Kashi Keefe← All resources
Creative·Multi-Tool·4 min read

The AI Product-Shot Prompt

Turn a phone photo of your product into a polished, ad-ready studio shot.

Your product is sitting on a kitchen counter under fluorescent light, photographed at a slight angle with your thumb almost in the frame. That is the reality of most product photos taken before a budget kicks in. The good news: AI image tools — specifically ChatGPT with DALL-E, Adobe Firefly, or Midjourney — can take that raw phone shot and rebuild it inside a clean, intentional studio environment. You do not need a photographer. You need a strong prompt and two minutes.

This resource walks you through the exact prompt structure, the variables that move the needle, and a set of variations you can adapt for different product categories and platform formats.

The Prompt

This is the core prompt. Use it in ChatGPT (GPT-4o with image input), Adobe Firefly's Generative Fill, or as a style reference in Midjourney. Swap the bracketed values for your actual product details.

Studio product photograph of a [PRODUCT NAME], isolated on a [COLOR] seamless backdrop. 
Soft-box lighting from the upper left, subtle shadow on the right side. 
Shot with a 90mm macro lens, shallow depth of field, f/2.8. 
High-end commercial photography aesthetic — clean, minimal, no props. 
Color grade: [WARM/COOL/NEUTRAL]. 
Output: square format, 1:1 ratio, 4K resolution. 
Do not add text, watermarks, or logos.

If you are uploading your existing photo for transformation (available in ChatGPT and Firefly), add this line to the end of the prompt:

Reference image provided. Retain the exact product shape, color, label, and texture. 
Replace the background and lighting environment entirely. 
Do not alter the product itself.

What Each Variable Controls

  • COLOR backdrop: White reads clinical and clean — good for supplements, tech, skincare. Off-white or warm cream softens it for food, wellness, lifestyle brands. Dark charcoal or black adds luxury weight for spirits, fragrance, or premium leather goods.
  • Lighting direction: Upper-left soft-box is the industry default for product e-commerce. If your product has a label on one side, flip it to upper-right so the copy stays readable. Side lighting reveals texture — useful for textiles or food.
  • Lens and aperture: 90mm macro with f/2.8 creates the blurred background isolation that reads as professional. Wider aperture (f/1.8) adds more blur. Close it down (f/5.6) if you need the full label sharp.
  • Color grade: Warm tones (slight amber, golden hour feel) sell lifestyle and food. Cool grades (blue-white, clean daylight) sell tech, medical, and minimalist brands. Neutral is safe for anything going to Amazon or a catalog.
  • Format: Square (1:1) for Instagram feed and Amazon listings. Vertical (4:5 or 9:16) for Stories and TikTok. Horizontal (16:9) for banner ads and website hero sections. Specify it every time — default outputs vary by tool.

Variations by Product Category

Pull the base prompt above and swap in the ending block that matches your product type.

// SKINCARE / WELLNESS
Backdrop: pale sage green or warm white marble surface. 
Add a single sprig of eucalyptus placed beside the product, not touching the label. 
Mist effect on the product surface suggesting freshness. 
Lighting: diffused natural light, slightly overexposed.
// FOOD / BEVERAGE
Backdrop: dark slate or weathered oak wood surface. 
Three-quarter overhead angle (45 degrees), not flat lay. 
Warm tungsten lighting, slight lens flare from upper right. 
Add condensation droplets to any glass or can surface. 
Styling: one complementary ingredient (e.g., coffee beans, lemon slice) placed at lower-left edge.
// TECH / APPAREL / ACCESSORIES
Backdrop: pure white infinity curve (seamless white floor-to-wall transition). 
Hard directional light from upper-right, strong shadow cast to the left. 
Reflection visible on surface below the product — polished floor effect. 
Color grade: high-contrast, clean cool white. 
Minimalist, no props, no texture elements.

Tips That Actually Matter

  • Upload the clearest photo you have. AI tools reconstruct better from a sharp, well-lit original — even if the background is messy. A blurry or dark source photo degrades the output label and texture detail.
  • Run the same prompt three times and pick the best result. Generative outputs vary. You are not editing a single file — you are sampling from a distribution. Budget for iteration, not perfection on the first pass.
  • Tell the model what NOT to do. Adding lines like 'do not add props,' 'do not alter label text,' and 'no watermarks' reduces the most common failures. Negative instructions carry real weight.
  • Use ChatGPT's image editing mode to fix small issues. If the AI changed your label color or added an extra button, paste the output back in and write a targeted correction prompt: 'The label on this product should be black with white text. Correct only the label — leave all other elements unchanged.'
  • Check the output at full resolution before using it. AI images sometimes produce plausible-looking but incorrect label text, especially on small type. Zoom in and verify the product name and any claims are accurate before publishing.
  • Firefly is the safest choice for commercial use. Adobe Firefly is trained on licensed content, which matters if you are running paid ads or selling the image. Midjourney and DALL-E have their own terms — read them if your use case involves resale or broadcast.

One prompt, multiple formats: Once you have a strong hero shot, re-run the prompt with format changes (1:1, 4:5, 16:9) to generate a full set of ad-ready assets in one session. You now have Instagram, TikTok, and banner covered without a reshoot.

← Back to the library