Skip to main content

Create Upscale

Upscales an image to higher resolution using Stability AI's upscaling engines.

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 the upscaled image will be saved. Image is automatically named as upscaled_img.png.
  • Init Image Path - Path to the image file to upscale. Must be a valid image file.

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.

Upscaling Engine​

  • Engine Id - Upscaling engine to use:
    • ESRGAN x2 Upscaler (esrgan-v1-x2plus) - Fast 2x upscaling using ESRGAN (default)
    • Stable Diffusion x4 (stable-diffusion-x4-latent-upscaler) - AI-powered 4x upscaling with prompt guidance

Output Dimensions​

  • Height - Desired height of the output image in pixels. Only specify height OR width, not both.
  • Width - Desired width of the output image in pixels. Only specify width OR height, not both.
warning

You must specify either Width OR Height, but not both. The other dimension will be calculated automatically to maintain aspect ratio.

Stable Diffusion x4 Settings​

The following options are only used when Engine Id is set to "Stable Diffusion x4":

  • Prompt - Text prompt for guiding the upscaling (required for SD x4 engine). Describes the desired quality and characteristics.
  • 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.
  • Cfg Scale - How strictly the diffusion process adheres to the prompt (default: 7). Higher values follow the prompt more closely.

How It Works​

The Create Upscale node enhances image resolution using advanced upscaling techniques. When executed, the node:

  1. Validates the connection or creates a temporary client using provided credentials
  2. Validates the init image path and save path
  3. Validates dimension settings (width OR height, not both)
  4. Reads the source image file
  5. Constructs the upscaling request:
    • For ESRGAN: Simple upscaling with dimension specification
    • For SD x4: Includes prompt and generation parameters
  6. Sends the multipart request to Stability AI upscaling endpoint
  7. Receives the upscaled image
  8. Saves the image as upscaled_img.png in the specified directory

Requirements​

  • Either a valid Connection Id from Connect node OR direct API Key credentials
  • Valid Init Image Path (file must exist)
  • Valid Image Save Path (directory must exist)
  • Either Width OR Height specified (not both)
  • For Stable Diffusion x4: Non-empty Prompt
  • Sufficient Stability AI credits for upscaling

Error Handling​

The node will return specific errors in the following cases:

  • Empty or missing Init Image Path
  • Empty or missing Image Save Path
  • Invalid Connection Id (when not using direct credentials)
  • Both Width and Height empty (must specify one)
  • Width or Height not valid integers
  • Init image file not found
  • Empty Prompt when using Stable Diffusion x4 engine
  • Seed, Steps, or Cfg Scale not valid integers (for SD x4)
  • 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​

Choosing Upscaling Engine​

ESRGAN x2 Upscaler:

  • Fast and efficient 2x upscaling
  • No prompt required
  • Best for: Quick upscaling, photos, general images
  • Good quality without AI interpretation
  • Lower cost

Stable Diffusion x4:

  • AI-powered 4x upscaling
  • Requires prompt for guidance
  • Best for: Maximum quality, artistic images, dramatic enlargement
  • Adds details intelligently based on prompt
  • Higher cost but better quality

Dimension Guidelines​

  • Specify target dimension based on use case
  • Engine will maintain aspect ratio automatically
  • ESRGAN: Up to 2x original dimension
  • SD x4: Up to 4x original dimension
  • Consider output file size and processing time

Prompt Guidelines (SD x4)​

  • Describe desired quality: "high quality", "detailed", "sharp"
  • Mention image type: "photograph", "illustration", "artwork"
  • Include style descriptors: "professional", "crisp", "clear"
  • Example: "High quality professional photograph, sharp details, 8k resolution"

Examples​

Example 1: Fast 2x Upscaling with ESRGAN​

Inputs:

  • Connection Id: (from Connect node)
  • Image Save Path: "/path/to/output"
  • Init Image Path: "/path/to/small_image.jpg"

Options:

  • Engine Id: ESRGAN x2 Upscaler
  • Width: 1024

Use Case: Quick 2x upscaling for web images or thumbnails

Example 2: High-Quality 4x Upscaling with SD x4​

Inputs:

  • Connection Id: (from Connect node)
  • Image Save Path: "/path/to/output"
  • Init Image Path: "/path/to/low_res_photo.jpg"

Options:

  • Engine Id: Stable Diffusion x4
  • Width: 2048
  • Prompt: "High quality professional photograph, sharp details, enhanced clarity, 8k resolution"
  • Steps: 50
  • Cfg Scale: 7

Use Case: Maximum quality upscaling for print or high-resolution display

Example 3: Portrait Upscaling​

Inputs:

  • Connection Id: (from Connect node)
  • Image Save Path: "/path/to/output"
  • Init Image Path: "/path/to/portrait.jpg"

Options:

  • Engine Id: Stable Diffusion x4
  • Height: 2048
  • Prompt: "High quality portrait photograph, detailed facial features, sharp focus, professional photography"
  • Steps: 60
  • Cfg Scale: 8

Use Case: Upscale portraits for professional use

Example 4: Product Photo Enhancement​

Inputs:

  • Connection Id: (from Connect node)
  • Image Save Path: "/path/to/output"
  • Init Image Path: "/path/to/product_thumbnail.jpg"

Options:

  • Engine Id: Stable Diffusion x4
  • Width: 2048
  • Prompt: "High resolution product photography, crisp details, professional e-commerce quality, sharp focus"
  • Steps: 50

Use Case: Enhance product photos for e-commerce listings

Example 5: RPA Use Case - Batch Thumbnail Upscaling​

Flow:

Read thumbnail images from folder
  → Loop through each image
    → Upscale (ESRGAN 2x)
    → Rename to product SKU
    → Upload to website

Inputs:

  • Image Save Path: "/temp/upscaled"
  • Init Image Path: msg.thumbnailPath

Options:

  • Engine Id: ESRGAN x2 Upscaler
  • Width: 1024

Use Case: Quickly upscale batch of product thumbnails for web display

Example 6: RPA Use Case - Archive Photo Restoration​

Flow:

Read low-resolution archive photos
  → Upscale (SD x4 with enhancement prompt)
  → Quality check
  → Save to archive with metadata

Inputs:

  • Image Save Path: "/archive/upscaled"
  • Init Image Path: msg.archivePhoto

Options:

  • Engine Id: Stable Diffusion x4
  • Width: 4096
  • Prompt: "Restored vintage photograph, enhanced details, professional restoration quality, clear and sharp"
  • Steps: 75
  • Cfg Scale: 9

Use Case: Restore and enhance old low-resolution archive photos

Example 7: RPA Use Case - Print Preparation​

Flow:

Read digital images for print
  → Check current resolution
  → If below print quality: Upscale (SD x4)
  → Convert to CMYK
  → Save in print format

Inputs:

  • Image Save Path: "/print_ready"
  • Init Image Path: msg.imagePath

Options:

  • Engine Id: Stable Diffusion x4
  • Height: 3000
  • Prompt: "High resolution print quality, sharp details, professional printing standard"
  • Steps: 60

Use Case: Prepare images for professional printing

Example 8: RPA Use Case - Social Media Asset Creation​

Flow:

Read original artwork/photos
  → Upscale for multiple platforms
  → ESRGAN for quick Instagram posts
  → SD x4 for high-quality hero images
  → Resize to platform specs
  → Upload to content library

ESRGAN for Quick Posts:

  • Engine Id: ESRGAN x2 Upscaler
  • Width: 1080

SD x4 for Hero Images:

  • Engine Id: Stable Diffusion x4
  • Width: 2048
  • Prompt: "High quality social media image, vibrant colors, sharp details, eye-catching"

Example 9: Multiple Resolution Outputs​

Flow:

Source Image (512x512)
  → Upscale to 1024 (ESRGAN)
  → Save as medium resolution
  → Upscale to 2048 (SD x4)
  → Save as high resolution

First Upscale (ESRGAN):

  • Width: 1024
  • Engine: ESRGAN x2

Second Upscale (SD x4):

  • Width: 2048
  • Engine: SD x4
  • Prompt: "Maximum quality, sharp details"

Use Case: Create multiple resolution variants for different use cases

Example 10: Reproducible Upscaling​

Inputs:

  • Connection Id: (from Connect node)
  • Image Save Path: "/path/to/output"
  • Init Image Path: "/path/to/image.jpg"

Options:

  • Engine Id: Stable Diffusion x4
  • Width: 2048
  • Prompt: "High quality upscale, enhanced details"
  • Seed: 42
  • Steps: 50
  • Cfg Scale: 7

Use Case: Use same seed to reproduce exact upscaling results

Best Practices​

  1. Engine Selection:

    • Use ESRGAN for quick, straightforward upscaling
    • Use SD x4 when you need maximum quality and detail enhancement
    • ESRGAN is faster and cheaper
    • SD x4 provides better results for complex images

  2. Dimension Planning:

    • Don't exceed 4x original size even with SD x4
    • Consider final use case (web, print, display)
    • Larger outputs require more processing time and credits
    • Maintain aspect ratio by specifying only one dimension

  3. Prompt Writing (SD x4):

    • Be specific about desired quality
    • Include image type and style
    • Mention target use case (print, display, etc.)
    • Use quality descriptors: "sharp", "detailed", "clear", "professional"

  4. Quality Optimization:

    • For SD x4: Use 50-75 steps for best quality
    • Higher Cfg Scale (8-10) for more prompt adherence
    • Generate test upscale before batch processing
    • Compare ESRGAN vs SD x4 results for your use case

  5. Cost Management:

    • ESRGAN is more cost-effective for batch operations
    • Use SD x4 selectively for high-priority images
    • Test with smaller dimensions first
    • Monitor credit usage

  6. File Management:

    • Upscaled images are named upscaled_img.png
    • Rename immediately if processing multiple images
    • Organize output by date or category
    • Keep original images as backup

  7. Batch Processing:

    • Process in batches with appropriate engine
    • Implement error handling for missing files
    • Add progress logging
    • Consider parallel processing for ESRGAN

Tips for RPA Developers​

  • Smart Engine Selection: Auto-select engine based on target size (2x = ESRGAN, 4x = SD x4)
  • Quality Checks: Implement automated checks for upscale success
  • File Renaming: Immediately rename output file to avoid overwriting
  • Progress Tracking: Log each successful upscale operation
  • Fallback Strategy: If SD x4 fails, try ESRGAN as fallback
  • Dimension Validation: Calculate and validate target dimensions before processing
  • Batch Optimization: Group similar images for consistent processing
  • Error Recovery: Retry failed upscales with different parameters

Common Errors and Solutions​

Error: "Width or Height cannot be empty"

  • Solution: Specify either Width OR Height (not both)

Error: "Prompt is required when using Stable Diffusion x4"

  • Solution: Provide a descriptive prompt for SD x4 engine

Error: "Image file not found"

  • Solution: Verify init image path is correct and file exists

Error: "Width must be a valid integer"

  • Solution: Ensure Width value is a whole number

Poor upscaling results:

  • Solution: Try SD x4 with detailed prompt instead of ESRGAN
  • Solution: Increase steps to 60-75 for SD x4
  • Solution: Adjust Cfg Scale (try 8-10)
  • Solution: Ensure source image is good quality

Output file already exists:

  • Solution: Rename or move previous upscaled_img.png before processing next image
  • Solution: Use unique output directories for each image