Art-Direct
Product Photos

7 Step Workflow

ChatGPT can turn a rough packshot into an editorial-quality product image, but only if you art-direct it like a real shoot. Moodboard first, product lock second, concepts before final prompts. Here's the workflow.

1
Step

Build a Focused Moodboard

Collect 6–10 images that reflect the mood, lighting, colour palette, composition and photography style you want. Choose references that feel visually connected, not a random collection of unrelated images.

  • Mood
  • Lighting
  • Colour palette
  • Composition
  • Photography style
2
Step

Upload a Clean Product Packshot

Use a high-resolution, front-facing image with even lighting, visible packaging edges, readable labels, minimal reflections and a plain background. Avoid screenshots, cropped packaging and blurry images.

  • Even lighting
  • Visible packaging edges
  • Readable labels
  • Minimal reflections
  • A plain background
3
Step

Define the Visual Direction

Upload the moodboard to ChatGPT and ask it to study the references. This gives ChatGPT a clearer creative direction before image generation begins.

Study these references. Identify the shared lighting, composition, colour palette and mood.
4
Step

Lock the Product Details

Upload the packshot and specify what cannot change. The product should stay consistent while only the environment around it changes.

Use this packshot as the exact product reference. Preserve its shape, proportions, colours, logo and label text.
5
Step

Generate Concepts Before Images

Ask ChatGPT for 5–10 distinct creative directions before generating a single image. Compare the ideas and select one direction first.

  • Setting
  • Composition
  • Lighting
  • Camera angle
  • Props or supporting elements
6
Step

Art-Direct the Final Prompt

Build the prompt using this formula:

Product + Setting + Lighting + Composition + Camera + Product Lock
Example
Create an editorial product image of the uploaded serum on polished chrome, surrounded by sculptural glass. Use hard side light, deep shadows and a close three-quarter camera angle. Keep the bottle, colours and label unchanged. Make the result feel photographed, not rendered.
7
Step

Generate, Refine and Retouch

Generate one or two images first, then review product shape, packaging text, scale and proportions, shadows and reflections, and overall realism. Re-prompt specific problems instead of changing the entire image.

Keep the composition. Restore the exact bottle shape and label. Make the shadows more natural.

Finish with colour correction, edge cleanup, packaging text and other fine detail work in post-production, just as with a conventional photoshoot.

Remember

ChatGPT can create the visual direction and a strong first image. It does not replace final art direction, quality control or post-production, especially when packaging contains small or detailed text.