Skip to main content

Overview

This node generates or edits images using ByteDance’s Seedream models (versions 4.0, 4.5, and 5.0 Lite). It can create new images from a text prompt or edit existing images by providing reference images, supporting resolutions up to 4K.

Inputs

ParameterDescriptionData TypeRequiredRange
promptText prompt for creating or editing an image.STRINGYesN/A
modelThe Seedream model version to use for generation. Each model has different capabilities and pricing.COMBOYes"seedream 5.0 lite"
"seedream-4-5-251128"
"seedream-4-0-250828"
seedSeed to use for generation (default: 0).INTNo0 to 2147483647
watermarkWhether to add an “AI generated” watermark to the image (default: False).BOOLEANNoTrue / False

Model-Specific Parameters

When you select a model, additional parameters become available:
  • Size Preset: A dropdown to select a predefined image resolution. The available presets depend on the selected model.
  • Width: The width of the generated image in pixels (default: 2048).
  • Height: The height of the generated image in pixels (default: 2048).
  • Max Images: The maximum number of images to generate (default: 1). When set to 1, sequential image generation is disabled.
  • Reference Images: Up to 10 (for Seedream 4.0 and 4.5) or 14 (for Seedream 5.0 Lite) reference images for editing. Images must have an aspect ratio between 1:3 and 3:1.
  • Fail on Partial: If enabled, the node will raise an error if not all requested images are generated successfully (default: False).

Resolution Constraints

  • Seedream 5.0 Lite and 4.5: Minimum resolution is 3.68 megapixels (e.g., 1920x1920).
  • Seedream 4.0: Minimum resolution is 0.92 megapixels (e.g., 960x960).
  • All models: Maximum resolution is 16.78 megapixels (e.g., 4096x4096).

Image Count Constraints

  • The total number of reference images plus generated images cannot exceed 15.
  • For Seedream 5.0 Lite, a maximum of 14 reference images are supported.
  • For Seedream 4.0 and 4.5, a maximum of 10 reference images are supported.

Outputs

Output NameDescriptionData Type
imageThe generated or edited image as a tensor. If multiple images were requested, they are concatenated into a single batch.IMAGE
This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub

Source fingerprint (SHA-256): 6a98d2c1f1e73e304970115bc6dd3ea24ca3cdf440803eed862b7a1ea8d394ce