Midjourney prompt improver
Paste a rough prompt, get a model-ready version that follows Midjourney's conventions. Free. No sign-in.
no sign-in · nothing stored · free
What good Midjourney prompts look like
Midjourney prompts describe an image rather than instructing a conversational assistant. Its current documentation recommends short, clear phrases and notes that fewer details allow more variety while specific details give more control. A useful rewrite therefore chooses the details that matter: subject, medium, environment, lighting, color, mood, and composition. It does not bury the image under a screenplay or promise that every named detail will appear. Parameters are a separate suffix at the end of the prompt.
Describe the finished image, not the act of making it
Use a compact visual phrase such as "colored pencil illustration of three orange poppies" instead of "please make me a picture and make it colorful." Name the subject first, then the medium or photographic treatment. Add an environment only when it contributes to the scene. Midjourney's own prompt guidance favors specific words and concrete numbers over vague plurals, so "three cyclists" communicates a clearer composition than "some cyclists."
Spend detail on visible decisions
Choose details that could be seen in the image: material, era, weather, light direction, palette, camera position, framing, or negative space. Business goals such as "premium" or "high-converting" need translation into visual choices. Do not stack near-synonyms to force an effect. A short phrase that defines the subject and two or three important art-direction decisions leaves room for variation without surrendering the brief.
State what you want before excluding anything
Midjourney recommends focusing the main text on desired content. If an unwanted object matters, use the `--no` parameter at the end rather than a conversational phrase such as "without any text." Keep exclusions precise. The official documentation warns that words after `--no` are read independently, so a compound phrase can have a broader effect than expected. It is often safer to describe the wanted clothing, setting, or era positively.
Put parameters at the end and keep their syntax clean
Parameters follow the descriptive text, with a space before each double dash. Use `--ar` for the aspect ratio, written as whole-number proportions such as `--ar 4:5` or `--ar 16:9`. Use `--stylize` or `--s` when you deliberately want to adjust the balance between prompt detail and Midjourney's artistic interpretation. Do not insert prompt text after the parameter suffix, and do not add punctuation to ordinary parameter values.
Use image references for guidance, not exact edits
On the Midjourney website, an uploaded image can be added as an Image Prompt. In Discord, a valid image URL goes at the beginning of the prompt. The accompanying text should describe everything wanted in the final image, not issue editing commands about the reference. Midjourney describes image prompts as inspiration for content, composition, and color rather than exact copying. For a precise alteration to an existing image, its Editor is the more appropriate surface.
Pin settings only when reproducibility needs them
Aspect ratio belongs in the prompt when the destination has a fixed shape. Other controls, including Raw, stylization, references, and model version, should be added for a reason rather than copied from a prompt recipe. Raw reduces Midjourney's automatic styling and can give a detailed art direction more control. Version syntax and feature compatibility change over time, so record a version for a repeatable production workflow and check the current compatibility table before reusing an old suffix.
Before and after
The rewrite turns "nice" into visible choices: one subject, a specific action, setting, light, palette, mood, camera position, texture, and usable headline space. The `--ar` suffix matches a wide website hero. A moderate stylize value is explicit rather than assumed, and `--no lettering` requests a clean base image without putting a negative instruction in the scene description.