How to Make an AI Model Hold Your Product Realistically

Create believable AI product photos with a model holding your exact item using matched references, planned grip geometry, controlled compositing and targeted hand repair.

Create product photos
RRasgo AI6 minutes

A model holding a product should look simple. In practice, the image has to solve four things at once: the person, the hand, the package and the physical contact between them. If any one is wrong, the product floats, fingers merge into the label or the packaging quietly changes shape.

The reliable approach is to separate the job into stages. Build a believable pose first, preserve the real product as a reference, then repair only the contact area.

1. Choose a grip that suits the product

Start with the way a real person would hold the item. The grip determines which fingers are visible, what part of the package is covered and whether the pose can carry the product's apparent weight.

Useful pairings include:

  • Small bottle or tube: a relaxed side grip, with the thumb on one side and two or three fingers on the other
  • Wide jar: a supporting palm underneath, or two hands if it looks heavy
  • Can or cup: fingers wrapped around the side, with part of the label still visible
  • Flat pouch: a light pinch near one corner rather than a fist around the middle
  • Box: fingertips along the side or a palm supporting the base

Avoid a delicate pinch on an object that looks heavy. If the logo must face the camera, say so in the brief rather than expecting the model to infer which panel matters.

2. Match the product reference to the intended angle

A straight-on packshot is useful for preserving a front label, but it cannot fully describe a product shown at a steep three-quarter angle. The model may invent the unseen side, cap or dispenser.

Choose a source close to the final camera position. For a three-quarter view, provide a three-quarter product photo. If your tool accepts several references, add a front view for label accuracy and a side view for geometry. Every reference must show the same current packaging.

The source should be sharp, evenly lit and large enough to reveal the edges. If you only have one image, keep the final viewpoint close to it rather than asking the system to reconstruct hidden surfaces.

This is the same principle used when creating lifestyle product photos from one image: the reference defines the product, while the prompt defines the new scene.

3. Plan the hand before placing the final product

There are two useful methods.

With reference-guided generation, upload the product, describe the person and specify the grip. This is fast when the packaging is simple and the output only needs moderate commercial accuracy.

For more control, use a staged composite. Generate the model with a neutral placeholder object of the correct size, then replace it with a clean cutout of the real product. This protects fine label text and exact packaging geometry.

A placeholder cylinder, box or pouch can establish hand position, finger spacing, wrist angle, scale, perspective and contact shadow. It should roughly match the real item's dimensions. Replacing a slim bottle with a wide jar will leave the fingers in the wrong position.

4. Write the prompt around physical contact

Describe the grip, product orientation, camera and lighting. State what must remain unchanged.

Create a vertical lifestyle product photo of a person holding the supplied serum bottle at chest height. Use a relaxed side grip with the thumb on the left edge and three fingers visible on the right. Keep the front label facing the camera and unobstructed. Preserve the bottle's exact silhouette, amber glass, black dropper, cream label, logo placement and proportions. Soft window light comes from camera left. Add a subtle contact shadow where the hand touches the bottle.

Replace those details with the features that identify your item. Keep the first attempt simple. One person, one hand, one product and a clean setting are easier to control than a crowded scene.

Generic quality terms do less work than a precise instruction about where the thumb touches the package.

5. Match scale and perspective

Wrong scale quickly makes a product image look fake. Give the system a physical measurement when you know it, such as “the bottle is 14 cm tall.” You can also describe its size relative to the hand.

Check the product's edges against the scene. A box held upright should follow the same camera perspective as the hand and torso. A cylindrical bottle may show a slightly oval top when viewed from above. If the product reference and model image were shot from different heights, later lighting fixes will not solve the mismatch.

When compositing, place the product first and adjust scale, rotation and perspective before rebuilding the fingers around it. Keep an untouched product layer so you can revise the interaction without cutting out the item again.

6. Repair the contact area with a small mask

Once the product is positioned, edit only the hand-product boundary. Mask the fingers that should pass in front of the product, plus a narrow margin. Do not include the whole hand, face or package unless they need to change.

Small selections reduce collateral edits. Adobe's official Generative Fill guidance uses a selection to define the changed area and places the result on a new layer, making comparison and rollback easier.

Diagnose a failed interaction before regenerating:

  • Extra or fused fingers: regenerate only the fingers, keeping the wrist and product fixed
  • Fingers pass through the package: redraw the occlusion boundary and reduce the mask
  • No sense of pressure: add slight fingertip flattening and a soft local shadow
  • Hand covers the logo: change the grip or rotate the product
  • Product floats inside the hand: tighten finger spacing and add contact shadows

For a deeper repair workflow, see the guide to fixing AI-generated hands.

7. Integrate the product without repainting it

A correct cutout can still look pasted on. Match brightness, colour temperature and contrast with adjustment layers or targeted controls. Add subtle reflected colour from nearby skin or clothing, but protect the label from broad generative edits.

Contact shadows should be short and soft, sitting beneath fingertips and along surfaces pressed against the palm. Glossy packaging may pick up a muted skin reflection. Matte cardboard should not gain sharp glass-like highlights.

If the product lighting points in the opposite direction from the scene, it may be safer to change the scene lighting. Heavy relighting can distort printed details and material cues.

Treat the logo and label as production assets, not flexible texture. The product logo and label preservation guide explains when to mask, composite or restore artwork from the source.

8. Check commercial accuracy

Review the result beside the original product image before colour grading or upscaling. Check:

  1. Product silhouette and proportions
  2. Cap, pump, handle or closure
  3. Label wording, logo and variant name
  4. Product colour and material
  5. Hand anatomy and finger count
  6. Finger overlap and contact shadows
  7. Scale relative to the hand and body
  8. Claims, quantity marks and regulatory text

An attractive image can still be unsuitable for a product page if the package is inaccurate. For paid social or lifestyle content, minor background invention may be acceptable, but the item should represent what the customer receives. If exact small print matters, use the original product layer or label artwork rather than trusting regeneration.

Build the interaction in layers

The shortest dependable workflow is to choose a believable grip, match the reference angle, establish the pose with a placeholder if needed, place the verified product, then repair only the fingers and contact shadows.

Rasgo can help create model-led product scenes and campaign variations from references, but the approval standard stays the same in any tool. The person must look natural, and the product must remain recognisably, verifiably yours.

Explore AI product photography for the workflow, then create product photos in Rasgo.

Create your next Rasgo visual

Turn the ideas from this guide into generated images, creator content, product shots, or video-ready concepts.

Create product photos