The workflow in seven steps
Shoot one good input photo
Daylight, plain background, the whole product in frame, no filters. A phone is enough.
Add detail shots
Back view, side view and close-ups of fabric, print, seams or hardware help the AI copy the product faithfully.
Pick the right tool
A general editor for backgrounds, or a fashion tool if the product has to be shown worn.
Choose the look
Studio background, a scene, or an AI model, plus a pose if the tool offers one.
Generate a small batch first
Three to five images at standard quality show whether your settings work before you spend more.
Check every image against the product
Color, print, length, logo, buttons. Discard anything that is not true to the real item.
Download and upload to your store
Put the strongest image first, name files clearly and keep the same style across the catalog.
The rest of this article explains each step, with the mistakes we see most often. The examples use Wizia AI Studio, but the same steps apply to any AI product photo tool.
Step 1: the input photo decides the result
AI product photo tools build on what they see. If the color in your photo is wrong, the color in every generated image will be wrong. If a sleeve is cut off, the tool has to invent it, and invented details are rarely accurate.
Five rules cover most of it:
- Good light. Shoot next to a window or outside in the shade. Avoid yellow indoor bulbs and hard shadows across the product.
- Plain background. A single color wall, a white sheet or a clean floor. Less noise means a cleaner cut out.
- Whole product in frame. Do not crop sleeves, trouser legs, straps or soles.
- No heavy filters. Shifted color and contrast carry over into the result.
- Several angles. Front, back and side, plus close-ups of fabric and details.
For clothing, shoot the garment on a hanger or laid flat, with no person in the frame. That gives the AI the cleanest view of the garment to put on a model. Our guide on how to photograph clothes covers setup and light in more detail.
Step 2: choose a tool for the job
There are two kinds of tools. General editors remove backgrounds, place products in generated scenes and fix light, for any product type. Fashion tools also put garments on AI models, which is what clothing needs, because shoppers want to see the fit.
If you sell cosmetics, home goods or accessories that do not need to be worn, an editor is usually enough. If you sell clothing, choose a tool that makes on-model photos. Our comparison of AI product photography tools lists the main options with prices from their own sites.
Steps 3 and 4: set up the generation
In Wizia AI Studio you start a new generation, give the product a name and pick its category, such as tops, shirts, dresses, shoes, trousers, swimwear or accessories. The category tells the AI how to understand the product, so choose it precisely.
Then upload the photos (JPG, PNG, WebP and HEIC are accepted) and choose:
- A model from the library, or switch the model off to get the product alone on a clean background.
- A scene, filtered to suit the product's category.
- A pose, or leave it on automatic.
Finally choose the number of images and the quality. A 1K or 2K image costs 1 credit and a 4K image costs 2 credits, and the total shows before you start. Other tools have their own forms or prompts, but they ask for the same decisions: what to show, where and at what size.
Step 5: start small, then scale
The most common way to waste credits is to generate a large batch with settings you have not tested. Start with three to five images of one product at 2K. Look at them on a phone screen, the way most of your shoppers will. If the model, scene and light suit your brand, generate the rest of the catalog with the same settings.
Keeping the same settings matters more than it seems. A category page where every product uses a different model, background and light looks messy. Pick one model and one or two scenes for each collection and stay with them. On Wizia's Pro+ and higher plans you can create a custom model and custom scenes, which keeps one face and one look across the whole catalog. Our article on a consistent AI brand model explains why that helps.
Step 6: check every image before it goes live
AI images can look perfect and still be wrong. Before you publish, compare each image with the real product in daylight:
- Color: does it match the product, not your screen's idea of it?
- Print and pattern: same scale, same placement, no invented elements.
- Shape and length: hem, sleeves, neckline and waist in the right place.
- Details: buttons, zips, logos, labels and stitching.
- Hands, feet and edges: no odd fingers, no merged straps, no blurred borders.
If a single image fails, regenerate that image. In Wizia you can regenerate one shot without redoing the whole batch. If many images fail in the same way, the problem is usually the input photo: reshoot it before you regenerate again, because better input helps more than more attempts.
Step 7: upload to your store
Download the images one by one or all together. Then add them to the product in your store admin. On Shopify, open the product and add the files in the Media section, then drag the best image to the first position. On WooCommerce, set the product image and add the rest to the product gallery. CloudCart and other platforms work the same way: the image files go into the product's images.
Two small habits pay off. Name files descriptively, for example linen-dress-sand-front.jpg, which helps search and keeps your media library tidy. And keep one clean main image on a plain background for every product, because marketplaces and category grids look best with it. Worn and styled images go below it.
Turn the steps into a weekly workflow
Once the first products work, the seven steps become a routine. A workflow that suits a store adding new pieces every week:
- Shoot all new arrivals in one session, in the same spot and the same light, so every input photo starts equal.
- Reuse saved choices per collection: the same model, the same one or two scenes, the same quality setting.
- Test one image per product, fix or reshoot what fails, then generate the full set. A full shot pack for one product is around 15 images, so about 15 credits.
- Check, export and upload in one pass. Full resolution exports need Starter or higher; Free plan images carry a watermark. Name files and write alt text as you upload.
- Reuse the same session for social posts and, if you need motion, an AI video made from the same product at 35 credits.
Split the work if you can: one person shoots, another checks the results against the real products. Fresh eyes catch wrong colours and details faster.
What it costs
With a general editor you usually pay a monthly plan. With Wizia you pay in credits: 1 credit for a 2K image and 2 credits for a 4K image. The Free plan includes 20 one-time credits to test on your own products, and its images carry a watermark. Starter costs €20 a month and includes 150 credits, enough for a small store adding a handful of products each week. Pro costs €98 a month and includes 750 credits. Full resolution downloads are on Starter and higher. Every plan is on the pricing page.
As an example, a store adding 20 new products a month with five images each at 2K would use 100 credits a month for images. That is an illustration, not a quote: your number depends on how many images you keep and regenerate.
Mistakes to avoid
- Skipping the input photo. A dark, cropped or filtered photo cannot become a good AI image.
- Changing settings for every product. It breaks the look of your category pages.
- Publishing without checking. One wrong color can mean a return and a bad review.
- Using only AI scenes. Keep a plain main image for grids and marketplaces.
- Forgetting the platform's image rules. Marketplaces such as eMAG or Allegro and Google Shopping have their own requirements for main images; check them before a large batch.
For more tools and comparisons, see the AI tools hub.