 Hello everybody welcome back to another AI video in this one we're talking about drag GAN this is a game changing tool developed by the researchers from google the max plank institute of informatics it's a cool name and even mit and let me tell you not only is this a game changer i think this is going to end adobe's run at the top straight up i think photoshop is in a world of hurt and that the adobe executives got called in over the weekend to come in and talk about this because this is absolutely crazy and i know that's a big statement but look at the video behind me this is absolutely nuts all right so what does it mean drag gan stands for drag generative adversarial networks that is a mouthful essentially it's billed as a revolutionary image editing model that allows users to intuitively edit images by dragging points on them if you look at the video running behind me you can see how it does that with the red and the blue points and how you drag them apart drag them in etc one of the cool things about drag and is that it uses a pre-trained gan which again is a generative adversarial network whatever that means and it basically creates images that stay within the realm of realistic visuals so basically you can't go too far out of bounds it's not going to look like an 80s horror film and just get all crazy on you it basically means that edits will look natural and realistic even if they are significantly different from the original images yeah this is crazy stuff guys apparently the two components are trained together in a process called adversarial training which kind of sounds like every relationship i've ever been in but whatever the generator basically tries to create images that are so realistic that the discriminator discrimination is bad except in gans apparently but the discriminator can't tell them apart from real images and as they're trained both become better at their tasks until the generator can create images the discriminator can't distinguish from real ones wow what a mouthful again i don't pretend to know the technical details that go into that the link to the full technical paper is in the description below apparently next once the once they're trained the generator can be used to edit images by dragging points on them that's very clear you can see that in the videos when a user drags a point on an image the generator uses it to create a new image similar to the original but with the desired edits i mean all in all guys this is just unbelievable like i could keep talking about stuff but just watch the video look at what this can do this seriously could be the end of adobe i mean i don't want to you know presume that they're in trouble but with these types of tools and you see tools like clip champ for video editing and davinci resolve photopea for image you know for image editing and a variety of other tools you know combine that with chat gbt and all these other you know mid journeys and dollies and stable diffusions adobe's in a lot of trouble please let me know what you think in the comments below thanks for watching be back soon