Generate a New Scene Without AI Altering the Product's Shape

Quick Answer

Give the AI the product's real geometry: supply sharp, uncropped photos from several useful angles, add real-world dimensions, and avoid requests that require it to invent an unseen side. Use a product-aware creation workflow for a new shot, edit an existing image when the change is local, then compare every result with the real product before publishing it.

What source photos help preserve a product's shape?

The strongest source set shows the product clearly enough that the AI has less geometry to infer. Start with a sharp, evenly lit photo in which the entire silhouette is visible, the product is not hidden by hands or props, and the lens has not exaggerated its proportions. Add side, back, top, or detail views when those surfaces matter. A request for a new camera angle that none of the source photos shows asks the model to invent that view.

In Nightjar, a Product groups several Product Photos around one reusable product identity. Set the clearest full-product view as the Main photo, then add views that reveal depth, openings, handles, seams, or other shape-defining details. Keep different colorways or visibly different designs in separate Products so the evidence does not conflict.

Add physical Product Dimensions such as 34 × 26 × 11 cm when scale and proportions matter. Nightjar passes those measurements to the generation context alongside the Product's factual description, but photos remain the primary visual evidence. Dimensions help guide scale. For size-specific errors, see why AI product images get scale wrong.

Should I use Product Photography or edit an existing image?

Choose the workflow according to how much of the image must change. In Nightjar, Product Photography is the creation workflow for new product-focused shots, while Edit Images is the workflow for changing or combining existing images on a board.

Task	Safer starting point	Why
Build a new listing or lifestyle shot	Product Photography with a saved Product	Multiple Product Photos and factual details can guide the new scene.
Place the product into a specific existing scene	Edit Images with the product and scene added separately	Direct references such as `@image1` and `@image2` make each image's role unambiguous.
Change one part of an otherwise good photo	Edit Images with a narrow instruction	A smaller requested change gives the system less of the image to reinterpret.
Keep the product pixels unchanged	Manual cutout and compositing	Keeps the photographed pixels exactly as shot.

Keep directions about the shot rather than redescribing the product from memory. Specify the setting, placement, and camera treatment, and avoid instructions such as "make the bottle slimmer" or "show the hidden back" unless a deliberate redesign is the goal.

How should I review and correct product-shape drift?

Compare the generated image side by side with the source photos at the intended publishing size. Check the outer silhouette, width-to-height ratio, curves, symmetry, openings, handles, seams, hardware, logo position, and the spacing between parts. For shape-sensitive products, an overlay with reduced opacity makes small contour changes easier to spot.

Nightjar's built-in visual review can reject and retry obvious failures, such as a missing or substituted product, a misspelled brand or product name, a substituted brand mark, or a catastrophic defect, without another user Credit.

If the shape keeps drifting, replace weak Product Photos, choose a clearer Main photo, supply the missing angle, simplify the scene, reduce occlusion, or request a less extreme camera position.