Inclusive model representation, by design
How Lume lets you cast the model your customers actually see themselves in
On templates that feature a human model, Lume lets you adjust how that model looks: gender presentation, apparent age, skin tone, hair, eye color, and build. Getting this right in an AI tool takes deliberate design, because an image model left to its own defaults tends to repeat one narrow idea of what a person looks like. This page explains the choices we made and the lines we will not cross.
Why representation matters for your store
- ✓Customers convert better when they can picture themselves using the product
- ✓A single casting for your whole catalog quietly narrows who feels invited
- ✓Regional storefronts can match local audiences without reshooting
- ✓Representation decided by you, not by whatever the AI happens to pick
How it is designed
The Monk Skin Tone scale, not ethnicity labels
Skin tone in Lume is a 10-point physical scale, the Monk Skin Tone (MST) scale developed by Harvard sociologist Dr. Ellis Monk and used across the industry to evaluate fairness in AI systems. You choose a tone, not a race. Every tone is presented the same size, in the same row, and none of them is the default.
No default look
When you open a model template, the model looks the way the template designer styled it. Nothing is pre-selected in the appearance controls, and leaving an option unset keeps the original look. There is no "standard" model that other looks deviate from.
Independent attributes
Skin tone, hair color, hair style, eye color, age, and build are all separate choices. Picking one never changes another, so the tool cannot bundle features into a stereotype. A model with deep skin can have straight silver hair and green eyes if that is your casting.
Fictional adults only
Every model is an AI-generated, fictional person, always depicted as an adult. The youngest apparent age you can choose is "20s". Lume will not recreate a real person, a celebrity, or a lookalike, and requests that try are blocked before generation.
No facial-geometry controls, on purpose
We deliberately do not offer eye-shape or nose-shape options. There is no neutral vocabulary for facial features the way the MST scale is neutral for skin tone; historically those descriptors have been tools of caricature. Combined with precise controls they are also how someone approximates a real person's face, which our policy forbids.
Guarded free text
A free-text field covers legitimate details the enums cannot, like freckles or glasses. It is filtered on the server for attempts to depict minors, real people, or explicit content, the generation model is instructed to ignore anything outside appearance, and every image passes automated safety filters before delivery.
Using it in the editor
- Open a template that features a model. A Model control appears in the editor toolbar.
- Adjust only what you want to change. Anything left unset keeps the template's original casting.
- The first time you apply changes, you accept our content policy: fictional models only, always adults, no stereotyped or explicit depictions.
- Generate. Your product, the scene, and the lighting stay as designed; only the model's appearance follows your choices.
Cast the model that fits your brand
Model templates work on every plan
Get Started