3 ms·
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 als
by gitfan86 4y 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 "