Create illusions conditioned on image.
Hint: Drag and drop image files from your computer, images from web pages, paste from clipboard (Ctrl/Cmd+V), or provide a URL. Accepted file types: jpg, jpeg, png, webp, gif, avif, heic, heif
Type # to reference inputs.
Customize your input with more control.
Your request will cost $0 per compute second.
Your request will cost $0 per compute second.
Illusion Diffusion extends Stable Diffusion with ControlNet conditioning to create optical illusion images from pattern inputs and text prompts. Trading straightforward text-to-image generation for pattern-aware composition, this model embeds your prompts into geometric patterns, creating images that reveal different interpretations based on viewing distance or angle. Built for designers and artists who need controllable visual illusions without manual composition work.
Use Cases: Optical Illusion Art | Pattern-Based Design | Hidden Image Marketing | Visual Puzzle Creation
Illusion Diffusion runs on fal's infrastructure with configurable inference steps (up to 80) and multiple scheduler options for quality-speed tradeoffs.
| Metric | Result | Context |
|---|---|---|
| Inference Steps | 40 (default) | Configurable 0-80 range for quality control |
| Output Resolution | Up to 1024x1024 | Square HD default, custom multiples of 8 supported |
| ControlNet Scale | 0-1 adjustable | Controls pattern influence strength |
Illusion Diffusion extends Stable Diffusion with ControlNet conditioning, allowing you to input both a base pattern image and a text prompt. The model interprets your prompt through the geometric constraints of the pattern, creating outputs where the pattern remains visible while the prompted content emerges at different viewing distances.
What this means for you:
Dual-input control: Provide a pattern image URL alongside your text prompt, the model blends both inputs rather than choosing between them
Adjustable pattern strength: ControlNet conditioning scale (0-1) lets you dial pattern visibility up or down without regenerating from scratch
Guidance timing control: Set when ControlNet influence starts and ends during generation (control_guidance_start/end parameters) for precise pattern integration
Multiple scheduler options: Choose between DPM++ Karras SDE and Euler samplers to optimize for your specific pattern complexity and prompt requirements
| Spec | Details |
|---|---|
| Architecture | Stable Diffusion with ControlNet |
| Input Formats | Pattern image URL (JPEG, PNG, WebP, GIF, AVIF) + text prompt |
| Output Formats | PNG images with embedded pattern conditioning |
| Resolution Options | Square HD default (1024x1024), custom dimensions as multiples of 8 |
| License | Commercial use permitted |
API Documentation | Quickstart Guide | Enterprise Pricing
AuraFlow Text to Image – Illusion Diffusion specializes in pattern-conditioned generation for optical illusion workflows. AuraFlow focuses on straightforward text-to-image generation with high prompt adherence, ideal for standard image generation without pattern constraints.