ChatGPT Image Prompts

Creating an image with ChatGPT is less like filling in a search box and more like briefing a creative partner. You describe the subject, setting, mood, and job the image needs to do. A clear brief makes the first result easier to improve.

A good image prompt does not need special code or impressive adjectives. It needs decisions: who or what appears, what viewers notice first, where the scene is, which style fits, and what mood it should convey. If text matters, specify the exact wording.

This guide shows how to write and refine prompts for portraits, products, social posts, presentations, and illustrations. Replace bracketed choices, keep what matters, and remove the rest.

The anatomy of a useful image prompt

Think of a prompt as a short creative brief. It can be compact, but it should answer the questions that affect the final image. A reliable structure is:

Subject + action or arrangement + setting + visual direction + composition + lighting or color + output requirements.

The subject is the main person, object, animal, or idea. The action or arrangement explains what is happening. The setting establishes context. Visual direction covers the medium or look, such as editorial photography or clean vector illustration. Composition, lighting, color, and output requirements control framing, mood, orientation, and usable space.

For example, “a bicycle” leaves most choices undecided. “A bright yellow commuter bicycle leaning against a brick wall outside a café on a rainy morning, realistic lifestyle photograph, eye-level three-quarter view, soft overcast light, portrait orientation, leave open space in the upper-left corner” gives the model a stronger plan.

Specificity helps when it controls an important choice, but contradictory styles and tiny details can distract. Start with decisions that affect purpose and recognition, then add detail when the first result shows a problem.

A prompt template you can adapt

Use this fill-in template for most projects:

Create a [type of image] showing [main subject] [action, pose, or arrangement] in [setting]. Use a [visual style or medium] with a [mood] feeling. Frame it as a [shot or composition], using [lighting] and a [color direction] palette. Make [priority detail] especially clear. The image is for [purpose or audience]. Use [orientation or aspect ratio], and leave [empty space or placement requirement] if needed. Avoid [specific unwanted elements].

You do not have to fill every bracket. A product image may need camera angle and background, while an illustration may need expression and color instead. Treat the template as a checklist rather than a rule.

Writing prompts for portraits and people

Portrait prompts work best when they describe a person respectfully and focus on visible, relevant characteristics such as clothing, expression, pose, setting, and lighting. Avoid arbitrary traits that do not support the intended image.

Create a warm, natural portrait of [person or character description] wearing [clothing], with [expression and pose], in [location]. Use [photographic or illustrated style], [framing], and [lighting]. The mood should be [mood]. Keep the person’s features natural and believable, with hands and facial details carefully rendered. Use a [color palette] background with enough separation from the subject.

Example:

Create a natural editorial portrait of a middle-aged community gardener wearing a faded denim jacket and green work gloves, smiling slightly while holding a basket of herbs in a shared garden. Use realistic photography, a waist-up frame, gentle late-afternoon light, and a soft background of raised beds. The mood should be welcoming and capable. Keep the expression relaxed, the hands anatomically clear, and the background detailed but not distracting.

For a fictional character, add continuity details:

Create a full-body character sheet for a curious young astronomer named Mira. Keep her short dark curls, round copper glasses, mustard sweater, navy trousers, and red canvas shoes consistent across three poses: standing, seated at a telescope, and waving. Use a clean illustrated style, plain light background, even lighting, and clear separation between poses. Do not add extra characters or text.

For a real person, obtain appropriate permission and be honest about the image. Do not imply that someone attended an event, made a statement, or endorsed a product when they did not. Review identity-related details before sharing.

Writing prompts for products and objects

Product prompts should make the item easy to inspect. State what must remain accurate, how it is positioned, and what background supports it. For a real brand, use only authorized design features and check for altered labels or accidental logos.

Create a clean product photograph of [product], showing [important features] from a [camera angle]. Place it on [surface] against a [background]. Use [lighting style] with soft, controlled shadows and a [color palette]. Keep the product’s shape, proportions, controls, and materials consistent. Leave [location] uncluttered for copy. No extra objects, labels, or invented logos.

Example:

Create a clean product photograph of a handmade blue ceramic soap dispenser with a brushed brass pump, shown from a slightly elevated front angle. Place it on a pale limestone bathroom counter against a softly blurred cream tile wall. Use bright diffused window light, gentle shadows, and a palette of blue, ivory, and brass. Keep the dispenser cylindrical, the pump centered, and the glaze visibly handmade. Leave the right third uncluttered for product copy. No extra bottles, labels, or logos.

For a catalog, consistency matters more than drama. Reuse the same background, framing, light direction, and camera distance, changing only product-specific details.

Prompts for social posts, presentations, and marketing

Communication images need room for words and a clear focal point. Specify where the headline will go. Add final text afterward when accuracy matters, because generated lettering may be misspelled or warped.

Create a [orientation] visual for [audience and use] about [topic]. Show [subject or scene] with [mood and style]. Place the main focal point in [location] and keep [opposite area] simple, low-detail, and suitable for a headline added later. Use [palette] and [lighting]. Do not include readable text, logos, watermarks, or decorative elements that compete with the subject.

Example:

Create a wide, optimistic visual for a community newsletter about repairing household items instead of replacing them. Show a person’s hands repairing a small table lamp on a tidy workbench, with useful tools arranged nearby. Use a friendly documentary-photography look, warm daylight, and a palette of wood, soft blue, and yellow. Keep the left third simple and lightly textured for a headline added later. Do not include readable text, logos, or watermarks.

For a presentation, request one main idea and a controlled number of supporting elements. For a square post, mention the central safe area so important content stays away from the crop edge.

Prompts for illustrations and visual concepts

Illustrations benefit from a defined audience and repeatable visual system. State whether the image is for a children’s book, handout, game concept, or poster. Describe line quality, shape, and detail instead of combining unrelated art labels.

Create an illustration for [audience and purpose] showing [scene]. Use [medium or illustration approach], with [line, shape, or texture direction]. The composition should emphasize [main idea]. Use [palette] and [lighting or atmosphere]. Keep the characters and objects [quality or accessibility requirement]. Avoid [unwanted visual traits].

Example:

Create a friendly instructional illustration showing three steps for planting a seed in a small pot: filling the pot, placing the seed, and watering gently. Use simple flat shapes, clear dark outlines, and a limited palette of terracotta, leaf green, cream, and sky blue. Arrange the three steps from left to right with generous spacing and no background clutter. Keep the tools recognizable and the hands clear. Do not include labels; leave space beneath each step for captions added later.

For a series, establish a “style anchor”: a short set of rules such as rounded shapes, muted colors, thin outlines, and paper texture. Change the scene while keeping those rules unchanged.

Customizing a prompt without losing control

Customize in layers. Write the purpose first, then the subject and composition, followed by style, lighting, color, and constraints. This keeps decorative language from hiding the requirement.

Use concrete nouns and observable descriptions. “A calm, premium scene” is less actionable than “a single stone beside still water, cool gray palette, wide negative space, and soft morning mist.” Use one mood at a time unless contrast is intentional.

When revising, change one or two variables at a time. If the scene is right but too dark, ask for brighter light; if the subject is too small, request a closer crop. Changing everything together hides what helped.

A useful refinement sentence is:

Keep the subject, setting, and overall style from the previous image. Change only [specific change]. Preserve [details that already work], and remove [problem detail].

Another is:

The composition is close, but the image feels [problem]. Make the focal point [clearer adjustment] without changing [protected elements].

These instructions turn vague dissatisfaction into a controlled edit and preserve continuity.

Practical tips for better first results

Lead with purpose. Name the use, such as a product listing, lesson aid, or event banner.

Name the priority. Tell the model what viewers should notice first.

Specify framing. Use terms such as close-up, waist-up, overhead, wide, centered, or three-quarter view.

Request usable space. Reserve an area for a title when needed.

Avoid contradictory styles. Choose one primary medium and compatible qualities.

Use constraints selectively. Mention likely problems such as extra fingers, logos, or watermarks.

Add important text afterward. Generated lettering may be inaccurate.

Save successful base prompts and revise from them.

Reviewing the result responsibly

Treat an AI image as a draft, not proof that every detail is correct. Inspect faces, hands, small objects, patterns, edges, and written words. Look for duplicates, impossible shadows, bent architecture, altered product features, and misleading details.

Review carefully when an image concerns health, safety, history, news, finance, or a real event. A polished image can show an incorrect procedure or falsely imply that a person, place, or product appeared somewhere. Disclose synthetic or substantial AI assistance when context requires it.

Use likenesses, private information, and recognizable creative work carefully. Obtain permission where appropriate. Avoid deceptive impersonation, fabricated evidence, and sexualized depictions of minors. Do not make unverified claims with an image. When accuracy matters, prefer a labeled diagram or reviewed authentic photograph.

A compact workflow you can repeat

Start with a brief, then expand it into subject, setting, composition, style, and constraints. Generate a first version, inspect it against the purpose, name the largest problem, and request one focused revision. Repeat until useful, then check it at its actual size.

This is faster than adding adjectives. Keep the final prompt with the audience, disclosure requirement, and features that must not change.

Conclusion

Effective ChatGPT image prompts are clear briefs, not magic formulas. Define the subject and purpose, choose a composition, and protect important details. Use templates, customize with concrete choices, and refine one variable at a time.

Separate creation from approval. Let the tool explore possibilities, but review important outputs for visual errors, misleading implications, unwanted likenesses, and inaccurate text. A focused prompt and careful final check make the result more useful and responsible.