Skip to content

Image Generation (Text-to-Image)

Overview

The Image Generation feature enables AI-powered creation of images directly from text prompts within Chat conversations and Agents. When using an LLM model with image generation capabilities (such as GPT-5.1, DALL-E 3, or other vision-enabled models) and the Image Generator internal tool is enabled, you can generate high-quality images through natural language descriptions.

Generated images appear inline as thumbnails in the conversation and can be viewed full-size or downloaded. Images are automatically saved to a dedicated folder within the attachment Artifact bucket for easy access and management.

Prerequisites

To use image generation, you need:

  • An LLM model that supports image generation (e.g., GPT-5.2, GPT-5.1, DALL-E 3, DALL-E 2)
  • The Image creation internal tool enabled in your conversation or agent
  • An image generation model configured and selected in your project (see AI Configuration → Image Generation)

  • Proper project configuration by your administrator

    Model

Not Seeing Image Creation?

If the Image creation option is not available, contact your administrator to enable image generation for your project.

Accessing Image Generation

Image generation is available in two primary locations within ELITEA:

From Chat Conversations

Enable image generation for ad-hoc conversations.

  1. Navigate to your conversation.
  2. Locate the chat input toolbar at the bottom of the screen.
  3. Click the Internal Tools icon (value icon) next to the attachment button.
  4. In the popup, find Image creation in the list.
  5. Toggle the Image creation switch to enable it.
  6. A success notification appears: "Internal tools configuration updated".
  7. Click anywhere outside the popup to close it.

  8. Use the model selector to choose a model with image generation capabilities (e.g., GPT-5.2, GPT-5.1, DALL-E 3).

    Chat Image Creation Configuration

What appears in the Internal Tools popup

The Image creation option is only visible in the popup when an image generation provider toolkit is configured for your project. If no such toolkit is available, the option is silently omitted from the list — and if no internal tools are available at all, the Internal Tools icon will not appear in the chat toolbar. Contact your administrator if you cannot see the option.

In Agent Configuration

Configure image generation as a persistent capability for an agent.

  1. Navigate to Agents in the main menu.
  2. Select the agent you want to configure or create a new agent.
  3. On the Configuration tab, scroll down to the TOOLKITS section
  4. Within the TOOLKITS section, locate the INTERNAL TOOLS subsection
  5. Find the Image creation toggle. If it is not immediately visible, click Show all to expand the full list of internal tools.
  6. Toggle the Image creation switch to enable it.
  7. Use the model selector at the top of the configuration page to choose a model with image generation capabilities (e.g., GPT-5.2, GPT-5.1, DALL-E 3).

  8. Click Save at the top of the configuration page to persist the change.

    Agent Image Creation Configuration

Consistent Image Generation Across Conversations

Enabling Image creation at the agent level means every new conversation that uses this agent has image generation available by default, without requiring users to enable it manually each time.

How to Generate Images

Once enabled, generating images is as simple as asking the AI assistant:

  1. In the message input field, type a prompt describing the image you want to create
  2. Example: "Generate an image of a sunset over mountains"
  3. Press Send or hit Enter
  4. The AI will automatically invoke the image generation tool

  5. Generated image(s) will appear inline in the conversation

    Generate image

Technical Limitations

  • Image Count: Maximum 10 images per generation request
  • Model Dependency: Available features depend on the specific image generation model
  • Quality vs Speed: Higher quality images take longer to generate
  • Size Constraints: Maximum image size varies by model configuration
Basic Image Generation

Example Prompts:

  • "Generate an image of a sunset over mountains"
  • "Create a diagram showing a microservices architecture"
  • "Generate an illustration of a friendly robot assistant"
  • "Make an image of a modern office workspace"

The AI will automatically invoke the image generation tool and return the generated image(s) inline in the conversation.

Advanced Image Generation

You can customize images by including details in your prompt:

Number of Images:

  • "Generate 3 variations of a logo design"
  • You can request up to 10 images per generation

Image Quality:

  • "Generate a high-quality image of a product mockup"
  • Options: auto (default), low, medium, high
  • Quality setting depends on the underlying model capabilities

Image Size:

  • "Create a 1024x1024 image of a landscape"
  • Default: auto (model-dependent)
  • Common sizes: 256x256, 512x512, 1024x1024, 1024x1792, etc.
  • Available sizes depend on the specific image generation model

Style:

  • "Generate an image in a watercolor style"
  • Describe the artistic style you want
Multi-Step Generation

You can refine images through conversation:

  1. Initial Generation: "Generate an image of a modern kitchen"
  2. Review: View the generated thumbnail
  3. Refinement: "Make it brighter and add more natural light"
  4. Iteration: Continue refining until satisfied

Viewing and Managing Generated Images

Generated images appear as thumbnails in the conversation. Hover over any image to access management options:

Fullscreen View:

  • Click the thumbnail to open fullscreen modal

  • Press Escape or click outside to close

    Fuul size image

Download:

  • Click the Download icon to save the image locally
  • Images are also accessible from Artifacts → attachment bucket, inside the folder

  • You can download images from the Artifacts page at any time

    Download image

Access from Artifacts:

  • Navigate to Artifacts in the main menu
  • Open the attachment bucket
  • Navigate to the folder inside the bucket
  • All generated images are stored here with timestamps

  • Download, delete, or manage images from this centralized location

    Artifact

Delete:

  • Hover over the generated image thumbnail in the conversation to reveal the action buttons overlay
  • Click the Delete icon (trash icon) that appears on hover
  • A confirmation dialog opens showing the image filename
  • The dialog includes an "Also delete from attachment storage" checkbox (unchecked by default)
    • Leave it unchecked to remove the image only from the conversation view
    • Check it to also delete the file from the attachment bucket in Artifacts

  • Click Delete to confirm

    Fuul size image

Storage and Deletion

Generated images are automatically saved in the Artifact Toolkit in the attachment bucket, inside a dedicated images folder. This folder is created automatically the first time you generate an image if it doesn't already exist.

Usage Scenarios

Creating Technical Diagrams

Use Case: Generate architecture diagrams, flowcharts, or system designs

Example Prompts:

  • "Create a diagram showing a three-tier web application architecture with load balancer, web servers, and database"
  • "Generate a flowchart for a user authentication process"
  • "Make a network diagram showing VPN connection between offices"

Best Practices:

  • Be specific about components and relationships
  • Mention diagram type (flowchart, sequence diagram, architecture diagram)
  • Specify colors or styles if needed
  • Request multiple variations to choose the best one
Generating Illustrations and Concept Art

Use Case: Create visual content for presentations, documentation, or creative projects

Example Prompts:

  • "Generate an illustration of a team collaborating in a modern office"
  • "Create a hero image for a SaaS landing page showing productivity features"
  • "Make an icon set for a project management application"

Best Practices:

  • Describe the mood and style (modern, minimalist, vibrant, professional)
  • Specify color schemes when important
  • Include context about the intended use
  • Generate multiple options for comparison
Design Mockups and Prototypes

Use Case: Quickly visualize UI/UX concepts or product designs

Example Prompts:

  • "Generate a mockup of a mobile app dashboard for fitness tracking"
  • "Create a landing page design for an AI-powered writing assistant"
  • "Make a product card design for an e-commerce website"

Best Practices:

  • Specify device type (mobile, tablet, desktop)
  • Mention key UI elements to include
  • Describe the color palette and brand style
  • Request specific layout patterns if needed
Data Visualization

Use Case: Create charts, graphs, or infographics from described data

Example Prompts:

  • "Generate a bar chart comparing quarterly sales for 2024"
  • "Create an infographic showing the software development lifecycle"
  • "Make a pie chart showing market share distribution"

Best Practices:

  • Clearly describe the data and labels
  • Specify chart type (bar, line, pie, scatter)
  • Mention color preferences for different data series
  • Include title and axis labels in your description
Educational and Training Content

Use Case: Generate visual aids for learning materials

Example Prompts:

  • "Create an illustration explaining how photosynthesis works"
  • "Generate a diagram showing the layers of the OSI network model"
  • "Make an image demonstrating proper ergonomic sitting posture"

Best Practices:

  • Focus on clarity and simplicity
  • Request labels or annotations
  • Use clear, step-by-step descriptions
  • Consider your audience's technical level
Marketing and Social Media Content

Use Case: Create branded visuals for marketing campaigns

Example Prompts:

  • "Generate a social media post image announcing a product launch"
  • "Create a banner image for an email newsletter about new features"
  • "Make a promotional image for a webinar on AI automation"

Best Practices:

  • Specify dimensions for the target platform
  • Include brand colors and style guidelines
  • Mention text placement areas if needed
  • Generate variations for A/B testing

Best Practices

Crafting Effective Prompts

Be Specific and Descriptive

Instead of: "Generate a logo"

Try: "Generate a modern, minimalist logo for a cloud storage company using blue and white colors, featuring a cloud icon"

Why: Specific prompts yield more accurate results that match your vision.

Include Context and Purpose

Instead of: "Make an office image"

Try: "Create an image of a modern open-plan office with collaborative workspaces for a tech startup presentation"

Why: Context helps the AI understand the intended use and style.

Specify Style and Mood

Instead of: "Generate a landscape"

Try: "Generate a serene mountain landscape at sunset with warm orange and purple tones in a photorealistic style"

Why: Style descriptions guide the aesthetic direction.

Request Multiple Variations

Instead of: Single generation and hoping for the best

Try: "Generate 3 different versions of a dashboard UI, each with a different color scheme"

Why: Multiple variations provide options to choose from.

Managing Generated Content

Organize Your Images
  • Download images promptly with descriptive filenames
  • Use browser download management or dedicated folders
  • Consider version numbers for iterative designs
  • Keep related images together for project work
Iterate and Refine
  • Review initial generation critically
  • Provide feedback for refinement ("make it brighter", "add more detail")
  • Build on successful elements from previous generations
  • Save intermediate versions you might want to reference
Document Your Process
  • Keep a record of effective prompts
  • Note which models work best for specific types of images
  • Document style preferences that yield consistent results
  • Share successful prompt patterns with your team

Performance Optimization

Model Selection
  • Use faster models for quick iterations
  • Switch to high-quality models for final outputs
  • Consider model capabilities for specific image types
  • Monitor generation times and adjust accordingly
Batch Requests Wisely
  • Generate multiple variations in one request when possible
  • Avoid requesting too many images at once (max 10)
  • Balance quantity with quality for better results

Troubleshooting

Issue: Image Generation Not Available

Possible Causes:

  • Image creation internal tool not enabled
  • No image generation provider toolkit configured for the project — the Image creation option is filtered out of the popup entirely when this toolkit is absent
  • If no internal tools are available at all, the Internal Tools icon is hidden from the chat toolbar
  • Network connectivity issues

Solutions:

  1. Enable the Image creation toggle in Internal Tools settings
  2. If the Image creation option is not visible in the popup, your project has no image generation provider toolkit configured — ask your administrator to add an image generation integration
  3. If the Internal Tools icon is missing from the chat toolbar entirely, contact your administrator to verify toolkit configuration
  4. Refresh the page and try again
Issue: Image Generation Model is Not Configured

Error Message: "Image generation internal tool requested but no image generation model configured"

Cause: The internal tool is enabled, but the selected model lacks image generation capabilities.

Solution:

  1. Select a different model with image generation support
  2. Verify model configuration with your administrator
  3. Check integration settings for the image generation endpoint
Issue: Generated Images Not Displaying

Possible Causes:

  • Image data format error
  • Browser compatibility issues
  • Base64 encoding problems
  • Network interruption during generation

Solutions:

  1. Refresh the page and regenerate the image
  2. Try a different browser (Chrome, Firefox, Edge)
  3. Check browser console for errors
  4. Ensure stable network connection
Issue: Low-Quality Image Results

Possible Causes:

  • Vague or ambiguous prompts
  • Quality parameter set too low
  • Model limitations
  • Size constraints

Solutions:

  1. Refine your prompt: Add more specific details and context
  2. Request higher quality: Explicitly ask for "high-quality" images
  3. Try different models: Some models excel at specific image types
  4. Specify larger sizes: Request specific dimensions (e.g., 1024x1024)
Issue: Download Fails

Possible Causes:

  • Browser blocking downloads
  • Base64 conversion error
  • Network timeout
  • Insufficient browser storage

Solutions:

  1. Check browser download permissions
  2. Try right-clicking the image and "Save Image As"
  3. Refresh and try download again
  4. Clear browser cache and retry
Issue: Image Generation Takes Too Long

Possible Causes:

  • Complex prompts requiring more processing
  • High-quality/large size settings
  • Server load or API rate limits
  • Network latency

Solutions:

  1. Simplify your prompt for faster generation
  2. Use "auto" quality setting initially
  3. Generate one image at a time instead of multiple
  4. Wait for current operations to complete
  5. Try during off-peak hours if rate limited

Additional Resources