Can eye shape be measured objectively?

What part of eye shape can be objective?

A method can place landmarks and calculate relationships visible in an image: relative position, angle, distance, curve or coverage inside a defined frame. The method must also say how the head, camera and eye were positioned. Without that protocol, two images of the same person can produce different apparent geometry.

Research on eyebrow position illustrates both sides. One photogrammetry study reported a reproducible method, yet its authors also noted the lack of a consensus objective measure because the face is curved (Naif-de-Andrade et al., 2012). The paper measured eyebrow position, not consumer eye-shape labels or lash suitability. Its value here is methodological: define the feature before claiming to measure it.

Why aren't almond, round or hooded simply measurements?

They are useful shorthand, but everyday labels combine multiple visible relationships and do not always use one published threshold. “Hooded” may refer to how much lid area is visible in a particular gaze and light. “Almond” may summarize the opening, corners and proportions. A classifier can still output a label, but the label should not be presented as more exact than its definition and validation.

• Observable: a corner appears higher than another in a quality-checked frontal frame.
• Method-dependent: a calculated angle between named landmarks.
• Label: a category assigned from a rule or model trained on examples.
• Meaning: whether the customer likes a lifted, open, elongated or textured effect.

Has research used images to recommend lash types?

Yes. A published multi-model system combined face-shape and eye-attribute outputs to recommend haircut and eyelash-extension types, and the paper reported effective identification and recommendation results (Alzahrani et al., 2021). That is relevant prior work. It is not proof that Central Oasis's camera is validated, that the categories transfer to every client or that appearance alone determines the right service.

The stronger product design separates what the image contributes from what the customer contributes. The image may supply quality-checked observables. The client supplies the desired effect, boundaries and practical preferences. The governed menu supplies bookable options. Desiree resolves what remains uncertain. Each output can then show which field supported it instead of hiding everything behind an “AI chose this” sentence.

Why can't an objective feature decide the desired look?

Because measurement and preference answer different questions. In a Korean mascara study, cosmetic preference patterns were strongly associated with natural eyelash shape (Kwak et al., 2002). An association at the study-group level is not a rule that predicts what one person wants, and mascara effects are not extension-service outcomes.

The useful connection is not “shape chooses style.” It is “starting structure and desired effect should be considered together.” The customer can reject the default, choose a contrasting effect or prioritize maintenance over a visual category. That is why See what we see should explain its visible inputs and booking suggestion while leaving the goal editable and the final decision with Desiree.

The photo may show [observable], but what I want is [effect]. My boundary is [too long/too dark/too full], and I prefer [maintenance or finish]. Please use the image as context, not as a decision about my taste.