Collection of GPT-image-generator 2.0 issues, bugs, and work-around tips (check first post)

> The problem now is: > I can no longer see the old internal prompts that were actually sent to the image generator. > > So it is possible that newer updates no longer interpret those internally expanded style concepts in the same way and perhaps now overreact to them instead — exaggerating details rather than improving them. I think your use case is very cool - but only in the short-term. Here’s why: In the long term: (1) As far as an imageGen model goes, I don’t think that additions to the training is possible. And if so, each new imageGen version will have new training data. (2) With respect to ChatGPT, each new version changes things that we are not aware of. More importantly, chat history is not infinite. No one can use ChatGPT to discover the hidden workings (secrets) of how things work. Otherwise, competitors will use it to their advantage. And there would be security concerns.