Masonry Logo
FAQ

AI product photography and image generation, answered

Straight answers to the common questions about AI product photography and AI image generation: what it costs, whether it is allowed on Amazon and Shopify, which model to use, and how to keep your product accurate.

AI product photography

What is AI product photography?
AI product photography uses image models to turn a plain photo of your product into studio shots, lifestyle scenes, and ad-ready images, without a physical shoot. You upload a clear photo, describe the scene, and the model relights and restages the product into a new background. The point is to keep the actual product accurate while changing everything around it.
How much does AI product photography cost?
AI product images typically cost from a few cents to about two dollars each depending on the tool and quality settings, versus a few hundred to several thousand dollars for a traditional studio session. For a catalog of 50 products at four scenes each, the AI route is usually single or low double digit dollars, and you can redo any shot the same day. A studio still wins for a flagship hero shoot where craft and control matter most.
Is AI-generated product imagery allowed on Amazon and Shopify?
Shopify places no restriction on AI-generated product images. Amazon allows AI-generated imagery but still requires the main listing image to accurately represent the actual product on a pure white background, with no misleading additions. The safe rule on any marketplace is that the image must truthfully show what the buyer receives, so keep the product itself accurate and use AI mainly for backgrounds, lighting, and lifestyle context. Always check the current marketplace policy, since these rules change.
How do I keep my product looking accurate in AI photos?
Start from a sharp, evenly lit photo of the real product on a plain surface, since the model restages what you give it. Judge every result by zooming in on the label, logo, color, and the point where the product meets the surface, because that is where AI errors show first. Be extra careful with transparent, reflective, or text-heavy products like glass bottles and jewelry, where models struggle most. Keep one honest, accurate photo on the listing alongside the AI scenes.
Do AI product photos look real enough to sell with?
For most opaque, matte products like boxes, mugs, bags, and apparel, yes, the results are routinely store-ready when you start from a good input photo. The weak spots are transparent and highly reflective items and fine printed text, which need more retries and close checking. The honest test is whether the product still looks unmistakably like yours at full zoom, if it does, it will sell.
Should my product photos be white-background or lifestyle?
Use both, for different jobs. A clean white-background shot is what marketplaces require for the main listing image and what buyers expect for a clear look at the product. Lifestyle and scene shots, the product in context, are what win on social, ads, and secondary listing slots because they show the product in use. AI makes it cheap to produce both from one input photo.
Is there a free way to try AI product photography?
Most tools offer a free tier or trial credits, which is enough to test your best-selling products before you pay. A practical free path is to shoot a clean photo with your phone near a window, remove the background with a free tool, then run a few lifestyle scenes on a trial. Test your hardest product first, since free tiers often fall down on glass and small text.
How do I turn a product photo into an ad creative?
Upload the product photo, describe the ad concept including the scene, mood, and any space for a headline, then generate and pick the strongest result. Keep the product centered and accurate, use clean directional lighting, and avoid baking in long text since models still garble dense copy. Generate a few variations so you can A/B test them on the actual placement.

Choosing a model

Which AI image model is best for product photos?
There is no single best model, because each one is strong at different things. GPT Image 2 is reliable for legible text on packaging and precise edits, Nano Banana 2 is fast and strong on photorealism and 4K, FLUX is fast and cheap for high volume, and Seedream is strong on text and 4K at a low price. The practical approach is to run your specific product through a few models and keep whichever preserves it best, which is what a multi-model tool like Masonry is built for.
What is the difference between Nano Banana 2 and GPT Image 2?
GPT Image 2 (from OpenAI) leads on legible in-image text, precise layout, and edits that change one thing without disturbing the rest. Nano Banana 2 (Google Gemini 3.1 Flash Image) leads on photorealism, generation speed, and cheaper 4K output. Choose GPT Image 2 when text and exact composition matter, and Nano Banana 2 when you want fast, photoreal product and lifestyle shots.
Why run one prompt across multiple AI models?
Because no single model is best for every product or style, and the only reliable way to know which one suits your specific input is to compare them on it. Running the same product through several models side by side lets you keep the one that preserved your product and lighting best, instead of subscribing to several tools or betting your whole catalog on one model's house look.

CLI and AI agents

Can I generate images from the terminal or inside Claude Code?
Claude Code cannot generate images on its own, since it is a text coding agent with no built-in image model. You can add that ability with a command-line tool like the Masonry CLI, which generates images and video from the terminal across many models, so an agent can create an asset mid-session without leaving the terminal. Install it with npx @masonryai/cli, run masonry login, then masonry image "your prompt".
Does the Masonry CLI work with AI coding assistants other than Claude Code?
Yes. The CLI is a normal shell command, so any agent or tool that can run shell commands, or a project skill that wraps it, can call it. That covers Cursor, other coding agents, and your own scripts and pipelines. No special integration is required.

Video and Masonry

Can AI make product videos, not just photos?
Yes. Image-to-video models can animate a still product photo into a short clip, such as a slow rotation, a dolly-in, or a lifestyle moment for paid social. Video models like Veo, Kling, and Seedance handle this, and you can generate them from the same product image you used for stills. Quality and length still trail photos, so treat AI video as short ad and social content rather than long-form.
What is Masonry?
Masonry is an AI creative platform for business image and video generation. It runs one prompt across 50+ models on a single canvas so you can compare results side by side and keep the best, includes a Product Studio for turning product photos into ecommerce hero shots and ad creative, and offers a CLI for generating from the terminal. It is built for business creative output, such as product photography and marketing assets, rather than one-off AI art.