Preface:
Im learning how create good LoRA´s at the moment. I think learning to make realistic persons is the first step to see if the LoRA´s are good. It´s hard to make a realistic person that is not overtrained, so i already did around 30 LoRA´s and here is the best so far.
V1:
The face results are awsome!
The problem: its very stiff and not yet perfect with other LoRA´s. I will update as soon as i get better results, i do this 24h/7 at the moment!
Feedback please!
Please if you have some time after testing it give me some feedback how the outcome is for you! Any support is very welcomed :)!
About V2
The new model has much more flexibility! The face looks mostly realistic! I personally don't think V2 is any more realistic than V1 but it has so much more flexibility and still looks very similar. I would upgrade to V2 if you have already downloaded V1.
Reason for new update
I had a bit of a "breakthrough" with my LoRA's and am happy with this model for now. I will now try new characters! If you suggest a person I'll see if I can implement them next, just comment!
Short overview:
Quality: /similar as V1
Flexibility: ++ Way better with any checkpoint!
Combination with other LoRA´s: ++ way better!
描述:
V1:
Good: Awsome Face | Works with any Checkpoint | High Resolution
Bad: Too stiff yet |
训练词语: K4rlottaNila
名称: K4rlottaNilaV1.safetensors
大小 (KB): 73858
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success
