Prompt hints: format something like this: "A character turnaround of a (X) wearing (Y). " In some of the prompts the token is spelled differently: due to some technical issues with xformers, I trained this badboy about 40 times, and some of the examples are from those trainings. All the same dataset, I promise. Just change any weird token names to CharTurnerV2 if you use them.
Token - "CharTurnerV2" place at the front for stronger effect, place at the end for a weaker effect. Token weighting works too
Add:
"Multiple views of the same character in the same outfit"
Add original CharTurner to mix back in some anime/turns
Add charTurner Lora (uploading soon) on top (at very low strength, like .3) to really force the turn. Will affect style.
Next Version: (after the v2 lora and 2.1 versions): Looking for more variety for the data set! Post a review or comment with your best turn around (especially if it was a hard one to get it to do!) and I will use it in the V3 dataset.
2.1 and LoRa versions coming soon.
Prompt hints: format something like this: "A character turnaround of a (X) wearing (Y). " In some of the prompts the token is spelled differently: due to some technical issues with xformers, I trained this badboy about 40 times, and some of the examples are from those trainings. All the same dataset, I promise. Just change any weird token names to CharTurnerV2 if you use them.
Token - "CharTurnerV2" place at the front for stronger effect, place at the end for a weaker effect. Token weighting works too
Add:
"Multiple views of the same character in the same outfit"
Add original CharTurner to mix back in some anime/turns
Add charTurner Lora (uploading soon) on top (at very low strength, like .3) to really force the turn. Will affect style.
Next Version: (after the v2 lora and 2.1 versions): Looking for more variety for the data set! Post a review or comment with your best turn around (especially if it was a hard one to get it to do!) and I will use it in the V3 dataset.