How to Create AI Product Bundle Images Without Duplicating Items

A product bundle image has a stricter job than a general lifestyle photo. It must show the correct items, quantity, and relationship between their sizes. If a three-piece skincare set becomes four bottles, or a travel-size tube grows taller than the full-size product, the image is unusable even if the lighting looks good.

The safest workflow is to lock the bundle as a composition first, then use AI to improve the setting and lighting around it.

Define the bundle before generating

Write an inventory for the exact offer. Treat it as a production specification, not just a prompt.

For each item, record:

• product name and SKU
• quantity
• package size or relative height
• colour or scent variant
• cap, pump, lid, or accessory
• whether the item is inside or outside its box

A bundle containing one cleanser, one serum, and one moisturiser is different from those three products plus a presentation box. State whether packaging is part of the sale or only a prop.

This inventory gives you a simple approval test. The image fails if any listed item is missing, duplicated, replaced, or redesigned.

Build one clean reference for each product

Prepare a sharp reference image for every SKU. A front view on a plain background is the minimum. Add a three-quarter view when the package shape, depth, cap, or side panel matters.

Keep the reference set unambiguous:

• Use the current packaging only.
• Do not mix old and new labels.
• Keep different sizes and colour variants in separate files.
• Remove hands, decorative props, and strong cast shadows.
• Include the full product outline.

If two bottles look almost identical, note their distinguishing details. For example: “Serum is the 30 ml amber bottle with a black dropper. Cleanser is the 150 ml white tube with a flip cap.”

The workflow for preserving logos and labels in AI product photos explains how to protect packaging details that cannot be approximate.

Assemble the exact bundle before styling it

Do not begin with a text prompt that asks the model to invent a five-item arrangement from scratch. That gives it three jobs at once: recreate each product, count them correctly, and compose the scene.

Remove the backgrounds from the approved product images and place the cutouts on a blank canvas. Arrange the exact quantity yourself. This rough composite needs correct products, scale, overlap, and reading order. It does not need perfect lighting yet.

Google’s official Gemini image-generation guide covers editing workflows that combine an input image with text instructions. OpenAI’s image-generation documentation also distinguishes editing supplied images from creating a completely new image. Editing a controlled layout gives the model less freedom to change the offer.

A useful sequence is:

1. Place the product cutouts on a neutral background.
2. Adjust size and spacing until the bundle is accurate.
3. Add rough contact shadows beneath each item.
4. Save this as the approved bundle master.
5. Edit the surface, background, and light around it.

This takes more setup than one prompt, but avoids chasing random duplicates across repeated generations.

Use positions instead of a vague product list

“Show the skincare set beautifully arranged” leaves the layout open. Describe where each item sits and what may overlap.

Use the uploaded bundle master as the exact product arrangement. Keep all three products unchanged. The tall white cleanser stands at the back left. The amber serum sits at the front centre. The short moisturiser jar is at the right. Preserve the exact item count, package sizes, labels, caps, colours, spacing, and left-to-right order. Place the group on pale limestone with soft window light from the left. Add realistic contact shadows. Do not duplicate, remove, replace, open, or redesign any item.

The prompt protects the offer first, then describes the scene. It also names the common failures.

Avoid piling on visual directions. A bathroom, beach, flowers, water splash, silk, gold tray, and sunset create more objects, overlaps, and reflections to solve. Start with one surface, one background, and one lighting direction.

Keep relative scale consistent

AI can make every item look equally important by making them similar in size. This is a problem when a bundle mixes full-size and travel-size products.

Choose one product as the scale anchor. Record the others as simple visual ratios. For example:

• cleanser height: 100%
• serum height: about 55% of cleanser
• moisturiser jar height: about 30% of cleanser
• presentation box width: about 1.4 times the cleanser height

These are layout guides, not exact measurements. Compare the composite with real dimensions or a verified reference photo.

Use a consistent baseline and camera angle. If one item was photographed from above and another at eye level, their tops, bases, and labels may not align. The guide to generating multiple product angles from one image can help prepare matching views first.

Control overlap, shadows, and reflections

Overlap adds depth, but it can hide information customers need. Keep product names visible and avoid placing one cap across another label. Small gaps are often easier to edit than tight stacks.

Every item should contact the same surface. Check that:

• shadows fall in one direction
• shadow softness matches the light
• each base touches its own shadow
• reflections suit the product material
• no item sinks into the surface
• transparent packaging reflects the correct environment

The troubleshooting steps in fixing floating products in AI product photos are especially useful when several objects share one surface.

Reflective bottles, chrome tools, and glass containers may need local corrections. Edit those areas individually instead of regenerating the group and risking a new count or label error.

Generate variations from the approved master

Once one bundle is correct, use it as the source for each campaign variation. Change the environment around the group, not the group itself.

Useful versions include a clean ecommerce hero image, square paid-social image, vertical story cover, wide website banner, and seasonal lifestyle scene.

Keep the same product order and approximate camera angle until the bundle is stable. For another angle, create a second approved master instead of asking the model to rotate the group and invent hidden surfaces.

Fix common bundle failures without starting over

An item is duplicated. Remove only the extra item, rebuild the background behind it, then check nearby shadows and reflections.

An item disappears. Composite the approved product back into position. Match scale and perspective before adding its shadow.

Two variants merge together. Increase the spacing and use clearer, separate references. Do not place nearly identical packages directly behind each other.

The count is right but the products changed. Return to the bundle master. Protect the product pixels and edit only the environment.

The arrangement looks flat. Adjust height, depth, and overlap in the composite. Do not ask the model to “make it dynamic” without defining which item moves.

Labels are correct but lighting is inconsistent. Relight locally using the workflow for relighting product photos without changing the product.

Approve accuracy before aesthetics

Review the final image against the inventory at full resolution. Count every product. Check each variant, closure, label, relative size, and package state. Then inspect edges, shadows, reflections, and background artefacts.

For marketplace listings or regulated products, keep a verified packshot available and check the rules that apply. AI lifestyle creative should not imply contents, quantities, or features that customers will not receive.

Lock the offer first and style it second. A controlled composite gives AI a reliable arrangement to work around, which is safer than asking it to rediscover the bundle in every generation.

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