Midjourney Prompt Builder
Build Midjourney & Stable Diffusion prompts from style, light and lens tags.
Done calculating? Check these tools
Midjourney
Midjourney is a powerful AI image generator known for creating stunning,…
Stable Diffusion
Stable Diffusion is an open-source AI image generation model. It's widely used…
Krea AI
Real-time AI generation canvas with instant image enhancement, video upscale,…
Related categories
About Midjourney Prompt Builder
Tired of writing bare-bones Midjourney / Stable Diffusion prompts? Describe your subject, then tap through a hand-curated bilingual tag library — style, lighting, lens, quality and palette — and watch the fragments assemble into a fluent English prompt. Midjourney mode appends --ar and --stylize automatically; SD mode outputs a clean comma-separated string. 50+ tags cover everything from photorealism to cyberpunk, Rembrandt lighting to neon backlighting, 85mm portraits to macro. Live preview, one-click copy, all local in your browser.
FAQ
- Why do the tags come out in English?
- Midjourney and Stable Diffusion are trained predominantly on English, so English prompts reproduce intent far more faithfully. The Chinese labels exist so Chinese-speaking users can read and pick accurately — the UI language, not the prompt language. Write the subject in English too when you can (a quick translation pass is fine); mixed Chinese is usually understood but less reliable.
- How should I pick the --stylize value?
- Think of it as the model's artistic licence: 0 executes your words literally (product shots, precise compositions); around 250 is the balanced default for most scenes; 750+ lets Midjourney impose its own aesthetic — more striking, less controllable. Use low values for architecture and industrial design, high values for concept art and mood pieces.
- Are SD and Midjourney prompts interchangeable?
- The descriptive part (style/lighting/lens tags) transfers well, but the parameter systems are entirely different: --ar and --stylize are Midjourney-only, while Stable Diffusion takes canvas size from the generator's width/height settings and steers style via samplers, CFG and LoRAs. The SD mode here outputs the plain tag string with no -- params precisely so nothing errors out or gets treated as literal text.