How to turn a ghost mannequin photo into an on-model photo with AI

Quick Answer

Use a clean ghost mannequin photo as garment evidence, then either combine it with a separate person image in an AI try-on editor or build a reusable on-model setup. The AI generates the body, pose, and apparent drape; a ghost mannequin view cannot verify size-specific fit or reveal hidden construction. Review the output against the real garment, and reshoot when accurate fit, movement, transparency, or unseen details matter.

Which Nightjar path should I use for a ghost-mannequin conversion?

Nightjar offers two current paths. Product Photography in Create is for repeat on-model production with reusable product and shoot direction. Try On is a prefilled two-image instruction in Edit for one defined garment-person pair.

Goal	Steps	What the path provides
Convert one ghost-mannequin photo with one person photo	In Generate > Edit, add both images to the board, choose Try On, then assign the outfit and person placeholders to the correct images.	A direct edit with explicit image roles. The person image must show enough of the body and intended crop for the AI to build the shot.
Reuse the garment or person across a catalog	Create a Product from the garment photos. In Generate > Create, select it, choose With model, then select a Fashion Model, Pose, and Camera Distance. Add a Photography Style for reusable camera, lighting, and mood direction, plus a setting when needed.	A Product groups several photos around one sellable item. A Fashion Model is a reusable AI person, while Pose and Camera Distance control body arrangement and crop.

For repeat production, save the Create direction as a Recipe, Nightjar's reusable setup for the Fashion Model choice, Pose, Camera Distance, photographic look, setting, and output options. The Product remains separate, so the same direction can be applied to another garment. If a person image or custom Fashion Model is based on a real person, confirm that you have the rights required for the source photograph, AI likeness creation, and intended commercial use; this guide to digital-likeness licensing explains the distinction.

What can AI learn from a ghost mannequin photo?

A ghost mannequin photo gives the AI visible evidence of the garment's outer edges, color, surface detail, and three-dimensional product-only silhouette. It does not show how the fabric responds to a particular body, how much ease a size has, or what happens in motion. It also cannot establish a back panel, lining, closure, print continuation, or other detail that is absent or obscured.

Supply a sharp, evenly lit front view with the full garment visible. Add accurate back, side, fabric, label, fastening, and pattern close-ups when those details must carry into the result. In Product Photography, several Product Photos give Nightjar richer evidence than one ghost-mannequin image. Text direction can clarify visible facts, but it cannot prove construction that no source photo shows.

The on-model result is generated styling imagery. Its folds, tension, body contact, and apparent length are synthesized rather than measured. Texture preservation remains an active virtual-try-on problem because garment warping and image synthesis can change fine patterns and surface detail, as this CVPR research on texture-preserving virtual try-on describes.

What should I inspect before publishing the on-model photo?

Compare the generated image with the garment and every source view at full size.

Identity and construction: Check color, material, print scale, logos, labels, seams, pockets, buttons, zips, straps, neckline, cuffs, and hem.
Silhouette and drape: Look for changed length, invented folds, impossible tension, altered volume, or a waist and shoulder line that imply unsupported fit.
Edges and anatomy: Inspect hands, hair, skin-to-fabric boundaries, crossed limbs, shadows, and any area the ghost mannequin originally left hollow.
Occlusion: Confirm that the pose has not hidden or redrawn the features a buyer needs to see.

Nightjar's built-in visual review can retry obvious eligible failures at no extra Credit cost, but it is not fine-grained fit, drape, construction, or likeness approval. The current Nightjar Terms of Service state that generated output may contain errors and make the user responsible for reviewing and validating it.

For a small local defect in an otherwise approved image, use a narrow Edit instruction or manual retouch, then inspect the whole garment again because a generative edit can change areas outside the requested fix. Composite approved product pixels when factual detail matters more than a fully generated result. Reshoot the garment on a fitted person when the buying decision depends on defensible size, fit, compression, transparency, support, movement, or construction that the source photos do not establish.