How do I generate AI photos for multiple sizes of the same product without reshooting each one?

Quick Answer

Reuse one approved product image across size-only SKUs only when the item looks identical at every size. If a size changes the product's proportions, fit, label, packaging, construction, or included accessories, provide measurements and size-specific reference views, then generate and review that size separately. AI cannot reliably predict fit or reconstruct size geometry from one photo, so exact geometry may still require a reference capture, 3D model, or reshoot.

How should I group size variants for AI product photography?

The right unit is visual difference, not the number of SKUs in the catalog.

Size variant	Recommended treatment
The product looks the same at every size, and scale is not visible	Reuse one approved image across those size SKUs.
Overall scale changes and the image must communicate that scale	Create a separate size record with its own physical measurements and reference view.
Proportions, fit, seams, hardware, label, packaging, construction, or accessories change	Treat each visibly different size as a separate product photography subject.
Exact geometry, assembly, or engineering dimensions matter	Use verified 3D or CAD assets, or photograph the affected size.

In Nightjar, a Product is a reusable record for one visually distinct sellable item. Size-only options may share one Product when their imagery is identical. When scale or construction becomes visible, create a Product for each affected size instead of asking one Product to represent conflicting appearances.

How do I give AI enough evidence for a size that looks different?

A size-specific AI image needs evidence of the difference. Add several Product Photos, the reference images stored with a Nightjar Product, such as a clean front view, side view, close detail, and an on-model or in-context view when scale matters. Choose the clearest view as the Main photo.

Add Product Dimensions using real-world measurements such as 34 × 26 × 11 cm. Do not enter a size range such as S–XL in that field. Measurements can help ground scale against a surface, prop, hand, or person, but the Product Photos remain the authority for shape and proportions. A measurement alone does not tell an image model how panels, seams, drape, labels, or included parts change between sizes.

You can avoid a full campaign reshoot by capturing a small factual reference set for each visibly different size, then using AI to produce the finished listing and lifestyle images. Apply the same Recipe, Nightjar's saved Create-form setup, to each Product so the Photography Style, background, camera treatment, and output settings remain consistent while the size evidence changes.

Can AI show how each apparel size fits on a model?

An AI on-model image can present a plausible garment appearance, but it is not evidence of real fit. A Fashion Model in Nightjar is a reusable AI person for photography consistency, not a body-measurement, sizing, or fit-prediction system. Reusing one Fashion Model and Pose can keep the campaign visually consistent, but brands still need real garment measurements, fit testing, and size-guide data.

For apparel whose grading changes length, ease, neckline, sleeve, rise, or drape, include a reference from the actual size. If no size-specific evidence exists, label the output for internal review rather than treating it as an accurate fit representation.

What should I review before publishing a size-variant image?

Compare every generated image with the physical sample or approved size reference. Check the silhouette and proportions, seam and hem positions, product scale, label and logo placement, packaging text, hardware, closures, and included accessories. Nightjar's built-in visual review can retry obvious eligible failures at no extra Credit cost, but a person must approve the size-specific details.

Reshoot the affected size when the differences cannot be established from reference photos, or when fit, regulated packaging, readable text, or exact construction must be documented. Use 3D or CAD when controlled geometry and repeatable views matter more than photographic inference.