MakeImgA useful image prompt is a short creative brief. It tells the model what belongs in the frame, how it should look, and what you need the image for. More adjectives do not automatically make a better result. A clear subject and a few decisions you can check afterward are more useful.
Use this template when you start with text to image:
Create [subject and action] in [setting], viewed from [camera angle or framing]. Use [light and palette] with [medium or visual style]. Leave [space or detail needed for the final use].
Replace the brackets with details that matter to your project. You can leave a field out when you do not care about it.
Start with the subject and the intended use
“A coffee cup” leaves the setting, composition, and purpose open. “A ceramic coffee cup on a small café table, square product image” gives the model a subject and a layout to attempt. If you need room for a headline, say where that room should be. If the image is a product concept, describe the object accurately before adding atmosphere.
Try this product brief:
A matte blue ceramic mug on a pale stone table, three-quarter view. Soft window light from the left, warm neutral background, natural shadows. Square product image with empty space above the mug for a headline.
The empty space is part of the composition. Add the headline later in your design tool if exact wording matters; generated lettering can vary.
Make visual choices you can inspect
Describe light, framing, surface, and palette with ordinary words. “Soft morning light” and “eye-level view” are easier to judge than a stack of mood words. Keep the first prompt focused so you can tell which instruction caused a useful or unwanted change.
For a portrait concept:
An adult person reading beside a large window, waist-up portrait at eye level. Diffused daylight, muted green and cream palette, documentary photography feel. Keep the face unobstructed and leave space on the right for layout text.
For an illustration:
A small bookstore on a rainy evening, seen from across the street. Warm light inside, cool blue reflections outside, hand-painted editorial illustration. Landscape composition with the storefront as the focal point.
These are starting briefs, not promises of an exact output. Review the generated image before publishing it.
Use a reference when the source matters
If you already have a product, sketch, or approved visual, use image to image. Upload the source and state both the change and the details to retain:
Keep the mug's shape and blue glaze. Replace the busy kitchen setting with a pale stone tabletop and a quiet cream background. Use soft daylight and preserve a natural contact shadow.
A reference can guide the result, but it does not guarantee exact product geometry, logo, or text. Compare those details with the original. For a longer reference-based workflow, read How to Use Image to Image AI.
Revise one decision at a time
After the first result, name the largest mismatch. If the subject is right but the crop is wrong, change the framing instruction. If the background distracts, simplify the setting. If a product's details drift, make the “keep” instruction more specific or use a clearer source image.
Save the prompt that produced a useful direction, then make one intentional revision. The model selector and aspect ratio options shown in the MakeImg composer depend on the active workflow; check them, along with the displayed credit cost, before each generation.
A quick prompt checklist
- Is the main subject unmistakable?
- Have you specified the framing and destination format that matter?
- Are light, palette, and medium described without conflicting directions?
- If using a reference, have you said what should stay and what should change?
- Can you identify the single change to make after reviewing the first result?
Open the text-to-image workspace and try one brief. If you have an existing picture to adapt, start with image to image.