The fact that GF can't create wuality images from cero, like MJ or Stable Diffusion is on purpouse..
Or a limitation of the dataset the model was trained with?
It's good at removing objects/replacing/extend.
But when it comes to create from scratch something, it doesn't even come close to the above mentioned.
Let's not talk about DALL.E 3...
You guys have plans to make GF better at creating things from scratch and for the love of god, that faces!!!


