I have watched and held my breath. I only speak up because there is a lot of misinformation here. JBC you have no idea of which you speak. You have no idea how a diffusion model is produced or deployed. The waters get muddied by disinformation. Portraying yourself as knowledgeable in this field is laughable at best and quite a dis-service to those of us who are data scientists and imaginative artists.

"A good AI collects Information then it builds up a so called model, you can compare it with a mathematical function of your brain, but they call that file a Model witch is one of many learning algorithm's comparable to a .Json file Not a Image it self, the learning process with given Information generates a sort of Intelligent Filter, just like you use in Photoshop just that it is a learning filter. You can also compare it with a encyclopedia of Information."

What kind of BS is this? Really? Would you really like to know how a diffusion model is built? Not your way. An "A.I." doesn't "collect information" ;and exactly how you equate this to "many learning algorithm's (not a possessive btw) comparable to a .Json file." makes no sense what so ever.

We've had this conversation before. More than once. It has been shut down for various reasons each time. Usually the misinformation and fear of the future arguments win the day. I understand that. There are unanswered in the court of law arguments currently underway. They will fall on both sides of the fight as there is more than one way to build a diffusion argument and many companies are building their own using GANs and not the addition/subtraction of noise method currently in use by Stable Diffusion and others.

So stop with the BS and try reading a white paper or two. I have linked to them in the past and got in sh!t over it so I will not do so again. There are reasons that Ken and others have stepped out of these conversations and folk with knowledge like myself fear to tread.