How to Relight a Product Photo With AI Without Changing the Product

A good product photo can have the right composition and still feel wrong because of its lighting. The label may be readable, the product may be sharp, and the background may suit the campaign, yet the highlights, shadows, and reflections make the object look pasted in.

AI can help relight an existing product photo, but there is a catch. A broad generative edit may redraw the packaging while it changes the light. The safest workflow treats the product as locked source material and changes illumination in controlled stages.

Decide what needs to change

Before editing, separate the lighting problem from the product itself. Write down the intended result in five parts:

• Direction: Where is the key light coming from?
• Quality: Is it soft and broad or hard and defined?
• Colour: Is the light neutral, warm, or cool?
• Contrast: How deep should the shadows be?
• Environment: What should the product appear to reflect?

This brief is more useful than asking for "better lighting." A precise target gives the model fewer decisions to make.

For example:

Large diffused softbox 45 degrees camera left, gentle fill from camera right, neutral daylight balance, soft contact shadow below the bottle, restrained highlight on the shoulder, clean pale-grey studio background.

If the entire product needs a new viewing angle, the task is no longer just relighting. Use a viewpoint workflow instead, and account for the limits of generating multiple product angles from one image.

Start with the cleanest source

Use the highest-resolution original available, ideally a sharp photo with an unobstructed silhouette. Compression artefacts, clipped highlights, motion blur, and existing colour casts all reduce the information the edit can preserve.

Keep an untouched copy beside the working image. It is your reference for:

• package proportions
• label position and wording
• cap, pump, clasp, or closure geometry
• true product colour
• surface finish
• small legal or ingredient text

If the source already contains a reliable product cutout, preserve its alpha edge. A clean edge makes it easier to adjust the background and surrounding light without altering the object's shape.

Protect the product before generating

The most reliable approach is to mask the edit. Protect the silhouette, logo, label text, and distinctive hardware. Then select only the background, cast shadow, or a narrow area where a highlight needs adjustment.

Adobe's Generative Fill documentation describes this selection-based approach and places generated results on a separate layer. Whatever editor you use, the principle is the same: constrain the changed area and keep the original pixels recoverable.

Don't ask one generation to replace the background, rebuild the shadow, relight the product, and sharpen the label. Large edit regions give the model permission to reinterpret too much.

A safer sequence is:

1. Correct the background exposure and colour.
2. Build or repair the contact shadow.
3. Adjust broad brightness and contrast on the product non-generatively.
4. Generate only small highlight or reflection changes where necessary.
5. Restore the original label and critical details at the end.

For branded packaging, the label should usually come from the source photo, not from generated text. The guide to preserving logos and labels in AI product photos covers that repair workflow in more detail.

Use a preservation-first prompt

A relighting prompt should describe both the desired light and the details that must remain fixed.

Try this structure:

Relight the existing product photo with a large diffused key light 45 degrees camera left and subtle fill from the front. Add a soft contact shadow directly beneath the product and a narrow natural highlight along the right edge. Preserve the exact product silhouette, dimensions, cap, label layout, wording, logo, colour, and material. Change only the lighting, shadow, and background illumination. No new objects, no packaging redesign, no text changes.

Terms such as "cinematic" or "luxury lighting" can be useful as secondary cues, but they are not a lighting plan. Direction, softness, contrast, and colour temperature do more practical work.

Generate a few restrained variations rather than one dramatic transformation. The best result is often the version that changes the least.

Relight according to the material

Different materials need different highlight behaviour. A single generic prompt rarely works across every product.

Matte paper or cardboard should have broad, subdued highlights. Bright specular streaks make it look laminated or plastic. Preserve fine surface texture and keep the shadow edge relatively soft.

Glossy plastic and glass need clean reflections that follow the object's curvature. A vertical bottle usually looks more convincing with a long controlled highlight than with scattered bright spots. Check that every reflection responds to the same light direction.

Metal reflects its environment as much as the light source. Describe large light and dark bands, not just brightness. Perfectly even illumination can flatten the form.

Transparent containers are especially difficult because the background is visible through the product. Edit the background first, then refine transmitted light and edge highlights locally. If the liquid colour or fill line changes, restore it from the source.

For each material, ask whether the new lighting still communicates what the product is made from. Visual polish is not a success if the edit turns frosted glass into glossy plastic.

Make the shadow agree with the light

A product can be accurately preserved and still look artificial because its contact shadow points the wrong way or starts too far from the base.

The shadow should:

• begin where the object meets the surface
• fall away from the key light
• become softer as it travels
• reflect the shape and height of the object
• use the scene's colour rather than pure black

If the item appears to float, repair the contact area separately. This guide to fixing floating AI product photos explains how grounding, occlusion, and shadow softness work together.

When compositing into a lifestyle scene, inspect nearby objects. Their shadow direction, highlight size, and colour cast form the lighting rules your product must follow. Matching those cues is more important than making the product brighter.

Check the edit at three scales

Review the result before exporting.

At thumbnail size, check whether the product reads clearly and belongs in the scene. At normal viewing size, inspect lighting direction, shadow contact, and colour balance. At 100 percent, compare the label, edges, closures, texture, and small print with the original.

A practical approval checklist is:

• Is the silhouette unchanged?
• Are every logo and word accurate?
• Does the material still look correct?
• Do highlights and shadows point to the same light source?
• Is the product colour faithful to the source?
• Are there invented seams, reflections, or accessories?
• Does the image remain credible beside other photos from the same product line?

Use non-generative curves, exposure, white balance, and local dodge or burn for final corrections. These tools adjust pixels predictably and are often better than another full AI pass.

Know when relighting is the wrong fix

AI relighting is useful for campaign variations, social creative, mood changes, and matching an approved product cutout to a new environment. It is less suitable when the image must prove exact colour, transparency, texture, or regulated packaging information.

For colour-critical catalogue work, complex transparent objects, jewellery with precise reflections, or products with dense small print, a controlled reshoot or traditional composite may be faster and safer. AI should shorten the route to an accurate image, not lower the accuracy standard.

In Rasgo AI, the same preservation-first approach applies: begin with a strong product reference, define the lighting precisely, keep edits local, and compare each output with the source before moving on. The winning relight is not the most dramatic one. It is the version that fits the scene while leaving the product unmistakably unchanged.

Explore AI product photography for the workflow, then create product photos in Rasgo.