User's Guide

1. What SkinBalance Does and Who It's For

SkinBalance is a color-science tool for photographers. Its job is to tell you, with precision, what adjustments your editing software needs to bring skin tones into accurate white balance.

It is not a beauty retouching tool. It does not smooth skin, reshape features, or apply filters. What it does is measure the actual color of detected faces in your photo, compare those colors against a reference library of calibrated skin tones in LAB color space, and return a set of numerical adjustments: Temperature, Tint, Exposure, and Saturation.

Who it's built for

• Portrait photographers who need consistent, accurate skin tones across a session or across multiple subjects
• Photographers working in mixed or challenging light where white balance is difficult to nail in-camera
• Anyone who has reached for a ColorChecker passport and wished there were a faster, software-only alternative

If you shoot people for a living, accurate skin tones are not optional. SkinBalance makes the correction process systematic instead of subjective.

2. Uploading a Photo and Reading Your Results

Step 1: Upload your photo

Drag and drop or click to upload a JPEG, PNG, or WebP file. SkinBalance runs face detection automatically on upload. You do not need to crop or mark faces manually.

Step 2: Let the analysis run

The app detects faces, samples the skin tone from each, converts those values to LAB color space, and sends them to the API for analysis. This takes a few seconds. When it's done, you'll see a card for each detected face.

Step 3: Read the per-face card

Each face card shows:

Confidence Score (0–100)

This is how close the sampled skin tone is to its ideal reference in LAB space. Luminosity (brightness) is evaluated separately from color, so the score reflects color accuracy independent of exposure.

Matching Tone

The closest reference tone from SkinBalance's calibrated library. The library spans seven core tones from Deep Skin through Lightest Skin, with 16 interpolated steps between each pair for a total of 103 reference points. The matching tone name tells you which reference the algorithm found as the best match for this face.

Sampled and Target Swatches

Two color swatches side by side: what the app measured (Sampled) and what the ideal should be (Target). These give you a visual confirmation of the direction the correction needs to go.

Suggested Adjustments

The per-face recipe: Temperature, Tint, Exposure, and Saturation values derived from a calibrated lookup table (LUT) built from empirical measurements. Values below the significance threshold are suppressed to avoid noise.

Values below these thresholds are not shown. They are real but too small to matter in practice.

3. Applying the Recipe in Your Editing Software

SkinBalance gives you a direction and a magnitude. The recipe is not locked to any specific application. Any software that lets you adjust Temperature, Tint, Exposure, and Saturation can use it: Lightroom, Capture One, DxO PhotoLab, Luminar, Affinity Photo, or any raw processor with equivalent controls.

Lightroom is used here as the reference example because it is the most common tool in professional portrait workflows, but the logic and direction of adjustments is identical across platforms.

Reading the values

The recipe shows values like:

• Temperature: +350 — warmer, shift toward orange/yellow
• Tint: −8 — shift toward green
• Exposure: +0.30 stops — brighten
• Saturation: +12% — increase color intensity

Positive Temperature moves warm. Negative moves cool. Positive Tint moves magenta. Negative moves green. These directions are consistent across all major editing tools, though the numerical scale varies by software.

Applying in Lightroom (example)

1. Open the image in the Develop module.
2. In the Basic panel, locate the White Balance section.
3. Move the Temperature slider by the suggested amount from its current position.
4. Do the same for Tint.
5. Adjust Exposure in the Basic panel.
6. Apply Saturation in the Basic panel or HSL panel.

Using the Corrected JPG

Once results are showing, click Balance Tones for N Faces in the Global Analysis Controls section at the bottom of the page.

SkinBalance runs a multi-pass correction loop on your uploaded image. Each pass applies the current recipe to a canvas, re-analyzes the result, and refines the adjustment until the scores stop improving or the corrections converge. The best-scoring pass is exported, not necessarily the last one.

Download the corrected JPG and compare it against your original. It is a proof of concept. The recipe in the Correction Summary is what you apply in your editing software to your actual working files.

4. Working with Multiple Faces

When SkinBalance detects more than one face, it analyzes each one independently. A group photo with five people will show five face cards, each with its own confidence score, matching tone, and suggested adjustments.

The Global Suggested Adjustments panel

Below the face cards, the Global Analysis Controls section shows averaged adjustments across all enabled faces. This is the single recipe you would apply to a flat JPEG to serve all faces at once. If Face 1 needs Temperature +400 and Face 2 needs Temperature +200, the global suggestion will be approximately +300.

Enabling and disabling faces

Each face card has a toggle switch. Disabled faces are dimmed and excluded from the global average. Use this to remove poorly detected faces (sunglasses, extreme angles, partial crops), isolate specific subjects, or focus the global correction on the faces that matter most.

The Score Spread Gate

When the confidence scores across enabled faces are too far apart (a spread of 15 points or more), SkinBalance suppresses the global adjustment panel and the correction button.

This is by design. A single flat JPEG carries one global adjustment. If one face scores 85 and another scores 50, averaging their recipes produces a correction that fully satisfies neither. Disable the outlier face, run the correction for the remaining faces, then address the outlier separately.

The gate recomputes in real time as you toggle faces on and off.

5. Lock Target: Pinning a Reference Tone

Lock Target is a precision tool for sessions where you need every subject matched to one specific reference. It eliminates target drift and gives you consistent results across multiple uploads.

How it works

By default, SkinBalance finds the closest reference tone for each face independently on every analysis. Lock Target freezes a single tone as the destination for all faces on all subsequent uploads until you unlock it.

How to use it

1. Analyze a photo. Look at the Target swatch on any face card.
2. Click the Target swatch to lock that tone globally.
3. The upload zone displays a banner showing the locked tone name, HEX value, and RGB values.
4. Upload another photo. Every detected face will now be analyzed and scored against the locked tone.
5. To unlock, click the swatch again or use the unlock button in the upload banner.

When to use it

• Group sessions with mixed skin tones: Lock to one subject, then re-upload photos of other subjects to see how far each deviates from the locked reference.
• Multi-photographer workflows: Lock the reference before you start so everyone is calibrating against the same target.
• Batch consistency: If you have 50 frames from the same session and you know the correct target, locking it means every upload compares against that fixed point.

6. The Saturation Alternative Toggle

Sometimes the algorithm detects a situation that is geometrically ambiguous: the skin tone's hue is already correct, but chroma (color intensity) is off. In LAB space, both a temperature cast and a saturation deficit can produce this same symptom. The skin data alone cannot distinguish them.

When this happens, SkinBalance flags the result as ambiguous and offers two alternative recipes:

• Default (Temperature-led): A temperature adjustment to resolve the chroma error. This is the higher-probability fix because warm or cool casts are more common than genuine saturation deficits.
• Alternative (Saturation-led): A pure saturation boost that addresses the chroma deficit directly. The right call when you know the scene's neutrals are genuinely neutral and not introducing a cast.

The toggle

When at least one enabled face is flagged ambiguous, a text link appears below the Global Suggested Adjustments: "Background neutral? Use Saturation +N instead." Clicking it switches all ambiguous faces from the temperature recipe to the saturation alternative simultaneously. Click again to return.

How to decide which to use

If the background in your frame is genuinely neutral (white wall, gray seamless, black backdrop), the deficit is almost certainly saturation. Use the toggle. If you are not sure, or if there is colored light or a non-neutral environment in the scene, stay with the default temperature recipe.

7. Common Scenarios

Portrait session, single subject

Upload your test frame. Check the confidence score. If it is above 85, apply the recipe and move on. If it is below 70, apply the recipe to the corrected JPG first and visually confirm before committing to your working files. The Exposure adjustment (if any) is worth checking against your histogram. A strong exposure suggestion alongside a low score can indicate blown highlights or crushed shadows in the sampled region.

Group photo with mixed skin tones

Upload the image. Review individual face scores before looking at the global adjustment. If all faces score within 15 points of each other, the global recipe is valid. If one face is a clear outlier, disable it and run the correction for the rest. For a group where one subject is backlit, disable them in the global panel, apply the global recipe to your edit, then return to that subject with local adjustments in your software.

Wedding or event photography (wedding, quinceañera)

You are often dealing with mixed ambient and flash, multiple skin tones, and fast turnaround. The workflow:

1. Pick a hero frame from each scene (ceremony, reception, portraits).
2. Upload it, run the analysis, and note the global recipe.
3. Apply that recipe as a base edit in your editing software.
4. Sync that base edit across all frames from the same scene.
5. Upload edge-case frames (shade to direct sun transitions, mixed ambient) to verify. Adjust if needed.

For batch consistency, use Lock Target on a key subject and re-upload as needed to verify consistency across frames.

Studio vs. natural light

In controlled studio light, the correction recipe tends to be smaller and more predictable. Confidence scores are generally higher out of the gate. In natural light, especially mixed or changing light, you will see more variance. Use the face toggles aggressively in multi-subject frames, and treat the score as a diagnostic rather than a verdict. Apply the recipe, download the corrected JPG, and verify visually.

Uploading an already-edited file

SkinBalance reads what is in the file. If you have already made white balance corrections and the image is still off, upload it as-is. The recipe will reflect what remains to be corrected from the current state, not from the original raw values.

8. FAQ and Troubleshooting