​
Stable Diffusion 3.5: Prompting and Options Guide

Stable Diffusion 3.5 offers advanced AI image generation capabilities. This guide will help you understand how to craft effective prompts and use the available options to create stunning images.

​
Crafting Effective Prompts

The prompt is the most crucial input for generating images with Stable Diffusion 3.5. Here are some tips for writing effective prompts:
  1. Be Specific and Descriptive: Provide detailed descriptions of what you want to see in the image.
  2. Use Natural Language: Write as if you’re describing the image to another person.
  3. Specify Visual Elements: Include details about colors, textures, lighting, and composition.
  4. Mention Styles or Artists: Reference specific art styles or artists to guide the aesthetic.
  5. Balance Detail and Creativity: While being specific, leave some room for AI interpretation.
Example prompt: A hyperrealistic close-up portrait of an elderly Inuit woman in her 80s, with deep wrinkles, weathered skin, and wise, piercing dark eyes. Her silver hair is braided and adorned with traditional bone beads. She’s wearing a fur-trimmed parka. Capture the soft, golden light of the Arctic midnight sun reflecting in her eyes and on her skin.

​
Available Options

Stable Diffusion 3.5 provides several options to fine-tune your image generation:

​
Image Size

Choose from predefined sizes:
  • square_hd
  • square
  • portrait_4_3
  • portrait_16_9
  • landscape_4_3
  • landscape_16_9
Specify a custom width and height in pixels. This option allows for more precise control over the output dimensions.

​
Inference Steps

Controls the number of denoising steps in the generation process.
  • Range: 1-50
  • Default: 28
  • Higher values generally result in more detailed images but increase generation time.

​
Guidance Scale

Determines how closely the image adheres to your prompt.
  • Range: 1-20
  • Default: 3.5
  • Lower values allow more creative freedom, while higher values stick closer to the prompt.

​
Seed

A number that initializes the random generation process.
  • Optional: Leave blank for a random seed each time
  • Specifying a seed allows for reproducible results
  • Useful for creating variations of an image by keeping the seed constant and adjusting other parameters

​
Using the Interface

  1. Enter Your Prompt: Type your detailed image description in the main text area.
  2. Adjust Settings: Click the “Settings” button to access and modify the available options.
  3. Generate: Click the “Generate” button to create your image based on the prompt and chosen settings.
  4. View and Download: Once generated, you can view, download, or share your images to a gallery.

​
Tips for Best Results

  • Experiment with different prompt phrasings to see how they affect the output.
  • Try various combinations of inference steps and guidance scale to find the ideal balance for your needs.
  • Use the seed option to create variations of an image you like by slightly modifying other parameters.
  • Enable sync mode when you want to ensure each generation completes before starting the next.
Remember, the key to getting the best results from Stable Diffusion 3.5 is experimentation. Don’t hesitate to try different prompts and settings combinations to achieve your desired outcome.