Github
This website is fully interactive playground! In this project, we generate optical illusions and perform stenography using off-the-shelf diffusion models - all of which work in the physical world too!
Diffusion Illusions won best demo award in CVPR 2023!
We also presented at OpenSauce 2023. See JacksFilms' reaction here
Source code: github.com/RyannDaGreat/Diffusion-Illusions. If you like it, please give it a star!
Our Paper: diffusionillusions.pdf - please give it a read!
More Details: Check the bottom of the page to learn more about this project!
If you wait a few seconds, the below images will flip upside-down - to reveal a new image!
You can
Try it on Colab!In this section, we have examples with two images. As the one on the top is rotated, the image changes! The one on the top is multiplied by the one on the bottom (pixelwise), simulating how light would be filtered through two transparent sheets with images printed on them. Effectively, we've made 4 images out of 2!
This works in real life too: you can overlay printouts of these images on top of each other for the same results, as seen in the youtube video.
You can
Try it on Colab!In this section, we have sets of 4 images. You can
This works in real life too: you can overlay printouts of these images on top of each other for the same results, as seen in the youtube video.
I won't spoil it for you in this paragraph; try it out!
Try it on Colab!Click the this link to learn much more about our project!
The project is comprised of three demonstrations:
Another project with a similar goal was released in February 2023: Illusion-Diffusion, authored by Matthew Tanick. That project only achieves the first part of this demo, aka image flipping.
Our other paper, "Peekaboo: Text to Image Diffusion Models are Zero Shot Segmentors", has an extremely similar formulation to the illusions on this website - it optimizes images using a variant of score distillation loss to perform zero shot segmentation.
A paper released after our project in December 2024 by Daniel Geng et al: Visual Anagrams has beautiful results and is worth checking out
To help explain how these were calculated, I've included some easy-to-read pseudocode that covers how each of these illusions were made. This code has been oversimplified for the sake of simplicity - for all the details please read our paper. The full source code can be found at github.com/RyannDaGreat/Diffusion-Illusions, or at any of the Google Colab links throughout this webpage.
Diffusion illisions are generated through an iterative process: we optimize the output images with gradient descent, starting from pure noise. Here are some timelapses showing how they evolve over time.
For reference, here are the end results of the above timelapse video:
Paper
Youtube
