Stable Diffusion Prompt Formula: The Complete Guide

Master positive prompts, negative prompts, and weights to generate perfect Stable Diffusion images every single time.

How Stable Diffusion prompts differ from Midjourney

Stable Diffusion gives you more direct control over image generation than Midjourney. While Midjourney interprets prompts creatively, Stable Diffusion responds more literally — which means you have more power, but also more responsibility to write precise prompts.

The key differences: SD uses positive AND negative prompts, supports prompt weighting, and responds strongly to quality tags like (masterpiece:1.4).

The Stable Diffusion prompt formula

The most reliable SD prompt structure is:

[Quality tags], [Subject], [Style], [Lighting], [Camera/Medium], [Artist reference]

Essential quality booster tags

Always include these at the start of your positive prompt for higher quality results:

• (masterpiece:1.4) — Signals highest quality generation
• (best quality:1.2) — Improves overall quality
• 8k uhd — Ultra high definition
• highly detailed — More fine details
• sharp focus — Better clarity
• photorealistic or hyperrealistic — For realistic images

Prompt weighting syntax

Stable Diffusion supports weighting specific parts of your prompt to give them more or less emphasis:

• (word:1.5) — Increases emphasis by 50%
• (word:0.8) — Reduces emphasis by 20%
• [word] — Slight decrease in emphasis
• (word) — Slight increase in emphasis (1.1x)

Example: (masterpiece:1.4), portrait of a (beautiful woman:1.2), (long red hair:1.3), cyberpunk city background, neon lighting

Negative prompts — the secret weapon

Negative prompts tell SD what NOT to include. This is one of the most powerful tools you have for improving quality. A good universal negative prompt:

ugly, deformed, noisy, blurry, distorted, out of focus, bad anatomy, extra limbs, poorly drawn face, poorly drawn hands, missing fingers, extra fingers, fused fingers, too many fingers, long neck, mutation, mutated, extra arms, extra legs, watermark, signature, text, logo

Complete prompt example

Positive:

(masterpiece:1.4), (best quality:1.2), cinematic portrait, beautiful woman, long auburn hair, wearing elegant dress, sitting in a café, warm golden hour lighting, bokeh background, 35mm film, photorealistic, highly detailed, sharp focus

Negative:

ugly, deformed, blurry, bad anatomy, extra limbs, watermark, text, low quality, worst quality, normal quality

CFG Scale explained

CFG (Classifier Free Guidance) scale controls how strictly SD follows your prompt. Lower values = more creative but less accurate. Higher values = more literal but potentially over-saturated.

• 5-7 — Creative, loose interpretation, good for artistic styles
• 7-9 — Balanced (recommended default)
• 10-15 — Very literal, can become over-processed