Free Midjourney Prompts

Midjourney prompts by category

Choose a collection based on the image brief, not only a broad style word. A portrait needs a clear subject, crop and light direction; a poster needs hierarchy and negative space; a product image needs the object and its materials to stay legible. These categories link to existing PicGens collections, each backed by its own public examples.

• Portraits
• Posters and graphic design
• Logos and brand identity
• Illustration and 3D
• Wallpapers and scenery
• Anime prompts
• Product photography and ads

How to write a good Midjourney prompt

Build the prompt in a readable order: subject, action or pose, setting, composition, light, visual treatment and only then parameters. The subject should say what the viewer sees; the action should clarify what is happening; the setting and composition place it in the frame. Add style and lighting cues that reinforce the same direction instead of stacking adjectives that pull the image in different directions.

• Subject: name the person, object or scene and the detail that should attract attention first.
• Action and setting: describe a visible moment and the environment that explains it.
• Composition: say close-up, full-body, overhead, centered, profile or wide establishing shot when framing matters.
• Lighting and color: use a light source and direction, then a restrained palette or contrast relationship.
• Finish: choose one medium or visual tradition, such as editorial photography, ink illustration or a paper-cut diorama.

Before and after: make a prompt more specific

These short pairs are illustrative rewrites, not copied gallery prompts. They show how adding one concrete instruction at a time makes a brief easier to interpret. Keep the subject and intent intact while improving framing, light and hierarchy.

Midjourney parameters cheat sheet

Parameters belong at the end of the prompt. Start with the scene description, change one control at a time, and record the version and settings when comparing runs. Current availability can differ by model version, so use the official compatibility notes before copying an older prompt unchanged.

• Official aspect-ratio guide
• Official stylize guide
• Official chaos / variety guide
• Official version compatibility chart

Use parameters to test one change at a time

A parameter is most useful when it answers a specific question about the image. If the framing is wrong, change --ar and leave the scene alone. If the image is too interpretive, reduce stylization rather than rewriting every noun. If the first set of options is too similar, try a small change to chaos. This makes each run easier to compare and keeps a prompt from becoming a list of controls that no longer has a clear visual brief.

• Frame first: choose a portrait, square or landscape ratio that gives the subject room to read.
• Control interpretation second: adjust stylize only after the subject, pose and setting are explicit.
• Explore alternatives third: change chaos when you want the initial options to diverge more.
• Keep style separate from subject: use --sref for the visual treatment and the written prompt for what should appear.
• Record version and settings with a result so a later comparison uses the same baseline.

How to reuse a style-reference code

The examples below come from distinct public Midjourney prompt rows in this catalogue. They are ordered by the likes captured in the current snapshot, so the list is a way to discover real prompts that include a style-reference code. It is not a ranking of code quality, a test of reproducibility, or a claim that the code alone caused a result. Follow each card to inspect the complete prompt and source.

• Keep the code with the subject text when you want to study the original combination.
• To explore a code on another subject, change the subject phrase and keep the rest of the prompt simple.
• For a style-reference image instead of a numeric code, follow Midjourney’s current Style Reference instructions.

• Official Style Reference guide

Midjourney vs Nano Banana vs GPT Image

The tools use different prompt workflows, so compare them against the task you need to finish. Midjourney prompts commonly combine descriptive scene text with parameters and Style References. Nano Banana examples often describe a direct edit to an attached image. GPT Image 2 examples can be useful when the brief includes a precise layout or short text. These are workflow distinctions, not a benchmark: results change with model versions, account settings and inputs.

• Midjourney vs Nano Banana comparison
• Nano Banana prompt gallery
• GPT Image 2 prompt gallery
• How to write AI image prompts

A simple Midjourney prompt review checklist

Before you run a prompt, read it once from the subject outward. The first phrase should tell the model what the image is about. Later phrases should refine the scene instead of introducing competing subjects. Check that the camera position matches the intended crop, that the light source can plausibly illuminate the scene, and that the palette supports rather than obscures the focal point. If text needs to appear in the image, keep the requested wording short and inspect spelling in the output. If the image needs to match an existing reference, use the right current reference workflow and describe the new subject clearly.

• Remove repeated adjectives that do not add a visible detail.
• Replace abstract praise such as “beautiful” with a concrete choice about pose, material, color or light.
• Move the aspect ratio and other parameters to the end of the prompt.
• After generation, compare the result against the brief and change the single clearest miss first.

Work through a portrait prompt in three passes

Begin with a clear photographic brief: a waist-up portrait of a violin maker in a small workshop, turned toward a window, holding an unfinished instrument. Decide which element should attract attention first and whether the workshop needs to be recognizable. For a first pass, keep the clothing, background and light description simple. Review the pose and the relationship between the hands and instrument before adjusting the visual treatment. If the crop loses the instrument, ask for more room below the shoulders; if tools dominate the scene, reduce the background detail. A second pass can change the light to a soft side source while retaining the pose and framing. A third pass can add a restrained color treatment or surface texture. Keep the text and settings beside each output so the comparison records what actually changed. This is an illustrative writing exercise, not a claim that PicGens generated or tested an accompanying image.

Turn a product idea into a usable poster brief

Separate the product, environment and typography into distinct instructions. For example, describe a simple bottle on a matte stone plinth, a warm background, a narrow side highlight, and clear empty space above for a short headline. The product needs enough room to remain readable after cropping. Keep the title wording separate from the description of the bottle, and say which text should appear if lettering is part of the request. If the output changes the product shape or invents a label, correct that visible miss before increasing the number of decorative elements. Compare the full image and a small thumbnail: a design that looks detailed at full size can lose its hierarchy in a feed. For an actual campaign, check product details and text against approved source material, then finish any inaccurate typography in a design editor. The prompt supplies a visual brief; it does not replace proofing or product review.

Record the context of a style-reference experiment

When studying a shared style code, keep the original prompt, source link and recorded model label together. A code can appear alongside several parameters or reference images, so the visible result should not be attributed to that number alone. Start a comparison by holding the subject and composition steady, then change one style instruction or control. Inspect color, texture, edge treatment and light separately instead of describing the entire result as better or worse. If a style makes the subject hard to read, simplify competing style language before changing the subject. Save the useful result with its actual settings and date. That record makes a later adaptation easier to understand even if the available model version or code interpretation has changed. The catalog examples below are selected from published records that pass the current detail index policy; a source link remains the place to inspect the creator’s context.