基于ANIMA base v1.0训练，训练分辨率为1024p
触发词@yunmi
Description
FAQ
Comments (3)
Love your models very much! ♥️
By the way, if you don’t mind sharing, I'd be really interested to hear what you found most important in training models like these. I’m currently experimenting with this myself, but all of them looked awful (the image being too flat, or other concepts mixing in), while your LoRAs all seem much better and universally applicable.
No pressure to share anything you’d rather keep private, of course! Even a few general pointers would be really helpful.
Thanks for your appreciation! To be honest, summarizing everything in a comment box is tough, so I actually compiled my entire workflow into two agent skills and comprehensive guides on GitHub:
Tagging & Decoupling : anima-style-lora-caption
Hyperparameter & Training : anima-lora-config
They are based on extensive trial and error with Anima Base, as well as insights gathered and summarized from community practices, so hopefully they help you dial in your models!
@reisen8848 🤯Wow, thank you so much. I honestly wasn't expecting you to share the whole workflow like this. This is incredibly generous of you.
I'm still pretty new to training models, so being able to look through an actual working setup like this is going to save me a ton of trial and error and help me understand the process much better. I really appreciate you taking the time to package everything up and share it so openly.
Thanks again, and looking forward to your future models and works!