Skip to main content

Create Images From Text

Generates images from text prompts using Stability AI's Stable Diffusion models.

Common Properties​

  • Name - The custom name of the node.
  • Color - The custom color of the node.
  • Delay Before (sec) - Waits in seconds before executing the node.
  • Delay After (sec) - Waits in seconds after executing node.
  • Continue On Error - Automation will continue regardless of any error. The default value is false.
info

If the ContinueOnError property is true, no error is caught when the project is executed, even if a Catch node is used.

Inputs​

  • Connection Id - The connection ID from the Connect node (optional if API Key is provided directly).
  • Image Save Path - Directory path where generated images will be saved. Images are automatically named as txt2img_0.png, txt2img_1.png, etc.
  • Prompt - Text description of the desired image. Be specific and descriptive for best results.

Options​

Authentication​

  • API Key - Stability AI API key (optional if using Connection ID). You can provide the API key directly instead of using a Connect node.

Model Selection​

  • Engine Id - Stability AI engine to use for image generation:
    • Stable Diffusion XL 1.0 - Latest SDXL model, highest quality 1024x1024 images
    • Stable Diffusion XL 0.9 - Previous SDXL version
    • Stable Diffusion 1.5 - Classic SD 1.5, balanced performance (default)
    • Stable Diffusion 2.1 - SD 2.1 for 512x512 images

Generation Settings​

  • Negative Prompt - Text describing elements to avoid in the generated image. Helps exclude unwanted features.
  • Height - Height of the image in pixels (must be in increments of 64). Leave empty for default engine resolution.
  • Width - Width of the image in pixels (must be in increments of 64). Leave empty for default engine resolution.
  • Cfg Scale - How strictly the diffusion process adheres to the prompt (default: 7). Range: 1-35. Higher values follow the prompt more closely.
  • Samples - Number of images to generate (1-10, default: 1). Generates multiple variations from the same prompt.
  • Seed - Random noise seed (default: 0 for random). Use specific values for reproducible results.
  • Steps - Number of diffusion steps to run (10-150, default: 50). More steps = better quality but slower generation.

Advanced Settings​

  • Clip Guidance Preset - CLIP guidance preset for image generation quality:

    • NONE - No CLIP guidance (default)
    • FAST_BLUE - Fast with blue color bias
    • FAST_GREEN - Fast with green color bias
    • SIMPLE - Simple CLIP guidance
    • SLOW - More precise guidance
    • SLOWER - Even more precise
    • SLOWEST - Maximum precision

  • Sampler - Sampling algorithm used for diffusion process:

    • NONE - Use engine default (recommended)
    • DDIM - Denoising Diffusion Implicit Models
    • DDPM - Denoising Diffusion Probabilistic Models
    • K_DPMPP_2M - DPM++ 2M Karras
    • K_DPMPP_2S_ANCESTRAL - DPM++ 2S Ancestral
    • K_DPM_2 - DPM 2
    • K_DPM_2_ANCESTRAL - DPM 2 Ancestral
    • K_EULER - Euler method
    • K_EULER_ANCESTRAL - Euler Ancestral
    • K_HEUN - Heun method
    • K_LMS - Linear Multi-Step

  • Style Preset - Style preset to guide the image generation:

    • NONE - No style preset (default)
    • 3d-model - 3D rendered style
    • analog-film - Analog film photography
    • anime - Anime/manga style
    • cinematic - Cinematic movie style
    • comic-book - Comic book art
    • digital-art - Digital artwork
    • enhance - Enhanced details
    • fantasy-art - Fantasy illustration
    • isometric - Isometric perspective
    • line-art - Line art/sketch
    • low-poly - Low polygon 3D
    • modeling-compound - Clay modeling style
    • neon-punk - Cyberpunk neon
    • origami - Paper folding style
    • photographic - Realistic photography
    • pixel-art - 8-bit pixel art
    • tile-texture - Seamless texture

How It Works​

The Create Images From Text node generates images from text descriptions using Stable Diffusion. When executed, the node:

  1. Validates the connection or creates a temporary client using provided credentials
  2. Validates the prompt and image save path
  3. Constructs the generation request with:
    • Main prompt (positive weight)
    • Negative prompt if provided (negative weight)
    • Dimension settings (width/height)
    • Generation parameters (steps, cfg_scale, seed, etc.)
    • Style and sampler settings
  4. Sends the request to Stability AI API
  5. Receives base64-encoded images from the API
  6. Decodes and saves images to the specified directory
  7. Images are named sequentially: txt2img_0.png, txt2img_1.png, etc.

Requirements​

  • Either a valid Connection Id from Connect node OR direct API Key credentials
  • Non-empty Prompt
  • Valid Image Save Path (directory must exist)
  • Sufficient Stability AI credits for image generation

Error Handling​

The node will return specific errors in the following cases:

  • Empty or missing Prompt
  • Empty or missing Image Save Path
  • Invalid Connection Id (when not using direct credentials)
  • Height or Width not valid integers
  • Height or Width not in increments of 64
  • Cfg Scale, Samples, Seed, or Steps not valid integers
  • Samples outside range of 1-10
  • Steps outside range of 10-150
  • API authentication errors (401)
  • Insufficient credits (402)
  • API rate limit errors (429)
  • API service errors (500, 503)

Usage Notes​

Prompt Engineering​

  • Be specific and detailed in your prompts
  • Include style descriptors (e.g., "oil painting", "digital art", "photorealistic")
  • Specify important details (lighting, composition, colors)
  • Mention artists or art styles for specific aesthetics
  • Use comma-separated descriptors for clarity

Negative Prompts​

  • Use to exclude unwanted elements (e.g., "blurry, low quality, distorted")
  • Common negative prompts: "ugly, deformed, low resolution, watermark"
  • Helps avoid common AI artifacts

Image Dimensions​

  • Must be multiples of 64 (e.g., 512, 576, 640, 704, 768, 832, 896, 960, 1024)
  • Different engines have recommended sizes:
    • SDXL models: 1024x1024 (square) or 1024x768, 768x1024 (landscape/portrait)
    • SD 1.5: 512x512 (square) or 512x768, 768x512
    • SD 2.1: 512x512 or 768x768
  • Larger images consume more credits and take longer to generate

Generation Parameters​

  • Cfg Scale: 7-10 for most use cases, 15-20 for very strict prompt following
  • Steps: 30-50 for good quality, 50-150 for maximum quality
  • Seed: Use 0 for random, or specific number to reproduce exact images
  • Samples: Generate multiple variations to choose the best result

Examples​

Example 1: Basic Image Generation​

Inputs:

  • Connection Id: (from Connect node)
  • Image Save Path: "/path/to/output"
  • Prompt: "A serene mountain landscape at sunset, oil painting style"

Options:

  • Engine Id: Stable Diffusion 1.5
  • Steps: 50
  • Cfg Scale: 7

Result: Generates a single 512x512 image saved as /path/to/output/txt2img_0.png

Example 2: High-Quality SDXL Generation​

Inputs:

  • Connection Id: (from Connect node)
  • Image Save Path: "/path/to/output"
  • Prompt: "Professional portrait photo of a software developer working on code, natural lighting, modern office, highly detailed, 8k quality"
  • Negative Prompt: "cartoon, anime, illustration, blurry, low quality"

Options:

  • Engine Id: Stable Diffusion XL 1.0
  • Width: 1024
  • Height: 1024
  • Steps: 40
  • Cfg Scale: 8
  • Style Preset: photographic

Result: High-quality photorealistic 1024x1024 portrait

Example 3: Multiple Style Variations​

Inputs:

  • Connection Id: (from Connect node)
  • Image Save Path: "/path/to/output"
  • Prompt: "A fantasy dragon perched on a castle tower, epic scene"

Options:

  • Engine Id: Stable Diffusion XL 1.0
  • Samples: 4
  • Steps: 50
  • Seed: 0

Result: Generates 4 different variations: txt2img_0.png through txt2img_3.png

Example 4: Specific Style - Anime​

Inputs:

  • Connection Id: (from Connect node)
  • Image Save Path: "/path/to/output"
  • Prompt: "A cute anime girl with blue hair, magical girl outfit, sparkles, vibrant colors"
  • Negative Prompt: "realistic, photo, 3d render"

Options:

  • Engine Id: Stable Diffusion 1.5
  • Style Preset: anime
  • Steps: 40
  • Cfg Scale: 10

Result: Anime-style character illustration

Example 5: Reproducible Results​

Inputs:

  • Connection Id: (from Connect node)
  • Image Save Path: "/path/to/output"
  • Prompt: "A red sports car on a coastal highway"

Options:

  • Engine Id: Stable Diffusion 1.5
  • Seed: 12345
  • Steps: 50
  • Cfg Scale: 7

Result: Using the same seed will generate the exact same image every time

Example 6: RPA Use Case - Product Visualization​

Flow:

Read CSV (product descriptions)
  → Loop through products
    → TextToImages (generate product image from description)
    → Rename file to product SKU
    → Upload to e-commerce platform

Inputs:

  • Prompt: msg.productDescription (from CSV)
  • Image Save Path: "/temp/product_images"

Options:

  • Engine Id: Stable Diffusion XL 1.0
  • Style Preset: photographic
  • Width: 1024
  • Height: 1024

Example 7: RPA Use Case - Marketing Material Generation​

Flow:

Read campaign requirements
  → Generate hero image (TextToImages)
  → Generate thumbnail variations (TextToImages with Samples: 5)
  → Select best image (manual or AI selection)
  → Resize and optimize
  → Upload to content management system

Inputs:

  • Prompt: "Modern tech startup office, collaborative workspace, diverse team working together, bright and energetic atmosphere, professional photography"
  • Negative Prompt: "dark, gloomy, empty, stock photo watermark"

Options:

  • Engine Id: Stable Diffusion XL 1.0
  • Width: 1920
  • Height: 1080
  • Samples: 3
  • Style Preset: photographic

Example 8: RPA Use Case - Social Media Automation​

Flow:

Read social media calendar
  → For each scheduled post
    → TextToImages (generate post image)
    → Add text overlay (Image Processing)
    → Resize for platform (Instagram, Twitter, etc.)
    → Upload to social media scheduler

Inputs:

  • Prompt: msg.postTheme + ", eye-catching, social media style, vibrant colors"
  • Image Save Path: "/temp/social_images"

Options:

  • Engine Id: Stable Diffusion XL 1.0
  • Width: 1024
  • Height: 1024
  • Style Preset: digital-art

Best Practices​

  1. Prompt Quality:

    • Start with the main subject
    • Add style descriptors
    • Include quality terms ("highly detailed", "8k", "professional")
    • Specify composition ("centered", "wide angle", "close-up")

  2. Negative Prompts:

    • Always use negative prompts to avoid common artifacts
    • Include: "blurry, low quality, distorted, deformed"
    • Add medium-specific exclusions (e.g., "cartoon" for photographic style)

  3. Engine Selection:

    • SDXL 1.0: Best overall quality, higher cost
    • SD 1.5: Balanced performance, good for most use cases
    • SD 2.1: Alternative style, good for certain artistic styles

  4. Cost Management:

    • Start with lower step counts (30-40) for testing
    • Generate multiple samples to get variations
    • Use specific seeds to reproduce good results
    • Monitor credit usage with GetUserCredit node

  5. Quality Optimization:

    • Use 40-50 steps for production quality
    • Cfg Scale 7-12 for most use cases
    • Enable appropriate style presets
    • Use proper dimensions for your engine

  6. File Management:

    • Ensure output directory exists before running
    • Implement file naming conventions for batch processing
    • Clean up old generated images periodically
    • Consider organizing by date or category

Tips for RPA Developers​

  • Batch Processing: Use loops to generate multiple images from data sources
  • Error Recovery: Implement retry logic for API failures
  • Credit Monitoring: Check credits before large batch operations
  • Quality Control: Generate samples first, then scale up production
  • A/B Testing: Generate variations with different parameters to find optimal settings
  • Prompt Templates: Create reusable prompt templates for consistent results
  • Caching Seeds: Store successful seeds for reproducible results