- 1 Post
- 10 Comments
11·7 个月前Donald trump works… But not Natalie Portman. Random examples.
It’s curious. Some people work, like Taylor Swift, and others don’t, like Natalie Portman. Is it a matter of training the database? Do we just have to keep training it? Or will we always miss certain personalities?
ok but I don’t understand, according to the following link, it is explained that prettyAi already uses several models: “As for now, we know that Perchance is using different models depending on the prompt of the user (specifically ‘Furry’, ‘Photorealistic’, and ‘Anime’ images)” https://rentry.org/perchance-ai-faq#%3A~%3Atext=*+The+`text%2Cspecifically+‘Furry’%2C+'Photorealistic So, what to understand?
Same thing. I feel like the new model is trying to produce images of a certain style, with a predefined lighting style, and also predefined faces. It’s hard to really have control over it and make it produce what you want. In any case, more difficult than with the old one. Maybe it needs more training. Yes, but then, how can we help it manage its “fine tuning”?
Wouldn’t it be possible to simply have a button that would allow you to switch from one model to another? From the new to the old?
Wouldn’t it be possible to simply have a button that would allow you to switch from one model to another? From the new to the old?
I agree! An “image input” option would really be a plus for this generator!
ok thanks