I have been playing with Stability Diffusion for the past week. (It’s a bit addictive.) It’s currently very time consuming to make photo realistic deep fake images. It probably easier to do so with photo shop. What I can see happens is that people will use Stable Diffusion to make a lot of creative images for political messages intend to attack and mock the opposite side instead of trying to mislead.
Thanks or the comment. I think that simple interfaces for SD like this are not particularly worrysome. But I think that now (1) inpainting/outpainting, (2) dreambooth (see this SFW example), (3) GUIs that make it easy to use these, and (4) future advancements in difusion models (remember that DLALE-2 was only released in April of this year) are the main causes for concern.
I have been playing with Stability Diffusion for the past week. (It’s a bit addictive.) It’s currently very time consuming to make photo realistic deep fake images. It probably easier to do so with photo shop. What I can see happens is that people will use Stable Diffusion to make a lot of creative images for political messages intend to attack and mock the opposite side instead of trying to mislead.
Thanks or the comment. I think that simple interfaces for SD like this are not particularly worrysome. But I think that now (1) inpainting/outpainting, (2) dreambooth (see this SFW example), (3) GUIs that make it easy to use these, and (4) future advancements in difusion models (remember that DLALE-2 was only released in April of this year) are the main causes for concern.