A quick lora trained using frames from the sonic movies. No fanart is included in the dataset.
MODEL INFO:
Use tags "movsonic, sonic the hedgehog"
Dynamic poses are often to be seen, use tags or ControlNet if you need to
contact info is on my profile.
Description
an experimental pony version of the lora. Same dataset, different parameters (since I lost the old config)
sometimes the anatomy might not be movie accurate, increase the lora weight a bit to fix this
if the contrast is too high, decrease the weight
use your UI's detailer if you need to
(Adetailer for auto1111, <segment> for SwarmUI, etc)