules
  • Dimensional specifications (scale-locked)
  • Aesthetic system parameters (lighting, palette, mood)
  • Composition language and framing rules

How it travels

  • Embedded in every generation request
  • Validated against the 85-point rubric
  • Version-controlled per campaign
  • Exportable for brand compliance audits
  • Immutable once signed off

What it prevents

  • Product shape drift between frames
  • Brand mark hallucination
  • Colorway inconsistency
  • Scale distortion
  • Style contamination between campaigns
04Quality Assurance

The 85-Point Standards Rubric

Every image passes the same craft test we would apply in a real studio. Before any frame reaches you, it has been evaluated across six categories totaling 85 individual checks. This is not AI slop. This is studio craft, systematized.

Surface & Material

15 pts
  • Acetate grain direction correct
  • Titanium brush pattern accurate
  • Lens material properties match
  • Surface reflections physically plausible
  • Transparency and refraction correct

Scale & Proportion

15 pts
  • Frame width matches spec
  • Temple length accurate
  • Bridge width correct
  • Lens diameter faithful
  • Overall proportions consistent

Geometry & Structure

15 pts
  • Frame angles match original
  • Hinge placement correct
  • Curvature data accurate
  • Nose pad positioning faithful
  • Temple bend angle correct

Brand Identity

15 pts
  • Logo placement exact
  • Engraving details present
  • Brand marks unaltered
  • Colorway matches spec
  • Model name visible where applicable

Aesthetic Consistency

15 pts
  • Lighting system applied correctly
  • Palette matches chosen aesthetic
  • Mood maintained across set
  • Composition language consistent
  • No style drift within campaign

Technical Quality

10 pts
  • Resolution meets delivery spec
  • No artifacts or noise
  • Edge definition sharp
  • Color space correct
85
Points of inspection per image
05How the backend translates your choices

DIP is not a wrapper around a text box. Every dropdown choice runs through a fixed translation layer built specifically for eyewear. That layer has four parts.

1. The Lock Sheet — the contract

Before any prompt is written, your product uploads (front, three-quarter, side, detail) are analyzed and frozen into a binding spec: frame geometry, bridge width, temple shape, lens tint, surface material, logo position, hinge detail, colour values. This spec is injected into every prompt as non-negotiable text. The AI image model is not allowed to reinterpret the frame. No melted hinges. No invented logos. No colour drift.

2. The 85-point Standards rubric — the guard rails

Every generated prompt is checked against 85 eyewear-specific rules before it leaves our system. Examples:

  • Frame must occupy the upper third of any close-up.
  • No facial feature may cross or distort the frame line.
  • Reflections on the lens must match the described light source.
  • Wardrobe must not compete with the frame colour family.
  • Camera distance must respect the lens focal length — no 24mm faces.
  • Hands near the face must not occlude the bridge.

If a prompt fails any of these, the backend rewrites it before the AI image model ever sees it.

3. The translation dictionary — the precision

Each dropdown choice maps to a tested string of technical instructions, not a vibe word.

Example — Camera · Intimate Portrait
  • 85mm equivalent focal length
  • subject distance 1.2 metres
  • eye-line aligned to upper third
  • shallow depth of field, frame sharp throughout
  • single soft key light, 45 degrees camera-left
  • controlled catchlight in lens
  • slow head turn, no rapid movement
  • product remains in focus across full motion arc
Example — Aesthetic · Sun-Bleached Coastal
  • natural daylight, 5600K
  • softened contrast curve
  • light salt-air haze, 8% diffusion
  • matte skin finish, no specular highlights on cheekbones
  • clean neutral reflections on lens
  • no neon, no saturated primaries
  • relaxed pacing, ambient movement
  • wardrobe palette restricted to bone, sand, washed indigo

The user picks one word. The backend sends the full paragraph. Change one dropdown and only that paragraph is rewritten — the Lock Sheet, wardrobe, framing and product stay frozen.

4. Stress-tested prompt assembly — the part nobody else does

Every combination of Story × Director × Aesthetic × Camera × Scene has been run, reviewed, corrected and re-run inside our studio. The prompt structure that reaches the AI image model is the version that survived that testing — not a fresh guess assembled at click-time.

This is why DIP delivers an accurate result instead of a lottery ticket. The brand makes the creative decisions. DIP handles structure, precision, and protection of the product.

06Not a prompt box

Curated aesthetic systems

Every aesthetic system in DIP was authored by us — not scraped, not generated, not crowd-sourced. Each one was engineered against the 85-point rubric to ensure it produces campaign-grade output. The customer picks from a dropdown. Behind the dropdown is a system.

Neo Futurism

Sharp metallic, chrome, shadow

Surrealism

Dreamlike, impossible, ethereal

Biomorphism

Organic, natural, soft

Solarpunk

Golden hour, nature, optimism

Constructivist

Bold angles, industrial

Color Field

Monochrome, tonal, minimal

Neo Classicism

Painter light, directional, timeless

07Trait-based, not likeness-based

Curated model archetypes

Our model archetypes are built from trait specifications — bone structure, posture, expression language — not real-person likenesses. This means they are consistent across a campaign, legally safe, and endlessly reusable.

The Intellectual

Sharp, focused, angular

The Provocateur

Bold, confrontational, raw

The Minimalist

Clean, serene, understated

The Aviator

Confident, open, adventurous

The Visionary

Dreamy, distant, forward-looking

08Why a generalist LLM cannot do this for you

Same starting point. Same ending tool.
Two different roads.

vsYOUbrand owner · directorTrack A — GeneralistChatGPT · Claude · DeepSeek · Geminiglorious. generic. guessing.general web textno eyewear domain model · no product lock"cinematic eyewearportrait, premiumeditorial lighting…"Guessing Gamebridge drift · hinge movedtint shifts · different pairunusable.Track B — SpecialistDIP eyewear knowledge basebridge · curves · acetate / metal / nylonspecialist. specific. binding.Lock Sheetsurface · scale · geometryidentity · wardrobe — lockednon-negotiable specTranslation Layercamera → focal length, distancelighting → angle, temp, falloffaesthetic → exact model stringscomposition → framing rulesAI Image ModelSeedance · Imagen · MidjourneyBinding Contract175 frames · one contract · zero drift

Same starting point. Same ending tool. Two different roads.

The LLM writes a prompt that sounds professional. DIP writes a contract the AI image model cannot break — because we know eyewear, and the LLM does not.

09The Invitation

Ready to stop playing
the lottery?

Upload one product photograph. Make a few choices. See what DIP generates in an afternoon.