Comment by dzink

From the beginning of using Stable Diffusion in local and cloud instances, I’ve been promoting SD to generate objects I know nobody has ever drawn before. “Airplane by Tesla”, “Taylor Swift flying in the clouds”, “Little girl riding on an ira descent unicorn and chasing butterflies in the clouds”, “Turkey as a Judge” etc. I highly encourage everyone to try doing that. The results are absolutely atrocious in the beginning and it takes many many runs short and long, with seeds guiding the model to get closer and closer to what I ask. It took a long time to get one instance of SD to make the invention look plausible, and then trying on a new model copy/instance takes the results back to crap. That makes me suspect your guidance trains the instance you are using and your prompts and feedback create the work substantially. The model truly generates novel content based on input of the generator and it takes effort to replicate a work it was trained on, likely by using multiple keywords the original was captioned with in many places. So you can get replicas if you try, but you can also draw replicas with brushes if you have enough skill as well. To reiterate: try to generate content you know nobody has ever drawn before (and google to verify it is truly original) and see how much effort it takes to get an actually good result. Now sub-trained models can be steered in different directions, so it’s possible that Midjourney or a heavily sub-trained cloud instance overfit to the originals, so this is not universal, but every copy of the model is likely different and molded by the prompts and feedback it’s been given.