How to Photograph Multiple Products Together With AI - Scene Guide

A single product shot only has to solve one problem: make one object look good. A multi-product scene has to solve several at once - composition, consistent scale across every item, and making sure the model doesn't quietly merge two products into one or duplicate an item that should only appear once.

Most AI-generated multi-product failures come down to vague prompting. "A flat lay of three skincare products" leaves too much to the model's imagination - it has to guess positions, relative sizes, and spacing, and it usually guesses wrong on at least one of those.

This guide covers composition rules by product count, ready-to-use prompts for flat lays, lifestyle scenes, gift sets, and seasonal collections, and the spatial language that keeps scale consistent across every product in the frame.

Why multi-product scenes are harder

Three failure modes show up repeatedly in multi-product generations: merged objects (two products blending into one shape at their point of contact), duplicated objects (an item that should appear once shows up twice), and inconsistent scale (one product rendering noticeably larger or smaller than it should relative to the others).

All three come from the same root cause - the model wasn't given enough explicit structure to know exactly what belongs where. Vague grouping language like "arranged together" or "displayed as a set" gives the model no anchor points, so it fills in the gaps with whatever composition looks statistically plausible from its training data, which is not the same as what you actually asked for.

Vague grouping language gives the model no anchor points - it fills the gaps with whatever looks statistically plausible, not what you asked for.

Composition rules by product count

Two products: use a simple left-right or front-back pairing with one clear size or height relationship stated between them - this is the most reliable multi-product configuration.

Three products: a triangular arrangement (one product forward, two set slightly back and to each side) reads cleanly and gives the model a stable structure to work from.

Five products: group into a grid or staggered rows rather than a single loose cluster, and describe each row explicitly - this is the hardest count to get right and usually needs 2-3 generation attempts.

Flat lay prompts for collections

Flat lays work well for product collections because the top-down angle removes most depth-related scale confusion - everything sits on the same plane, so the model only has to solve for size and spacing rather than perspective.

Lifestyle scene prompts

Lifestyle scenes place a product family in a real-world context rather than a studio setting - a bathroom shelf, a kitchen counter, a gift table. These need more spatial description than flat lays since depth and perspective are now part of the composition.

Describing spatial relationships

The most reliable spatial language uses three components together: an explicit position for each item ("left," "center-back," "foreground right"), an explicit size relationship ("the same height as," "roughly half the width of"), and an explicit spacing instruction ("two inches apart," "evenly spaced with no overlap").

Skip vague terms like "nicely arranged" or "styled together" entirely - they carry no information the model can act on, and are the single most common cause of chaotic multi-product output.

Gift set and bundle prompts

Gift set photography needs to communicate that the products belong together as a purchasable bundle, which usually means a tighter, more deliberate arrangement than a lifestyle scene - often with packaging or a box visible in frame.

Seasonal collection prompts

Seasonal collection shots pair a consistent product arrangement with seasonal styling elements - the arrangement logic stays the same as any multi-product shot, but props and color palette shift with the season.

Maintaining consistent scale

The single most effective technique for scale consistency is anchoring every other product's size to one reference item stated explicitly in the prompt - "the soap bar is roughly half the height of the candle" rather than describing each product's size in isolation.

Without an anchor, the model treats each product description independently and has no reason to keep proportions consistent between them, which is exactly what produces the "one product looks oddly huge" failure so common in multi-product generations.

If scale keeps drifting after adding an anchor, try including a shared reference object in the frame - a coaster, a ruler-width prop, or a hand - the same technique used for single-product scale problems, just applied once across the whole scene rather than per item.

Nano Banana Pro for multi-product placement

Google's Nano Banana Pro (built on Gemini's image capabilities) is currently the strongest option specifically for multi-object scenes. It holds object count, position, and relative scale more reliably across a generation than most diffusion-based models, which makes it well suited to exactly the kind of structured, multi-item composition covered in this guide - gift sets, flat lay collections, and product family shots. See our complete guide to Nano Banana Pro for model-specific prompting techniques.

Need a full shot list for a product line rather than a single scene? The Product Description to Shot List tool turns a product description into a complete set of ready-to-use prompts.

Build my shot list →