Midjourney prompts: wording and reusable settings

Write Midjourney prompts as clear descriptions of the picture you want: name the subject, medium, lighting, and composition, then test a small change instead of piling on adjectives. Enter your description in the Imagine bar on Create, as Midjourney’s Getting Started Guide↗ (opens in a new tab) explains. Keep the accepted wording with the image and settings you actually used. fenbs can hold that tested recipe so a different assistant can help you continue later.

A reusable prompt is more than a sentence you liked. It needs a clear purpose, an example output, and a note about what you checked. Start with one repeatable need, such as editorial illustrations for a newsletter, rather than trying to write a single description for every kind of picture.

How do you write and test the first prompt?

Midjourney documents a subscribed website workflow through Create and the Imagine bar. The steps below are a practical writing method for that workflow. They are not a claim that one formula always produces a particular style or a perfect image.

1. Choose the picture’s job. For example, you need an illustration for an article about taking a quiet lunch break.
2. Name a specific subject: a person sitting on a park bench with a lunch bag.
3. Add a medium and lighting direction, such as colored pencil illustration in soft midday light.
4. Describe the composition you need, such as the person on the right with open space on the left.
5. Enter the description in the Imagine bar on Create, generate, and review the output against those requirements. Save the wording and observations before revising it.

Keep the first attempt simple enough that you can explain a failure. If the subject is right but the scene is crowded, change the composition wording first. If you change the subject, medium, lighting, and settings together, the next result will not tell you which adjustment helped.

What belongs in a clear visual description?

Use concrete, visible details. “A blue ceramic cup on a wooden table” gives you something to inspect; “an incredible, beautiful, perfect image” gives you little basis for a decision. Keep only the words that describe a requirement or a deliberate visual choice.

This outline is a writing aid, not a special Midjourney command. Adapt it to the job. A product study may need a clear viewpoint and background; an editorial concept may need a stronger relationship between two objects. Avoid adding details that have no role in the finished image.

Which example prompts are useful starting points?

Use these as experiments and judge the results yourself. Each describes a different use while leaving room to change a subject or material deliberately. None is presented as an already tested result.

A person eating lunch on a park bench, colored pencil editorial illustration, soft midday light, person on the right with open space on the left

A blue ceramic cup on a pale wooden table, studio photograph, soft window light from the left, simple background, cup viewed from table height

Two paper boats crossing a quiet pond, cut-paper illustration, gentle morning light, one boat near the foreground and one farther away, spacious composition

For a fair comparison, keep the subject and composition from one example and change only its medium. Then inspect whether the required space, viewpoint, and subject still work. Save the weaker result too when it reveals a mistake you want to avoid next time.

When you choose an accepted version, write a short reason: “The cup is clearly visible, and the left side has room for the headline.” Our prompt versioning guide explains how to keep that reason with a changed instruction. Save the accepted wording in fenbs so another assistant can start from the tested version.

How do you reuse wording from an existing creation?

Open the image from Create or Organize. Under Creation Actions, the Use buttons can return its prompt text to the Imagine bar, as the Modifying Your Creations guide↗ (opens in a new tab) documents. Read the recovered wording before changing it, and identify what the next attempt is meant to improve.

The Organizing Your Creations guide↗ (opens in a new tab) describes searching your own prompt text and generation attributes. Use a distinctive subject phrase when finding an earlier experiment. If an expected creation is missing, inspect the filters before concluding it is gone.

The documented reuse belongs to creations; it does not establish a separate saved-text-prompt library. For a library across tools, see how to build a prompt library. That guide handles organization, while this page focuses on writing and reviewing the image description itself.

Which settings should stay with the accepted words?

Record the full model or version name, the exact prompt, any reference images, and the settings you actually used. If your experiment used a seed or image-weight value, copy the observed value into that record. Do not invent a missing value or treat it as proof that another version will produce identical pixels.

• Purpose: where the image will be used and which requirements it must meet.
• Wording: the complete submitted description, including any parameters present in your tested prompt.
• Inputs: reference files and their locations, plus any seed or image-weight values actually recorded.
• Result: the accepted creation or saved image, the reason for choosing it, and remaining checks.
• Next test: one planned change, with the previous accepted recipe kept intact.

Reference features deserve their own tests rather than being treated as descriptive words. If a particular reference-driven look is the next question, use the separate Midjourney style reference guide. Do not assume a copied sentence preserves every choice made outside that sentence.

How to keep tested Midjourney prompts with fenbs

Keep your best work so you can carry on from exactly where you left off, on any device and in any other connected AI. Save a small, tested recipe that another person can understand, rather than an unexplained list of favorite adjectives.

1. Create a free account.
2. Save the accepted prompt with a simple name that says what picture it helps you make.
3. Add the model, reference links, settings you used, and a link to the accepted image. Write what passed and what still needs review.
4. Connect a supported assistant from Integrations and ask it to read the saved recipe before suggesting a revision.
5. After the next test, save what changed and which version you chose. Keep the next step with the work.

fenbs does not generate Midjourney images or automatically import your creations. Midjourney’s own prompt reuse may be enough for a quick variation. Use fenbs when the accepted wording and review findings need to reach another assistant; find more guidance in Prompts and skills for AI work.