Quick answer
Use ChatGPT as a prompt editor for Midjourney, not as a guarantee of a particular image. Ask it to organize your idea into subject, setting, composition, style, lighting, constraints and parameters. Generate a few variations, compare results, and change one variable at a time.
ChatGPT can help turn a visual idea into a clearer Midjourney brief,
but it should not be treated as a guarantee of a particular image. The
most reliable workflow is iterative: define the subject and composition,
generate a small set of prompt variations, compare the results, and
change one variable at a time.
A practical prompt structure
Ask ChatGPT to organize each prompt around:
- subject and action;
- setting and time of day;
- composition or camera angle;
- medium, style, or visual reference;
- lighting and color palette;
- important constraints;
- Midjourney parameters supported by the version you are using.
Use this setup prompt:
Act as a visual-brief editor. Turn my idea into five distinct Midjourney prompts.
Keep the subject, audience, and intended use clear. Vary only the visual direction
between options. Put the most important nouns first. Do not promise that a parameter
or style will produce an exact result. Ask for missing information before drafting.
For each option, explain the main trade-off in one sentence.Check the current Midjourney documentation for parameter names and
behavior before using them. Interfaces and model versions change, and a
parameter that worked in one version may not transfer unchanged.
15 reusable templates
Replace the bracketed fields with your own details.
1. Editorial portrait
Editorial portrait of [person or role] [action], in [setting], [camera angle],
[lighting], [palette], [mood], realistic skin and fabric texture, [orientation].2. Product image
Commercial product photograph of on [surface], [background],
[light direction], controlled shadow, [brand palette], room for copy on [side],
clean [orientation] composition.3. Architecture concept
Architectural visualization of [building] in [environment], showing [materials]
and [key features], [viewpoint], [weather or time], natural lighting, realistic
scale cues, [landscape or portrait] composition.4. Character sheet
Character design sheet for [character], front, side, and three-quarter views,
[clothing], [fixed traits], three expressions, consistent proportions, neutral
background, clean concept-art layout, no readable text.5. Book-cover concept
Minimal book-cover concept about [theme], showing [central image], generous
negative space, [palette], refined editorial design, portrait orientation,
blank areas reserved for title and author text, no generated lettering.6. Food photograph
Top-down photograph of [dish] with [ingredients and props], [surface], soft
window light, natural texture, balanced negative space, [palette], recipe-card
composition, [orientation].7. Travel poster
Vintage-inspired travel poster for [place], featuring [landmark or scene],
bold simplified shapes, [limited palette], strong negative space, screen-print
texture, vertical print composition, no readable text.8. Isometric workspace
Isometric 3D illustration of [workspace], including [objects], clean geometry,
soft ambient occlusion, [accent colors], warm indirect light, organized but lived-in,
square composition, no readable text.9. Fantasy landscape
Wide fantasy landscape of [place], [central subject], [distant element],
atmospheric perspective, [time of day], [lighting], restrained [palette],
cinematic but coherent composition.10. Botanical study
Scientific botanical plate of [plant], showing [flower, roots, seed pod, detail views],
fine ink lines, restrained watercolor, warm paper texture, consistent scale, clean
caption areas, no readable lettering.11. Social campaign image
Campaign image for [audience and message], showing [subject and action],
[visual style], clear focal point, [palette], space for a designer to add copy,
brand-safe and inclusive representation, [aspect ratio].12. Vector icon set
Set of ten vector-style icons for [topic], one consistent stroke weight,
simple geometric forms, [palette], clear silhouettes, equal spacing on a neutral
grid, no gradients, no readable text.13. Cinematic still
Cinematic still from an original scene: [character] [action] in [location],
[lens or framing], [lighting], [weather], [color grade], restrained production
design, visual continuity, no logos or readable text.14. Surreal collage
Surreal editorial collage combining [element A] with [element B], [material or
cut-paper treatment], centered focal point, [palette], soft shadows, coherent scale,
calm [mood], no recognizable copyrighted characters.15. Variation generator
Create six Midjourney prompt variations for this brief: [brief]. Keep the subject,
action, and intended use unchanged. Vary only [lighting / composition / medium].
Label each variation and state what changed. Avoid unsupported claims about exact
faces, hands, typography, logos, or product accuracy.How to refine the result
Start with the shortest complete prompt. If the subject is wrong,
clarify the subject and its position before adding more style words. If
the composition is weak, specify the camera angle, foreground,
background, and negative space. If the mood is off, change the light and
palette. If an image contains unwanted text or logos, describe a clean
surface and review the generated image rather than assuming a negative
phrase will solve it.
Keep a record of the prompt, model version, parameters, seed when
available, and intended use. Generate a small comparison set and change
one variable at a time. Review likeness, trademarks, cultural
representation, privacy, and commercial rights before publishing or
selling the result.
For more reusable workflows, explore the PromptsHeaven prompt
database and Complete
Playbook Bundle.
Final takeaway
ChatGPT is most useful as a prompt editor and idea organizer. Give it
a concrete visual brief, ask for controlled variations, check parameters
against the current tool documentation, and let the generated images, not
a list of impressive keywords, show you what to improve.
Frequently asked questions
How do I use ChatGPT to write Midjourney prompts?
Give ChatGPT your visual idea and ask it to organize each prompt around subject, setting, composition, style, lighting and palette, constraints and supported parameters. Ask for several distinct options and the main trade-off of each.
Can ChatGPT guarantee a specific Midjourney image?
No. The post says ChatGPT should not be treated as a guarantee. Results come from an iterative process: generate a small set, compare, and change one variable at a time. Expect to refine the prompts over several rounds.
Do Midjourney parameters work the same in every version?
Not necessarily. The post says interfaces and model versions change, and a parameter that worked in one version may not transfer unchanged. Check the current Midjourney documentation before using parameters. Check the current documentation before relying on a parameter.
How do I fix a Midjourney image that is wrong?
Start with the shortest complete prompt. If the subject is wrong, clarify it and its position. If composition is weak, specify camera angle and negative space. If the mood is off, change the light and palette.
How do I avoid unwanted text or logos in images?
Describe a clean surface and review the image rather than assuming a negative phrase will solve it. The templates also use phrases like no readable text and blank areas reserved for later design.
11 thoughts on “How to Generate Midjourney Prompts with ChatGPT: 15 Templates”