Face Looker — Turn Any Real Face Photo Into a Cursor-Tracking Character
Complete Tutorial with Free Resources
Learn how to create an interactive gaze tracking demonstration that creates an engaging visual experience where a face follows your cursor movements in real-time.
Video Preview
Tutorial Preview

Complete Setup Guide
Step-by-step walkthrough from image generation to interactive cursor tracking
What You'll Learn
- ✓Generate 121 gaze direction images using Replicate
- ✓Set up the interactive demo with JavaScript
- ✓Implement coordinate mapping and normalization
- ✓Create smooth cursor tracking system
- ✓Customize the UI with modern design
- ✓Deploy your own face tracking character
Download Resources
Face Looker Resources
Everything you need for interactive gaze tracking setup
Face Tracking Demo
- ✓Complete demo.html file
- ✓JavaScript gaze tracking code
- ✓Image generation guide
- ✓Replicate API integration
- ✓Customization instructions
Extra Resources
- ✓Sample face image sets
- ✓Python API examples
- ✓cURL API examples
- ✓UI customization tips
- ✓Mobile touch support guide
Tutorial Overview
Interactive Gaze Tracking Setup Guide
This comprehensive tutorial covers how to create an interactive gaze tracking demonstration where a face follows your cursor movements in real-time, using pre-rendered images and JavaScript-based tracking.
Introduction
Face Looker is an interactive gaze tracking demonstration that creates an engaging visual experience. As you move your cursor around the screen, the face follows your movements in real-time, creating an illusion of eye contact and engagement.
This is achieved through a clever combination of pre-rendered images and JavaScript-based gaze tracking.
How It Works
Face Looker uses a grid-based gaze tracking system that works in three main steps:
- Image Generation: A face image is processed through an AI model to generate 121 different gaze directions, creating a grid of images where the eyes look in different directions.
- Coordinate Mapping: When you move your cursor, the system calculates the relative position of your cursor to the face container. This position is normalized to coordinates between -1 and 1.
- Image Selection: The normalized coordinates are quantized to match the pre-generated image grid (ranging from -15 to 15 in steps of 3), and the corresponding face image with the matching gaze direction is displayed.
Technical Details
- Grid Configuration: 11×11 grid (121 images total)
- Gaze Range: -15 to +15 degrees in both X and Y axes
- Step Size: 3 degrees between each image
- Image Format: WebP format for optimal performance
- Naming Convention: gaze_px{value}_py{value}_256.webp
The system uses JavaScript to track mouse/touch movements, calculate normalized coordinates, quantize to the nearest grid point, and dynamically swap images for smooth gaze following.
Installation Guide - Option 1: Web Interface (Easiest)
This is the simplest way to get started with Face Looker. Follow these steps:
Step 1: Generate Your Face Images
- Navigate to the Replicate web interface: https://replicate.com/kylan02/face-looker
- Prepare your image: You'll need a 512×512 pixel face image. Make sure the face is clearly visible and centered.
- Upload your image: Click the upload button and select your prepared face image.
- Generate: The AI model will process your image and generate 121 different gaze direction images.
- Download: Once processing is complete, download the generated ZIP file.
Step 2: Extract and Organize Files
- Extract the ZIP file to a location of your choice.
- Locate the
faces/folder inside the extracted contents. - Place the folder in your project directory alongside
demo.html.
Step 3: Configure the Demo
- Open
demo.htmlin a text editor. - Find line 95 (or search for
BASE_PATH). - Update the path to point to your extracted faces folder:
const BASE_PATH = 'my_faces/'; - Save the file.
Step 4: Run the Demo
- Open
demo.htmlin your web browser (simply double-click the file or use "Open with" your preferred browser). - Move your cursor around the screen and watch the face follow your movements!
That's it! No server setup, no complex installation - just open and enjoy.
Installation Guide - Option 2: API (Python)
- Install the Replicate Python client:
pip install replicate - Get your API token from replicate.com/account/api-tokens
- Set the token:
# Windows PowerShell $env:REPLICATE_API_TOKEN="your_token_here" # Linux/Mac export REPLICATE_API_TOKEN=your_token_here
- Run Python script:
import replicate output = replicate.run( "kylan02/face-looker:latest", input={"image": open("your_face.jpg", "rb")} ) print(output) # Download URL for the ZIP file
Installation Guide - Option 3: API (cURL)
curl -X POST https://api.replicate.com/v1/models/kylan02/face-looker/predictions \
-H "Authorization: Token $REPLICATE_API_TOKEN" \
-H "Content-Type: application/json" \
-d '{"input": {"image": "https://your-image-url.com/face.jpg"}}'Final Result
The final implementation features:
- Beautiful UI: Modern design with a dotted black and white background theme
- 3D Card Effect: Interactive text section that responds to mouse movements with a 3D tilt effect
- Responsive Layout: Two-section design with the face on the right and information on the left, separated by a clean divider line
- Smooth Performance: Optimized image swapping for seamless gaze tracking
- Cross-Platform: Works on desktop and mobile devices with touch support
The demo creates an engaging, interactive experience that demonstrates the power of AI-generated gaze tracking in a simple, accessible web interface.
Use Cases
Face Looker has numerous practical and creative applications:
- Interactive Websites & Portfolios: Create engaging "about me" sections where your face follows visitors
- Educational Demonstrations: Teach concepts of coordinate mapping and normalization
- Marketing & Advertising: Create attention-grabbing banner ads and interactive product showcases
- Entertainment & Gaming: Interactive character displays and game UI elements
- Accessibility Tools: Eye-tracking demonstrations and assistive technology prototypes
- Art & Creative Projects: Digital art installations and interactive exhibits
- Customer Service: Virtual assistants with engaging presence and chatbot interfaces
Conclusion
Face Looker demonstrates how modern AI and web technologies can create engaging, interactive experiences with minimal setup. The combination of pre-rendered images and real-time JavaScript creates a smooth, responsive gaze tracking system that's both impressive and accessible.
Whether you're building a personal portfolio, creating educational content, or developing interactive marketing materials, Face Looker provides a simple yet powerful way to add engaging visual interactions to your projects.