Setting Values in a Stable Diffusion Prompt

Setting the right values and parameters in a Stable Diffusion prompt is crucial for generating high-quality AI images. As an emerging art form, crafting effective prompts requires an understanding of the AI model and how different prompt components guide the image generation process.

This article provides beginner-friendly guidance on the key elements to include in a Stable Diffusion prompt. We’ll cover the anatomy of a prompt, best practices for setting sampler values, and techniques to make your prompts more robust. By the end, you’ll have the knowledge to start creating amazing AI artworks!

Anatomy of a Stable Diffusion Prompt

A typical Stable Diffusion prompt contains several components that steer the image generation process:


This is the text that describes the content you want the AI to render. The more detailed and specific the description, the better, as it gives the model more signals.

Good example: “A majestic white Bengal tiger with blue eyes, resting on a tree branch in a sunny jungle”

Sampler Values

These parameters fine-tune how the AI generates images, like the number of steps taken to render details. We’ll explore the key samplers next.


An integer that introduces some randomness, so the same prompt and sampler settings produce different images.

Setting Sampler Values

Samplers control how many “steps” the AI takes to transform noise into the final image. Adjusting these impacts the quality and variety of outputs.


The number of diffusion steps. Higher values like 50 produce more detailed results but take longer to generate. Start with 30-50.


Scales the image down for generation then upsamples for the final render. Lower scales around 7 increase image quality.

CFG Scale

Controls model “creativity” and variation. Higher values induce more randomness. Useful from 6-12.

Advanced Prompt Engineering

Leveraging other techniques in your prompt can further enhance Stable Diffusion outputs:

Text Weights

Wrap keywords in brackets to make them more influential, like [detailed painting]. Heavier weights on key terms boost relevance.

Negative Prompts

Add elements you want to avoid prefixed with “-“, like -poorly drawn, -bad anatomy. This helps exclude unwanted features.

Artistic Style

Referencing a specific artist, era, or genre biases the model to adopt those aesthetics.

Prompt Format Examples

Here are some full prompt examples using the above guidance:

An oil painting portrait in the style of Van Gogh of a girl with red hair and green eyes, very detailed face [detailed painting], 50 steps, CFG scale 10, seed 42

A futuristic cityscape digital art, matte painting, by Asher Brown Durand, -poorly drawn buildings, 30 steps, scale 7, seed 101  


With an understanding of the key components that comprise a Stable Diffusion prompt, you can start crafting effective prompts tailored to your creative vision. Experiment with descriptive language, sampler parameters, and techniques like weights and negative prompts to produce high-fidelity, aesthetically pleasing AI generations.

The world of AI art awaits! Why not put this new knowledge into practice and see what captivating images you can create today using Stable Diffusion?