What was the Dreambooth finetuning call
Hello,
I was wondering what --class-prompt and --instance-promt were used to achieve this? And have you added handpicked photos to a CLASS_DIR?
Hi there
instance_prompt= "archer style"
class_prompt= "illustration style"
The class images where generated beforehand with SD 1.4 with the prompt "illustration style" and 30 Steps, DDIM Sampler
Thank you!
Hello, awesome model and awesome work!
Just wondering what is the batch size and learning rate for achieving this result at 4000 steps?
Hi there
instance_prompt= "archer style"
class_prompt= "illustration style"The class images where generated beforehand with SD 1.4 with the prompt "illustration style" and 30 Steps, DDIM Sampler
Can you explain how you generated the class images and what they are for? I have spent weeks with Dreambooth but still don't understand the classification images.
Edit: I think I got it now.
You mean that you created images that are similar to what you use for training, but a totally different style, right?