Diffusion Illusions: Hiding Images in Plain Sight
The "Diffusion Illusions" project uses diffusion models to create four types of optical illusions, won an award at CVPR 2023, and will be presented at SIGGRAPH 2024, encouraging user interaction and experimentation.
Read original article
The project "Diffusion Illusions" explores the creation of optical illusions and steganography using diffusion models. It features four main types of illusions: Flip Illusions, which change appearance when flipped; Rotation Overlay Illusions, where rotating an image reveals different pictures; Folding Tile Illusions, which involve mechanical linkages of tiles that can be manipulated; and Hidden Overlay Illusions, where overlapping images reveal a hidden fifth image. The project has gained recognition, winning the best demo award at CVPR 2023 and being accepted for presentation at SIGGRAPH 2024. Users can interact with the illusions through a website that allows them to manipulate images and see the effects in real-time. The project also includes a source code repository for those interested in the technical aspects and implementation of these illusions. The authors encourage users to experiment with the illusions both digitally and in real life, providing resources for printing and further exploration.
- "Diffusion Illusions" project utilizes diffusion models for creating optical illusions.
- Four types of illusions are featured: Flip, Rotation Overlay, Folding Tile, and Hidden Overlay.
- The project won the best demo award at CVPR 2023 and is set to be presented at SIGGRAPH 2024.
- Users can interact with the illusions online and access source code for further exploration.
- The project encourages real-life experimentation with printed images.
Related
Show HN: I built a large JavaScript powered flipdisc display. Here's a guide
Flipdisc displays, or flip dots, use electromagnetic pulses to switch colors. A project details building a large interactive display for offices, covering construction, power, software, and design considerations. It aims to explore real-time visualizations and user interactions, hoping to make flipdisc technology more accessible.
VFusion3D: Learning Scalable 3D Generative Models from Video Diffusion Models
VFusion3D is a project developing scalable 3D generative models using video diffusion, to be presented at ECCV 2024. It offers pretrained models and a Gradio application for user interaction.
DifuzCam: Replacing Camera Lens with a Mask and a Diffusion Model
The paper introduces DifuzCam, a lensless camera design using a diffusion model for image reconstruction, enhancing quality and fidelity, with potential applications in various imaging systems.
Show HN: Typescript library for editing videos with code
Diffusion Studio is a browser-based video processing framework for AI, featuring video/audio editing, layering, and real-time playback. It supports various media formats and requires a license for commercial use.
Tutorial on Diffusion Models for Imaging and Vision
The paper "Tutorial on Diffusion Models for Imaging and Vision" by Stanley H. Chan discusses diffusion models' role in generative tools, targeting students and researchers in machine learning and computer vision.
3 commentsRelated
Show HN: I built a large JavaScript powered flipdisc display. Here's a guide
Flipdisc displays, or flip dots, use electromagnetic pulses to switch colors. A project details building a large interactive display for offices, covering construction, power, software, and design considerations. It aims to explore real-time visualizations and user interactions, hoping to make flipdisc technology more accessible.
VFusion3D: Learning Scalable 3D Generative Models from Video Diffusion Models
VFusion3D is a project developing scalable 3D generative models using video diffusion, to be presented at ECCV 2024. It offers pretrained models and a Gradio application for user interaction.
DifuzCam: Replacing Camera Lens with a Mask and a Diffusion Model
The paper introduces DifuzCam, a lensless camera design using a diffusion model for image reconstruction, enhancing quality and fidelity, with potential applications in various imaging systems.
Show HN: Typescript library for editing videos with code
Diffusion Studio is a browser-based video processing framework for AI, featuring video/audio editing, layering, and real-time playback. It supports various media formats and requires a license for commercial use.
Tutorial on Diffusion Models for Imaging and Vision
The paper "Tutorial on Diffusion Models for Imaging and Vision" by Stanley H. Chan discusses diffusion models' role in generative tools, targeting students and researchers in machine learning and computer vision.