How to Fix Product Photo Colours With AI Without Misrepresenting the Product

A product photo can be sharp, well composed and still be misleading because the colour is wrong. Warm room lighting can make silver jewellery look yellow. Mixed LEDs can push white packaging toward green or magenta. A phone camera can make rose gold look more copper than it does in person.

AI-assisted editing can fix these problems quickly, but colour correction is one of the easiest places to over-edit a real product. The safest goal is not to make the image more attractive. It is to remove the colour cast introduced by the camera and lighting so the photograph moves closer to the physical SKU.

Current Lightroom tools make that workflow easier with automatic corrections, AI-powered subject and background masks, white-balance controls and precise colour selection. The important part is knowing which adjustments should affect the whole photograph and which should stay away from the product.

Fix white balance before changing individual colours

If every neutral part of the image looks too warm, cool, green or magenta, start with white balance rather than trying to correct the product colour directly. Shopify's current 2026 product-photography guide specifically recommends matching white balance to the light source and using a grey card in a test shot when accurate colour matters.

In Lightroom, you can use Auto white balance as a starting point or sample a neutral grey area with the White Balance Selector. Adobe recommends choosing an area that should be neutral rather than clicking a bright specular highlight or pure white patch. Then use Temperature and Tint for small corrections.

For repeatable catalogue photography, a grey card is more valuable than trying to remember what the product looked like later. Photograph it under the same lighting as the product, correct that reference frame, then copy the basic white-balance setting to other images from the same setup.

Use AI masks to separate the product from the background

Sometimes the product colour is already close to reality, but the surrounding image needs work. A white background may look grey, or a warm tabletop may be throwing an unwanted cast across the composition. This is where AI masking is more useful than generative recolouring.

Lightroom's current masking tools can automatically select the subject or background. You can then adjust exposure, temperature, tint and other settings only inside that mask. This lets you brighten a background or neutralise a colour cast without automatically changing the metal, fabric or packaging itself.

The August 2026 Lightroom update also adds new Feather and Edge controls for masks, giving more control over selection boundaries. That is useful around reflective products, where a rough mask can create an obvious colour seam along polished edges.

Be careful with Point Color and selective saturation

Lightroom's Point Color and Color Range tools can isolate a narrow hue and change its hue, saturation or luminance. They are useful when one photographed colour is slightly off, but they can also make a real SKU less accurate very quickly.

Imagine a rose-gold bracelet that looks too orange under warm lights. Reducing orange saturation may appear to solve the problem, but the same adjustment can affect skin, reflections and other warm tones. If you push Hue until the image simply looks nicer, you may end up advertising a finish that does not match the product.

Use selective colour as a precision tool, not a styling preset. Keep the real item beside the screen if possible, make a small adjustment, and stop when the photograph matches the product rather than when the colour looks most luxurious.

A practical colour-accuracy workflow

1. Photograph under one controlled light source. Avoid mixing window light, ceiling bulbs and LEDs with different colour temperatures. Mixed light is harder to correct because different parts of the product can carry different colour casts.

2. Capture a grey-card reference. Place a neutral grey card in the same light for one frame. You do not need it in every final image, but it gives you a reliable neutral reference for that lighting setup.

3. Correct global white balance first. Use the grey card or another trustworthy neutral area, then fine-tune Temperature and Tint. Do this before touching individual product colours.

4. Use AI masks for local problems. If the background or one area needs a different exposure or colour-temperature adjustment, isolate it with a subject, background or colour-range mask instead of applying the edit to the whole image.

5. Make selective colour edits only when necessary. Use Point Color or Color Range conservatively, and compare the change against the physical product or a trusted reference photograph.

6. Export in sRGB for normal ecommerce use. Shopify's current guidance recommends sRGB because it is the standard colour space for consistent web display. Even then, remember that customer screens vary, so perfect colour matching on every device is impossible.

Jewellery needs a stricter colour check

Jewellery makes colour correction difficult because polished metal reflects the environment. A silver bracelet can pick up warm tones from a wooden room or cool tones from a blue wall. Those reflections are not necessarily the metal's actual colour, but removing all of them can make the surface look flat and artificial.

For gold, rose gold, silver and black finishes, verify both colour and material appearance. The highlights, reflections and roughness should still look like the physical piece. If you sell several metal variants, a genuine photograph of each finish remains the best colour reference. AI can help normalise the photography, but it should not decide what your plating looks like.

Where generative AI should stop

If the photograph only has a colour cast, you usually do not need a generative image model to rebuild the product. White balance, masks and selective colour tools preserve far more of the original photographic evidence.

Generative editing makes more sense when you intentionally want a new product variant or creative scene. Lustra Studio's colour-variation workflow is an example of that separate task: changing a metal or material finish while trying to preserve the underlying jewellery. That is different from correcting a photograph of the real finish you already sell.

The takeaway

Colour correction should make a product photo more truthful, not more dramatic. Start with controlled lighting, use a grey card when accuracy matters, correct white balance before individual colours, and use AI masks to keep local adjustments away from parts of the image that are already correct.

For ecommerce, the real SKU remains the final reference. If an edit makes the product prettier but less accurate, it is not an improvement.

Sources: Shopify, “Product Photography: A DIY Guide for Beginners,” updated 30 June 2026; Shopify Help Center, “Taking product photographs,” accessed 25 August 2026; Adobe Lightroom Help, “Create masks,” updated 8 June 2026; Adobe Lightroom Help, white balance and Point Color documentation, reviewed 25 August 2026; Adobe Lightroom August 2026 release notes.