I agree with you, but also the following is "just" an implementation detail: > only context it has is its prompt The only real context is its latent represent

I agree with you, but also the following is "just" an implementation detail: > only context it has is its prompt The only real context is its latent representation of the prompt, there's nothing fundamentally limiting visual, auditory, symbolic, and mixed prompts as long as they map to a common latent space and the generator is trained on it.