Create an Image (Advanced)#
Generate high-quality images using the advanced Flux model. This tool offers enhanced prompt coherence, superior aesthetics, and excellent text rendering capabilities with support for style transfer, shape guidance, and custom models.
Overview#
- Advanced text-to-image generation
- ControlNet shape guidance
- Style image transfer
- Custom model (LoRA) support
- Text rendering optimization
Quick Start#
- Visit Eden Create: Create an image (Advanced) tool
- Enter your descriptive prompt
- Adjust width and height settings
- Click "Create"
See API Reference section for API usage examples.
Ask an Eden agent to "Create an image (Advanced)".
Basic Usage#
Required parameters:
- Prompt
- Width (256-2048px)
- Height (256-2048px)
Optional features:
- Style image transfer
- Shape guidance via ControlNet
- Custom LoRA models
- Starting image
Technical Guidelines#
Resolution Options#
Best practice resolutions:
- 1536x640 (21:9)
- 1344x768 (16:9)
- 1216x832 (3:2)
- 1152x896 (4:3)
- 1024x1024 (1:1)
- 768x1344 (9:16)
- 640x1536 (9:21)
Higher resolutions possible in 64-pixel increments (e.g., 1920x1088, 1408x1408)
Advanced Features#
Style Transfer#
- Enable with
use_style_image
- Control influence with
style_strength
(0.5-0.9) - Optimal default: 0.75
Shape Guidance#
- Uses ControlNet preprocessors
- Options: edges, depth, lines, soft lines, human pose, scribble
- Strength adjustable via
controlnet_strength
Custom Models#
- Support for up to 2 LoRA models
- Independent strength controls
- Best for specific styles or subjects
Technical Details#
Base Model: Flux-dev (Black Forest Labs)
Default Parameters:
- Steps: 25
- Prompt Conditioning: 2.5
- Style Strength: 0.75
- ControlNet Strength: 0.6
API Reference#
Endpoint#
curl -X POST "https://api.eden.art/v2/tasks/create" \
-H "Content-Type: application/json" \
-H "X-Api-Key: YOUR_API_KEY" \
-d '{
"tool": "flux_dev",
"args": {
"prompt": "a man on the moon, planting an american flag on the surface, rocketship in the background",
"width": 1024,
"height": 1024,
"n_samples": 1,
"seed": 111
}
}'
Response Format#
{
"task": {
"_id": "task_12345",
"createdAt": "2024-01-29T21:52:07.171578+00:00",
"updatedAt": "2024-01-29T21:52:07.171608+00:00",
"user": "user_id",
"requester": "requester_id",
"tool": "flux_dev",
"parent_tool": null,
"output_type": "image",
"args": {
"prompt": "a man on the moon, planting an american flag on the surface, rocketship in the background",
"width": 1024,
"height": 1024,
"n_samples": 1,
"seed": 111
},
"mock": false,
"cost": 2,
"handler_id": "fc-xxxxx",
"status": "pending",
"error": null,
"result": null,
"performance": null
}
}
Common Issues and Solutions#
- Low Prompt Adherence: Increase prompt conditioning value (2.5-3.0)
- Style Transfer Too Strong: Reduce style_strength below 0.75
- LoRA Dominance: Lower lora_strength for better prompt balance
- Poor Shape Guidance: Try different preprocessor or increase controlnet_strength
Use Cases#
Creative Projects#
- High-quality artwork
- Concept development
- Style exploration
- Professional illustrations
Content Creation#
- Marketing materials
- Social media content
- Brand assets
- Visual storytelling
Professional Applications#
- Design work
- Advertising content
- Product visualization
- Creative direction
Tips for Best Results#
Prompt Engineering#
- Use clear, descriptive language
- Include style and composition details
- Specify quality modifiers
- Reference specific artists or styles
Parameter Optimization#
- Start with default settings
- Adjust prompt conditioning for adherence
- Fine-tune style strength
- Test different ControlNet preprocessors
Next Steps#
Once you have your generated image:
- Review Quality - Check prompt adherence
- Adjust Parameters - Fine-tune for better results
- Test Variations - Generate multiple samples
- Export and Use - Save in your preferred format
Ready to create advanced images? Visit the Advanced Image Generation tool to get started!