1. Separate a concept from an exact product
Text-to-image is useful for exploring a scene with a fictional product. To represent a real item, use an image-input model with a reference you have permission to use. Check packaging, logos, dimensions, and colours against the original; generation can change all of them.
2. Write the visual brief
Describe the subject and material, camera angle, light source, background, and intended crop. Keep one primary subject. Specify the intended space for copy rather than asking the image to solve every layout detail at once.
3. Choose settings before comparing
Set the aspect ratio for the destination and choose the resolution or quality level in the model’s editor. Save those settings with the prompt. Review the estimate before each run, because a different quality setting can change the cost.
4. Change one variable at a time
If the composition works but the lighting does not, revise the lighting instruction only. This makes the next result easier to diagnose. Keep the winning prompt and settings so you can use them as a starting point later.
Example brief
A studio product photograph of a fictional matte green ceramic mug on a pale stone pedestal. Three-quarter view, handle visible on the right, soft window light from the left, warm cream background, realistic contact shadow, generous empty space above the mug, no lettering or logos.
An illustrative starting prompt, not a guarantee of an exact output. Adjust it for the model and supported settings.
Before you publish
- Inspect edges, handles, reflections, shadows, and duplicated objects.
- Check the full-resolution output and the final crop.
- For a real product, verify every visible product detail; do not present an invented detail as a product feature.