Makeup Shade Accuracy: Lipstick and Foundation Photos with AI

By the ORA Lab team · Updated 7 November 2026 · 8 min read

Key takeaways

  • In makeup, the shade is the product. A lipstick image that is a little too warm or a foundation image a little too light sells a product the buyer will not receive.
  • Shades drift at every step: the light, the camera's white balance, editing, the AI tool and the buyer's own screen. Control the steps you can.
  • Capture reference colour properly: neutral light, a grey card or colour chart in the frame, and a real swatch next to the product.
  • Swatches on several real skin tones do more to sell a shade than any packshot, and they must be checked as carefully as the product itself.
  • Scenes, props and models can be generated. The shade, the texture and the label cannot be changed, so compare every image against the real product.

Most product categories can survive a small colour shift. Makeup cannot. A buyer choosing between two nude lipsticks, or between two foundation shades a step apart, is making the whole decision on colour. If the image is off, the product that arrives is wrong for them, however good the formula is. Shade accuracy is the single most important quality of a makeup product image, and the one most often lost.

This guide is about keeping shades true, from the first reference photo to the final AI-generated scene. For the wider picture of beauty imagery, including skincare and fragrance, start with our beauty and cosmetics guide.

Where shades go wrong

StepWhat happensWhat you can do
LightingWarm or cool light shifts the shade before the camera sees itUse neutral, consistent light for reference photos
CameraAutomatic white balance guesses, and guesses differently each shotSet white balance manually against a grey card
EditingRetouching and filters push colours warmer or more saturatedCorrect to the reference, never by eye alone
AI generationThe tool relights the product for the scene and shifts the shade with itUse a product-anchored tool and compare against the reference
The buyer's screenPhones and laptops show colour differentlyYou cannot control this, so control everything before it
The steps where makeup colour drifts

The last step is the reason the others matter so much. Screens vary, so a small error introduced in the studio becomes a large error on a buyer's phone. Every step you control should get the shade as close to true as possible.

Capturing reference colour

  1. 1.Use neutral, even light: daylight-balanced studio lights or soft window light at a consistent time of day, never mixed sources.
  2. 2.Put a grey card or colour chart in the first frame of each setup, and set white balance from it.
  3. 3.Photograph the product and a real swatch side by side: a lipstick bullet next to a swatch on card, a foundation bottle next to a smear of the formula.
  4. 4.Shoot the bullet, the pan or the liquid itself, not only the packaging, which is often a different colour from what is inside.
  5. 5.Edit on a calibrated screen, and correct to the grey card and the physical product in front of you.
  6. 6.Save these reference images as the master for every shade. Everything else is checked against them.

Swatches across skin tones

A packshot shows the shade in the tube. A swatch shows the shade on skin, which is what the buyer really wants to know. For brands selling in India and the Gulf, that means showing each shade on a range of real skin tones, not one. Our article on why AI product photos look fake covers why generated skin often looks too smooth; for swatches, real texture matters even more.

ProductSwatch to showWatch out for
Lipstick and lip colourLips on several skin tones, plus an arm swatchLip lines smoothed away, finish changed from matte to gloss
Foundation and concealerA stripe on the jawline or inner arm across matching skin tonesShade adjusted to match the model rather than the product
Blush and bronzerApplied on cheeks across light, medium and deep skinIntensity exaggerated or washed out
Eyeshadow palettesEach pan swatched in a row, plus one applied lookPans reordered or shades merged
Nail colourOn nails with two or three skin tonesShimmer or finish changed
Swatch images by product

Finish and texture

Packaging and labels

Packaging has to be as accurate as the shade: the shade name and number, the brand name, the ingredient text and any certification marks. Generated text is one of the most common AI errors, so labels must come from your real packaging photos. The lighting glossary explains how to light glossy and metallic packaging without glare across the label.

The checks every shade image must pass

The wider question of when an image is accurate enough to sell is covered in our accuracy guide. In makeup, the bar is simple: if the buyer would choose a different shade after seeing the real product, the image was wrong. If you want shade-true images and scenes across your full range, and bring your hardest-to-photograph shades.

Frequently asked questions

Why do lipstick colours look different in product photos?
Colour shifts at every step: warm or cool light, the camera's automatic white balance, editing, AI relighting and the buyer's screen. Use neutral light, a grey card, a real swatch in the frame and a calibrated screen to keep the shade true.
How do I photograph foundation shades accurately?
Shoot the bottle beside a real swatch of the formula in neutral, consistent light with a grey card in frame, set white balance from the card, and show each shade on matching real skin tones. Never adjust the shade to suit the model.
Can AI keep makeup shades accurate?
Only if the tool works from real product photos and does not relight the product away from its true colour. Compare every generated image with the physical product and the reference photo before it goes live.
How many skin tones should makeup swatches show?
Enough to represent your real buyers. For brands selling in India and the Gulf, that means a range from light to deep skin, with foundation and concealer shown on skin tones that match each shade.

Keep reading