Quoted By:
>stable diffusion model is like 3GB
>trained on 5 million images (probably more but assume for argument)
>assume for the sake of argument if every image was 500KB
>this is 2.5 terabytes of images
Can humanslop defenders explain their claims with evidence that AI is LITERALLY saving the images in a database and photobashing them together during generation? If you know a file compression algorithm that compresses to 0.01% then you would never have to work another day for the rest of your life, you would be so rich.
