Midjourney in Action: Mastering Prompts for Stunning Visuals

Midjourney is currently the most popular AI image generation tool, renowned for its high-quality output and strong artistic sense. In this chapter, we will start from scratch and systematically master Midjourney's core operations and prompt engineering techniques.

Basic Operations

Registration and Usage

Midjourney is primarily used through Discord. After registering a Discord account, join the official Midjourney Discord server. In a channel, type the /imagine command followed by your prompt to start generating images. Midjourney has also launched a web version (alpha.midjourney.com), offering a more intuitive interface.

Basic Commands

**/imagine prompt:[your description]** — The core command for generating images. Midjourney will generate 4 images for you to choose from.

**U1-U4** — Upscale the selected image to get a high-resolution version.

**V1-V4** — Generate variations based on the selected image, maintaining the overall style but with some changes.

Prompt Structure

A good Midjourney prompt typically includes the following parts:

Subject Description

Describe the core content of the image. The more specific, the better: 'An orange cat' works better than 'a cat', and 'An orange British Shorthair cat curled up on a windowsill basking in the sun' is even better.

Environment and Background

Describe the scene: indoor/outdoor, time (dusk/dawn), weather, surroundings. For example, 'on a rainy Tokyo street', 'in a sunny Provence lavender field'.

Artistic Style

Specify the desired visual style: 'oil painting style', 'cyberpunk', 'Studio Ghibli animation style', 'minimalism', 'surrealism'. You can also reference specific artists or photographers: 'in the style of Studio Ghibli', 'cinematic lighting like Roger Deakins'.

Technical Parameters

Add technical visual descriptions: 'shallow depth of field', 'golden hour lighting', 'aerial view', 'macro photography'.

Advanced Parameters

Midjourney supports adding parameters after the prompt to finely control the generation:

**--ar 16:9** — Set the aspect ratio. Common ratios: 1:1 (social media avatar), 16:9 (cover/wallpaper), 9:16 (phone wallpaper/short video cover), 3:2 (photographic composition).

**--v 6.1** — Specify the model version. Newer versions usually have better quality but may have different styles.

**--s 100** — Stylization level (0-1000). Higher values give the AI more artistic freedom; lower values make it more faithful to your description.

**--q 2** — Quality parameter. Default is 1; setting it to 2 takes longer but yields higher quality.

Practical Prompt Examples

**Product Showcase**: 'a minimalist cosmetic bottle on a marble surface, soft natural lighting, clean background, product photography, 8k --ar 4:5 --s 50'

**Social Media Graphic**: 'cozy coffee shop interior in autumn, warm tones, steam rising from a latte, soft bokeh, lifestyle photography --ar 1:1'

**Concept Design**: 'futuristic Chinese city skyline at sunset, traditional architecture mixed with high-tech elements, volumetric clouds, cinematic --ar 21:9 --s 200'

Iteration and Optimization

AI image generation rarely produces perfect results on the first try. The correct workflow is: first generate a batch of images with a simple prompt, find the direction closest to what you want, then gradually add detailed descriptions and adjust parameters, achieving satisfactory results through 3-5 rounds of iteration. View the image generation process as a dialogue with the AI, not a one-time command.

After mastering Midjourney, in the next chapter we will learn about the AI design workflow—how to use Canva AI and Instant Design to complete daily design tasks.