Midjourney step-by-step: prepare a scene, choose a mode, and check your first image
Use the Midjourney Image entry to prepare a prompt or reference image, check current mode options, and iterate with a composition and detail checklist.
The on-site Midjourney product entry currently displays Midjourney Image and keeps its legacy URL. The name does not confirm a particular upstream version such as V8.1.
This tutorial focuses on scene preparation and output checks. The current catalog lists text-to-image, image-to-image, and image-edit modes. Use the modes, required fields, and quotes actually offered in the workbench.
Step 1: prepare a visual brief
Describe subject, setting, style, lighting, composition, and exclusions. List logos, text, and product shapes that must be preserved as separate checks.
A coastal lighthouse in morning mist, soft side lighting, realistic illustration, main subject on the right, empty space on the left for a headline, no added text or watermark.
Step 2: match the mode to your input
| Mode | Prepare | Check before submission |
|---|---|---|
| Text-to-image | A scene description | Current generation type, mode, and quality options |
| Image-to-image | Reference image and intended changes | Image requirements and current input fields |
| Image editing | An image and a specific editing request | Current editing form and limits; do not assume arbitrary local editing tools |
The catalog includes fields such as prompt, generation_type, images, mode, and quality. Requirements and choices can differ by mode. Discord commands, version flags, and aspect-ratio requests in prose are not proof of an available form control.
Step 3: submit and record your settings
Check the selected name, mode, input requirements, and cost before submitting. Record the prompt and actual settings for comparison. Generated images may not satisfy every written constraint.
Check the first image
| Check | What to inspect |
|---|---|
| Subject | Count, position, and shape |
| Composition | Empty space and visual balance |
| Lighting and palette | Consistency with the intended scene |
| Details | Hands, edges, logos, and text errors |
| Exclusions | Whether unwanted elements still appear |
Change one major factor per round: clarify the subject first if it is unclear, or adjust spatial relationships if composition fails. Save useful outputs with their settings rather than adding style words without diagnosing the problem.
Common issues
- Missing parameter: follow the current mode form, not a historical version tutorial.
- Reference mismatch: specify which features must remain and check image input requirements.
- Incorrect text or branding: inspect it directly and compare other image models if necessary; no model is guaranteed to preserve every detail.