Highly recommend using it on CivitAI's generator (1.6 - 2 strenght)

Base prompt: Joelzasa style illustration.

Dataset: 400 images

After 12 hours of training, I can safely say this LoRA works… alright. It's definitely good at translating the core elements of the art style but clearly gets stuck when it comes to and changing proportions to match the actual style itself. What I do note is that an increase in dim and alpha can certainly give better results in terms of the LoRA learning more tiny details and nuances in the dataset images. Anyways, hope you like it.

Follow the original artist (joelsza.art):

https://www.instagram.com/joelsza.art?utm_source=ig_web_button_share_sheet&igsh=ZDNlZDc0MzIxNw==