This reminds me of Stand-up Maths and Steve Mould creating a puzzle that has two solutions with two different images: https://www.youtube.com/watch?v=b5nElEbbnfU
If this interests you, you might also like the below:
https://diffusionillusions.com
https://dangeng.github.io/visual_anagrams/
There are also videos with the process explained
It looks like this could be used for some form of steganography. You need the right mirror form for decoding.
I love how many optical illusions have been revisited with new generative techniques
Happy to see that Disney is still as active as always in research
Oh great. Now I suddenly feel the urge to acquire a physical oddly shaped lens or mirror.
Disney invited me to talk about my GenAI startup and research in front of a bunch of their execs across ABC, ESPN, Pixar, Streaming, etc. All of their folks were super nice and gracious to our small startup except for one.
Steve May basically scoffed at how little my small team could accomplish. Mind you we were using mocapped skeletal animation and object animation curves to fully steer video diffusion over a year and a half ago. Before image to video modalities. He picked apart our training and engineering and gloated that they could do better.
The incident is seared into my brain.
I can't help but think of Disney as the Empire and Pixar as the Death Star.
Don’t get me wrong, this is very neat. But why?
[flagged]
This reminds me a bit of a project I am currently doing where I swap adjacent pixels in the last image of short video sequences until it resembles the first image:
https://www.instagram.com/marekgibney/reel/DILksFYNSkE/