|
|
|
|
|
|
by gitfan86
1391 days ago
|
|
|
You could "hack it" by using a couple of other models as part of your pipeline. Similarly to how you have to use GAN after SD to "fix" faces sometimes. You also could put a language model on top of your prompting system. So "gandolff kicking ass" gets translated into " Page XXX, Paragraph XX from LOTR " |
|
|