JSON prompts offer significant advantages over traditional text prompts by providing structured, machine-readable data that enables precise control over image generation parameters. Unlike free-form text which can be ambiguous and subject to interpretation, JSON allows for consistent, reproducible results across different AI models and sessions. This structured approach ensures that specific settings like style, lighting, mood, and composition are explicitly defined, reducing variability and improving the reliability of generated images. Additionally, JSON prompts facilitate easier automation, batch processing, and integration with other tools, making them ideal for professional workflows and consistent creative output.
The prompt is the core description of the image you want to generate. It serves as the primary instruction for the AI model, defining the main subject, scene, and key elements.
Impact: This is the most influential parameter as it directly communicates your creative vision to the AI. A well-crafted prompt can significantly improve the relevance and quality of the generated image.
Examples:
Best Practices: Be specific but concise. Include key visual elements, setting, and mood. Avoid contradictory descriptions.
The negative_prompt specifies elements you want to exclude from the generated image. It helps refine the output by preventing unwanted features.
Impact: Acts as a filter to improve image quality by removing common artifacts, distortions, or undesired elements. Particularly useful in models like Stable Diffusion.
Examples:
Best Practices: Focus on technical issues and unwanted elements. Use commas to separate multiple exclusions.
The style parameter influences the artistic rendering technique and visual approach of the generated image.
Impact: Determines the overall aesthetic and technique, affecting brushwork, color application, and artistic interpretation.
Available Options:
Examples: Using "oil painting" with a landscape prompt creates a traditional art piece, while "cyberpunk" transforms the same scene into a futuristic metropolis.
The lighting parameter controls the illumination and shadow effects in the generated image.
Impact: Lighting dramatically affects mood, depth, and realism. It can create atmosphere, guide attention, and enhance the three-dimensional feel of the image.
Available Options:
Examples: "Golden Hour" creates warm, nostalgic scenes perfect for landscapes, while "dramatic shadows" adds tension and mystery to portraits.
The mood parameter conveys the emotional tone and atmosphere of the generated image.
Impact: Influences color palette, lighting intensity, and overall emotional response. Helps AI models interpret the desired atmosphere beyond literal descriptions.
Available Options:
Examples: "Eerie" enhances horror scenes with dim lighting and unsettling compositions, while "joyful" brightens colors and adds playful elements to celebrations.
The color_scheme parameter defines the dominant color palette and visual tone of the image.
Impact: Controls the overall color temperature, saturation, and emotional impact. Multiple selections allow for complex color combinations.
Available Options:
Examples: Combining "earth tones" with "pastel" creates soft, natural landscapes, while "neon" with "high contrast" produces vibrant cyberpunk scenes.
The composition parameter guides the arrangement and placement of elements within the image frame.
Impact: Affects visual flow, balance, and focal points. Proper composition can make images more engaging and professional-looking.
Available Options:
Examples: "Rule of Thirds" creates more dynamic landscapes, while "close-up" emphasizes details in portraits.
The aspect_ratio parameter controls the proportional dimensions of the generated image.
Impact: Determines the shape and format of the output, affecting how the composition fits within the frame.
Available Options:
Examples: "16:9" is ideal for cinematic landscapes, while "9:16" works well for mobile-optimized portraits.
The quality parameter influences the level of detail, resolution, and refinement in the generated image.
Impact: Higher quality settings produce more detailed and polished results, though they may require more processing time.
Available Options:
Examples: "Ultra HD" produces highly detailed technical illustrations, while "minimalist" creates clean, modern designs.
The seed parameter is a numerical value that initializes the random number generator in diffusion models.
Impact: Ensures reproducible results. Using the same seed with identical parameters will generate very similar images.
How to Use: Enter any integer value. The same seed with the same prompt and settings will produce consistent results across generations.
Examples:
Best Practices: Use fixed seeds when you want to iterate on specific compositions or when sharing prompts for consistent results.
Custom Keys allow you to add additional parameters not covered by the standard options.
Impact: Extends functionality for advanced users and specific model requirements. Enables fine-tuning of generation parameters.
How to Use: Click "Add Custom Key" to create new key-value pairs. Enter the parameter name and its value.
Examples:
Best Practices: Only add custom keys if you understand their purpose. Incorrect values may produce unexpected results or errors.