AI Product Photo Prompts

Start with the product that must stay recognizable

A product photo brief should distinguish the object being preserved from the scene being replaced. Use Image to Image with a clear product upload. Identify its defining features: the outline of a handle, the cap shape, the material finish, and the actual color. Avoid broad requests such as “make this premium” when the important change is simply a cleaner background.

According to OpenAI’s official image prompting cookbook, reference-based edits benefit from explicit preservation instructions. The examples here apply that principle to original product briefs. They are suggestions for directing a result, not guarantees of commercial accuracy. Compare each generated image with the real item before publishing a listing, especially when packaging wording or subtle construction details matter.

Separate background, light and contact

Specify the background as an environment: a seamless studio sweep, a pale plaster wall, or a kitchen counter. Then describe the light direction and softness. Finally explain how the object meets the supporting surface, including its contact shadow or reflection. This makes the desired physical relationship clearer than a string of adjectives about visual quality.

Preservation language can be concrete: “Keep the bottle’s straight sides, amber glass, black cap and printed label unchanged; replace only the surrounding set.” If another reference supplies a room style, say so explicitly. Otherwise the model may interpret a second product photograph as a different version to merge. Use the smallest reference set that explains the intended edit.

Image models and edit credit examples

The table quotes one uploaded product image and one output. GPT Image 2 offers a fixed 1K low-quality edit. GPT Image 2.5 Flare and Sunburst offer selectable size and quality levels; their matching configurations cost the same. Nano Banana 2.1 and Nano Banana Pro use fixed pricing for each size tier rather than the same selectable quality controls. These are workflow options, not a quality leaderboard.

Dollar estimates allocate purchased credits at Basic monthly: $29.90 for 149 credits, approximately $0.2007 per credit. They are not separate checkout prices. New accounts start with 0 credits. Check pricing before submitting; each quote below covers one result.

GPT Image 2 editing supports up to 9 references, GPT Image 2.5 up to 16, Nano Banana 2.1 up to 10, and Nano Banana Pro up to 4. Extra references can change the quote, so this table should not be applied blindly to a larger set. Choose the desired image ratio in the controls; use the aspect ratio guide for the accepted choices by family.

Six original product edit prompts

Replace a distracting background

“Use the uploaded enamel mug as the product. Keep its rounded handle, cream body, dark rim and existing surface marks unchanged. Replace the cluttered background with a pale warm-gray studio sweep. Keep the mug resting naturally on the surface with a soft contact shadow beneath it; do not add props or alter the camera angle.” This suits a cleaner presentation when the original product view already works.

Create soft studio lighting

“Preserve the uploaded leather wallet’s dimensions, stitching, brown color and folded shape. Light it with a broad soft source from the left and gentle fill from the front. Keep the leather texture visible without glossy plastic highlights. Use a quiet neutral background and a believable shadow under the wallet.” This suits a material-focused view where lighting should reveal construction rather than change the product finish.

Repair a floating appearance with a shadow

“Keep the uploaded soap dispenser exactly recognizable, including its green glass, metal pump and proportions. Place it on a pale stone ledge with a soft shadow directly at the base, extending gently away from the light. Match the shadow direction to the bright window on the left. Keep the dispenser upright and avoid a dark halo.” This brief focuses on contact with the set, rather than dramatic relighting.

Add a controlled reflection

“Use the uploaded perfume bottle without changing its clear glass, cap shape, liquid color or label. Place it on a polished dark surface with a restrained reflection immediately below the base. Keep the reflection softer than the bottle and consistent with its outline. Use broad side lighting with readable glass edges, and keep the background simple.” Inspect both the object and its reflection; a plausible reflection must not introduce a different bottle.

Make an ecommerce white-background image

“Preserve the uploaded running shoe’s sole profile, laces, fabric texture and original blue color. Present the same side view against a clean white studio background with only a faint natural contact shadow. Show the whole shoe with comfortable space around it. Do not add badges, extra lettering, decorative props or new design details.” This suits a catalog draft; confirm the final result against your marketplace’s own requirements before upload.

Place the product in a lifestyle scene

“Keep the uploaded stainless travel cup’s lid, tapered body and brushed finish unchanged. Place it beside a closed notebook on a real wooden breakfast table near a window. Use soft morning light, a gentle contact shadow and a quiet background with no people. The cup remains the main subject, fully visible and at a believable scale.” A restrained set gives the object context without introducing hands, food or other competing details.

Choose a setting around the task

A fixed GPT Image 2 edit suits an economical composition draft. GPT Image 2.5 suits a brief needing explicit size and quality choices. Nano Banana 2.1 offers another fixed-tier starting point, while Nano Banana Pro suits comparisons within its smaller reference limit. Test the actual preservation requirements before selecting a recurring workflow.

Review silhouette first, then color, materials, markings, shadow and crop. If a handle changes, revise the preservation brief or choose a clearer source image before increasing the output size. If the background is wrong, keep the product instructions stable and change the setting description. A focused revision makes the next result easier to assess and helps avoid spending credits on unrelated variations.

Fixing a warm plastic look or a gray cast

People comparing image models often describe two opposite drifts in product edits: a warm, glossy finish that makes ceramic, wood or fabric look like plastic, and a flat gray cast that dulls the real colors. Words such as "realistic" or "high quality" rarely fix either one. Specific material and light language does.

For a plastic look, name the actual surface ("matte ceramic glaze with fine speckles", "brushed aluminium with soft directional highlights", "uncoated cotton with visible weave") and ask for neutral daylight white balance instead of warm studio light. For a gray cast, ask for a clean light background, true-to-life product color, one small bright highlight and a soft contact shadow under the base.

Change one instruction per attempt and compare each result with your original photograph.

Frequently asked questions

Sources

OpenAI: image generation prompting cookbook

Related articles

GPT Image 2 vs 2.5 · Image aspect ratios · Product video model choices