Image Generation

Process text and generate human-like responses with advanced AI models.

What is the Image Generation?

The Image Generation node uses state-of-the-art image models provided by AWS Bedrock, OpenAI, and Google AI to convert text prompts into images. Whether you're creating art, mockups, or visual storytelling, this node gives you the power to bring text to life visually.

How to use it?

1. Select a Model:

   Choose from a variety of supported image generation models:

   • AWS Bedrock Image Models (e.g. TITAN, Stable Diffusion)
   • OpenAI Image Models (e.g., DALL·E)
   • Google AI Image Models (e.g., Imagen)

2. Configure Model Parameters:

   • Prompt: A detailed description of the image you want to generate.
   • Image Size: Common sizes include 512x512, 1024x1024, etc.
   • Seed: The Stability AI Stable Diffusion 3.5 Large model offers a seed value. It ensures the same prompt returns the same output. We need to change it if we want a different image for the same prompt.
   • Model-Specific Options:: Most models have specific parameters, like style presets or negative prompts. See the options on your specific models and documentation for the model.

3. Provide Input:

   • Connect a string input to the Prompt anchor or provide a static prompt input directly.
   • (Optional) Use workflow triggers or additional control inputs if supported (e.g., mask for inpainting, reference image for variations).

4. Set Up Output:

   • The output is an image directly in the ui or for api executions an URL to the generated image.

5. Execute the Component:

   • Run the workflow to invoke the selected model and generate the image(s).

Example Task: Sportswear Brand Visual Generation Workflow

Objective: Create a workflow that generates marketing visuals for a sportswear brand using branded visual guidelines and a text-to-image model.

Step-by-Step Setup

1. Add Brand Visuals Guide (Text Input):

   • Add a Text Input node to your canvas.

   • Set the template to:

     Brand visuals:
     
     Modern, bold, and high-performance. Show athletes in motion — sprinting, lifting, training — with energy and focus. Use dramatic lighting, bold colors, and urban or rugged outdoor settings. Clothing should look sleek, functional, and stylish. Emphasize movement, power, and confidence.

2. Add Prompt Builder (Text Input with Variable):

   • Add a second Text Input node.

   • Set the template to:

     Generate visuals for our sportswear brand. We release a new pair of running shoes. Consider the following guide on our brand visuals:
     {input-0}

   • Add an input variable and connect it to the output of the Brand Visuals node (from Step 1).

3. Add Image Generation Tool:

   • Add the Image Generation tool node (e.g., Stability AI Stable Diffusion 3.5 Large).
   • Set the model, credentials, region, and image size as needed.
   • Set the Seed (e.g., 0) for reproducibility.
   • Connect the output from the Prompt Builder node (Step 2) into the Prompt field of the image generation tool.

4. Add Output Node:

   • Add an Output node and set the type to File.
   • Connect the Image output from the Image Generation node to the input of the Output node.

5. Connect Components:

   • Link the output of the Brand Visuals node to the input variable of the Prompt Builder node.
   • Link the Prompt Builder output to the Prompt input of the Image Generation node.
   • Link the Image output to the File Output node.

---

This results in:

Marketing visuals graph

🔗 Related Nodes: Input, Text, Image Generation, Output

---

Execution

To execute the workflow:

• Ensure the brand visuals and promotional context (e.g., new running shoes) are correctly filled in.
• Run the flow to generate a bold, high-energy marketing visual aligned with your sportswear brand.

Additional Information

• Authentication: Set up necessary API keys and permissions:

  • AWS Bedrock Authentication
  • OpenAI API Keys
  • Google AI Authentication

• Multimodal Inputs: Some models support image inputs. Check the model's documentation for supported formats.

• Parameter Tuning: Experiment with creativity and coherence values the models provide.

• Ethical Considerations: Be aware of potential biases in AI-generated content. Review outputs before use in sensitive applications.

• Stay Updated: Providers frequently release new models with improved capabilities.

For advanced usage, explore:

• Prompt engineering techniques

Troubleshooting

• If you encounter rate limits, check your API usage and consider upgrading your plan.
• For unexpected outputs, review your prompt and try adjusting the temperature or topP values.
• Ensure your AWS region is compatible with the chosen Bedrock model.