WeShop AI: Product Photos from Flat Lay to Model

You have a flat-lay photo of a dress and need pictures for a new listing. A clean scene could make the item easier to see. An on-model view could help someone picture an outfit. You do not need to produce every possible view on the first attempt. Start with the picture that answers the shopper's next question.

This guide explains how to approach that work in WeShop AI, using an illustrative dress brief. The suggested prompts and review notes are examples you can adapt, not results from a shoot we ran. Product controls are based on the linked official guides; check the available options and usage information in your current interface.

1. Give each picture a job

Before uploading anything, write down where the picture will go. A listing needs a clear view of the item. A campaign can use more atmosphere, but the shopper should still recognize what is for sale. If you try to solve both at once, you may end up with a striking background and a dress that is too small to inspect.

For a first set, plan a simple product view, one on-model view and a real detail photo of the part shoppers need to check. Keep those jobs separate. The detail photo might show a fastening or fabric weave; there is no reason to generate that detail if you already have a good photograph of it.

Example brief: “One black dress, visible straps and hem, neutral background for the listing. One relaxed on-model view for styling. Preserve the neckline, waist construction and fabric appearance. No added accessories covering the garment.” Add the dimensions and crop requirements of the intended channel after checking its current guidance.

2. Make the source photo easy to read

Lay out the garment so its edges are visible and the important parts are not folded underneath it. Check the straps, sleeves, neckline and hem before taking the photo. Remove props that cover the product. A bag across the waist may look styled, but it also hides information you will need when checking the result.

Use even lighting and keep a version of the original photo without a strong filter. If the source makes black look grey, a longer prompt will not give you a dependable color reference. Place any close-ups beside the source in your working folder so you can compare buttons, lace or stitching later.

Open the photo at full size. If a printed logo is already unreadable or the garment edge is blurred into the background, take a clearer photo when you can. Upscaling may make the file larger, but it does not prove what the missing lettering originally said. Use photos you own or have permission to process.

3. Choose a product scene or an on-model view

The official AI Product guide describes placing a product in a new scene. For a clothing photo shown on a person, the AI Fashion Model tutorial demonstrates uploading the garment, choosing Virtual Try-On, and selecting a model and location.

Choose the route by what you need to show. If the dress itself is the subject, start with a simple product scene. If you want to show a possible outfit, try an on-model image with a pose that leaves the clothing visible. Keep real garment measurements and actual fitting information available: a generated person does not establish how the dress fits a customer.

Do not add a long prompt just because you can. Some workflows rely mainly on the uploaded image and interface selections. Where a prompt field is available, use it to explain the setting and the details to preserve. Where it is not, choose the closest available settings and inspect the result against the same brief.

4. Start with a background that gives the product room

A busy room, dramatic lighting and several props give you more things to check. For the first attempt, choose a background with enough contrast to see the garment edges. A dark dress against a dark wall may hide the straps even when the image otherwise looks convincing.

If your chosen workflow accepts a scene prompt, try: “Show the dress in the product reference against a simple light studio background, with soft, even light and the whole garment visible. Preserve its neckline, straps, hem and visible fabric details. Add no lettering or props over the dress.” Adjust the wording to the source you actually have.

For an on-model view, use the available model and location controls to establish the setting. Favor a relaxed pose that lets you inspect the item. Hair over the neckline, crossed arms or a bag held at the waist can make a clothing error harder to notice.

If you are unsure about the direction, generate one candidate and review it. Look at any displayed point cost before submitting. Avoid treating an advertised example cost as a fixed price for every model, resolution or account.

5. Check the garment before choosing your favorite

Put the original and generated picture next to each other. First check whether the same item is present. Then look at lighting and composition. A beautiful room is not a reason to accept a changed neckline.

• Count visible buttons, straps and fastenings. A plausible extra button still changes the product.
• Compare the neckline, waist and hem. Look for a straight edge becoming scalloped or a long skirt becoming a mini dress.
• Check prints, lettering and logos closely. Reject invented text rather than hoping it will be too small for anyone to notice.
• Look at fabric around hands, hair and furniture. Watch for edges that merge into another object or disappear.
• Compare color with the original, allowing for the fact that lighting and screens affect appearance. If you cannot judge a colorway reliably, keep a real product photo in the set.

An on-model picture needs a few more checks. Inspect hands, limbs and where the garment meets the body. Make sure the pose does not create a new slit or hide a strap. Do not use the generated image as evidence of stretch, compression or exact fit unless those claims are supported separately by the real product.

For a back view, use an actual back reference. A convincing invented back can still have the wrong zipper or seam. If the source does not show it, leave that view out until you can verify it.

6. Fix one visible problem at a time

Write down what failed. “The left strap is thinner than in the reference” is useful. “Not premium enough” is difficult to review and may send the next attempt in a completely different direction.

Where the workflow supports an edit request, try: “Restore the left strap width to match the product reference. Keep the neckline, pose and background unchanged. Do not add new details.” This is an instruction to attempt a correction, not a guarantee of an exact repair. If an edit path is unavailable, regenerate from the original with a simpler pose or setting.

Use the source of the problem to decide what to change:

• If a detail is missing in the original, improve the reference first. Do not expect repeated generation to establish its true design.
• If the source is clear but the result adds decoration, simplify the scene and restate the clothing details that matter.
• If a hand covers or distorts the waist, change the pose before reworking the whole background.
• If the garment is accurate but disappears into the scene, adjust the background contrast rather than redesigning the item.

Check the status of the previous task before sending it again. A delayed or failed generation is different from a completed result with the wrong clothing. Read the status and available recovery options so you do not confuse multiple attempts or spend points repeating an unresolved task.

Keep the useful results. You may only need to redo one view, not the entire set.

7. Build a consistent set without disguising differences

Once one direction works, write a short style note: background, lighting, framing and the amount of space around the item. Reuse that note for the next product. The AI Product guide describes a Fixed Location option for repeating a setting; check whether it is available in your selected workflow.

A repeated setting is a starting point, not proof that every result matches. Compare the pictures together. A white shirt, black dress and shiny bag may need different handling to remain readable. Check each product's actual color and edges rather than forcing the set to look identical through heavy filters.

For several colorways, use the real reference for each one. Do not assume that recoloring a finished picture preserves every trim, fastening and fabric detail. Keep a note tying each chosen image to the exact item and colorway, especially when similar names make the files easy to mix up.

If the project grows into several pose branches or video ideas, the WeShop Canvas workflow guide explains how to organize that next stage. Get one product image right before expanding the project.

8. Check the saved files in their final layout

Open the files you plan to use, then preview them at the intended crop. Check that the garment is still large enough to inspect, its hem is not cut off and any layout text has room without covering it. Zoom in separately to inspect seams and hands. Looking only at thumbnails misses small errors; looking only at full-size files misses crop problems.

Give the files names that identify the item, view and revision, such as “black-dress_front_listing_v02”. Keep rejected trials outside the folder you send to the person publishing the listing. Include a short note saying which image is for the product view and which is for styling.

Keep real product photos, measurements and descriptions alongside the generated scenes. Check the sales channel's current requirements for image use and disclosure before posting. A useful final set lets the shopper see the item clearly and understand what is supported by the actual product information.

For the recorded controls, use the official fashion model tutorial. To explore the product, visit the WeShop AI introduction. Start with one clear source photo and one picture you can confidently match to the garment.