Stable Diffusion offers a straightforward approach to writing prompts and provides the flexibility of allowing separate sections for positive and negative prompts. In this guide, we will help you with writing prompts and include tips and techniques to write the best prompts.
What we cover
Write Prompts on Stable Diffusion
- Start by loading Stable Diffusion on your PC. For demonstration, we have considered Automatic1111’s Stable Diffusion WebUI.
- After it is loaded, look for two input boxes that are assigned to enter positive and negative prompts.
- Once you have written the prompts, click on the Generate button to start the image generation process using Stable Diffusion.
Tips for Writing Better Prompts on Stable Diffusion
Get the prompt structure right
The most crucial part to consider while writing a prompt on Stable Diffusion is the clear and precise structure to guide Stable Diffusion effectively. Hence, make sure that your prompt has the following:
- Subject: The main focus of the image needs to be included at the start – A dog playing with a ball
- Medium: Describe the type of art for more accurate images – Digital art
- Style: Mentioning the style helps generate the image as imagined – Hyperrealistic, fantasy, etc.
- Resolution: It is important to specify keywords that represent resolution for Stable Diffusion to generate clear – highly detailed, sharp focus, etc.
- Colour and Lighting: Alongside the above, it is essential to specify the colour tones and lighting effects to make the image feel lively –Iridescent gold, studio lighting, etc.
- Additional Details: Never skip any detail that is needed for image generation, so ensure to include it in the positive or negative prompt field.
Write the negative prompt for more accurate results
The negative prompt specifies what should be excluded from the image to improve quality. With the inputs provided as a negative prompt, Stable Diffusion ensures that the specific content isn’t included in the image generation.
Example:
- Positive: “A tall and handsome prince, smile, holding a sword, natural light, detailed war armour, war ground background.”
- Negative: “disfigured, deformed, ugly”
Stable Diffusion Prompt Techniques
Keyword Weight
Modifies how strongly the model considers a particular keyword in the prompt. The “factor” adjusts the keyword’s weight. A value greater than 1 increases its importance, while a value less than 1 decreases it.
- Syntax: (keyword: factor)
- Example: “cat (cat: 1.5)” – This increases the priority of cats in the generated image.
() and [] Syntax
Specify the importance of the subject or specific keywords using parentheses and brackets. Parentheses enhance the keyword’s impact, and brackets reduce it. Multiple parentheses or brackets can be used for a more significant effect.
Examples:
- “(sunset)” is equivalent to “(sunset: 1.1)”, slightly increasing the importance of “sunset”.
- “[sunset]” is equivalent to “(sunset: 0.9)”, slightly decreasing the importance of “sunset”.
- “((sunset))” increases the impact further, equivalent to “(sunset: 1.21)”.
Keyword Blending
Blend two different keywords, specifying how they transition during the image generation process. This technique allows for a smooth transition between two keywords at a specified factor. The factor ranges from 0 to 1, where 0 means the image is entirely keyword1, and 1 means the image is entirely keyword2.
- Syntax: [keyword1: keyword2: factor]
- Example: “Watercolor painting of [a mountain: a forest: 0.4]” – The first 40% of the generation process focuses on “a mountain,” and the remaining 60% shifts towards “a forest,” blending elements of both.
Wrapping Up
Congratulations on learning how to write prompts on Stable Diffusion!
The image generation possibilities are endless with Stable Diffusion through the help of prompts. The tips and techniques we have provided help you write better prompts. While they come in handy, we suggest you experiment and get creative by thinking out of the box.
Related Reading
What Is a Negative Prompt in Stable Diffusion
What Is Denoising Strength Stable Diffusion
What Is CFG Scale in Stable Diffusion
