Does the prompt you used to Training affect your reasoning using Controlnet?
#44
by
Pupba
- opened
I used diffusers base code to fine-tun the SDXL-ControlNet.
At this time, I tried the prompts of about 500 data sets the same and tried differently, but the results did not change significantly.
I'm going to prepare more datasets this time, but will the prompt detailing the images greatly affect the image creation that I target?
Also, does using prompts used to train ControlNet for inference greatly affect the outcome?