That was my idea too and it looks like I imagined it to 😂

They also created ghidra! Probably second best

To train a diffusion model that only outputs one image with difference is I think not possible you could do an image to image and then fix the seed so you would get a consistent result and then picking the nearest result that is nearly an identical copy

Sir, you just made my day thank you!

“Eigentlich fertig” was for an IP subnet calculator that I programmed with a fellow student