Colour Accuracy in Virtual Wig Try-On

Colour is not a single value carried unchanged from a wig to a screen.

The visible colour of a reference depends on the fibres, highlights, shadows, camera exposure, white balance, surrounding colours, screen and viewing environment. A virtual fitting has to place the wig inside the client’s scene without allowing that scene to erase the reference’s commercial identity.

That is why “true colour” needs a method, not a slogan.

The two images begin under different conditions

A catalogue photograph may be produced with:

• controlled lamps;
• a neutral background;
• calibrated exposure;
• colour correction;
• a known camera;
• a prepared wig.

A client photograph may be made:

• inside a car;
• beneath warm indoor light;
• against a dark jacket;
• beside a bright window;
• with automatic phone processing;
• on a screen whose colour is unknown.

If a light caramel wig is placed into a dark source photograph, physically coherent shading can make it appear chocolate brown. If the engine holds the catalogue brightness rigidly, the wig may look disconnected from the room.

The task is to preserve identity while respecting the scene.

Hue, saturation and brightness tell different stories

Colour discussions become clearer when three dimensions are separated.

• Hue describes the colour family, such as warm brown or gold.
• Saturation describes intensity.
• Brightness describes lightness in the captured image.

In one small ELLËS re-audit, a caramel fitting looked visibly darker than its catalogue thumbnail. Sampled areas remained in a similar warm hue and saturation family; the largest perceptual difference was brightness under the darker source scene.

This finding does not prove the mapping is correct. It identifies a more precise problem: a reference can remain numerically warm while no longer looking commercially “caramel” to the client.

Why a global brightness correction is risky

Increasing the brightness of the whole result can make the wig clearer, but it can also alter the face.

Hair and skin share overlapping colours. A simple colour mask may not reliably separate a warm brown wig from areas of the face. A global correction can flatten facial modelling, move highlights or change the apparent skin tone.

ELLËS experiments described in the audit found that skin measurements differed depending on the selected facial zones and method. The observed changes were not always a uniform lightness shift; local highlights and shadows moved differently.

The correct conclusion is not that skin correction is solved. It is that a simple universal offset is not an adequate representation of the problem.

Capture quality should be assessed before generation

The lowest-risk intervention happens before an expensive fitting is created.

A source-image check can assess:

• overall exposure;
• strong warm or cool cast;
• backlighting;
• clipped highlights;
• deep facial shadow;
• whether the face occupies enough of the frame.

The result can be expressed in plain language:

• High confidence: light is even enough for a useful comparison.
• Medium confidence: the fitting is usable, with visible lighting limitations.
• Low confidence: retake the photograph before relying on colour.

This is a confidence statement about capture conditions. It is not a certification that the screen reproduces the physical wig.

Neutral-reference calibration

A more rigorous smartphone workflow could ask the client to include a known neutral reference once, such as a suitable grey card.

The system could then estimate the camera and room’s colour cast before applying the catalogue reference. This approach has a cost: every additional instruction creates friction, and an ordinary white sheet is not a calibrated reference.

A phased approach is more practical:

1. Automatic light-quality assessment.
2. Clear retake guidance when confidence is low.
3. Optional neutral-reference calibration for high-value consultations.
4. Formal validation against physical samples before stronger colour claims.

Professional colour-measurement systems exist in cosmetics and manufacturing. Their presence shows that measured colour is possible; it does not mean a phone photograph without controlled conditions reaches the same standard.

What to show beside a fitting

A useful Session Record can contain:

• selected reference name and ID;
• date and time;
• session identifier;
• source-light confidence;
• a note that screens and ambient light affect appearance;
• the actual generated result.

This is a consultation summary, not a legal determination. It creates traceability and prevents the fitting from being separated from its context.

The client can then ask a better question: “Is this light suitable for judging the caramel tone?” rather than treating every preview as equally reliable.

Testing colour fidelity properly

A robust test set should vary one factor at a time.

Repeatability

Generate the same client-reference pair several times with identical inputs. Compare:

• overall colour family;
• highlight placement;
• root depth;
• silhouette;
• curl pattern.

Lighting

Use the same person and reference under:

• bright neutral daylight;
• warm indoor light;
• low light;
• backlight.

The goal is not identical pixels. It is stable product identity with a visible confidence difference.

Skin tones

Test across diverse skin tones and inspect multiple facial zones. A correction that improves the forehead but worsens the cheek is not a complete correction.

Reference families

Include:

• dark uniform colours;
• highlighted brunette;
• caramel;
• blond;
• red;
• mixed-root references.

Light and multi-tonal references often expose errors that a dark reference hides.

Human review

Ask people who know the physical reference to compare the catalogue piece, the physical wig under controlled light and the fitting. Instrument readings should support perception, not replace it.

How ELLËS should phrase the evidence today

The internal audit reviewed a small set. Later tests showed improved repeatability in one repeated pair and good catalogue correspondence in three of four reviewed references. A darker caramel result remained a perceptual concern under low ambient light.

Those observations support continued deployment and testing. They do not support:

• universal colour certification;
• a claim of pixel-perfect matching;
• performance across every phone and light;
• performance across every skin tone.

An honest statement is:

ELLËS assesses the source light and keeps the selected reference visible within the client’s scene. Low-confidence conditions are identified rather than hidden.

The commercial value of uncertainty

Stating uncertainty can feel weaker than promising exact colour. In a high-consideration purchase, it is often stronger.

A boutique already knows that physical colour changes under light. A system that pretends otherwise is less credible than one that distinguishes a high-confidence fitting from a weak photograph.

Colour accuracy becomes defensible when three things meet:

1. a stable catalogue identity;
2. an assessed capture condition;
3. a traceable result with visible limits.

Read what virtual wig try-on should prove for the full evaluation framework. See textured and Afro hair in digital fitting for why geometry and colour must be tested together across diverse hair and skin.