Project Highlights
- Upload a real-world image and transform it into a stylized Where’s Waldo? puzzle
- Multi-stage pipeline combining segmentation, depth estimation, stylization, placement, and blending
- Layer-aware Waldo placement with occlusion reasoning and visibility constraints
- Diffusion-based illustration generation using ControlNet and custom Waldo-style checkpoints
- Interactive web demo with playable Waldo search puzzles
- Timed gameplay mode with multiple generated scenes and replay support