This is a blue-and-white style 2D model
The recommended weight is 0.8-1
Recommended resolution 768 x 1024
For training reasons, v2 has been replaced with a version of v1 with better prompt word compliance
Description
More images were used for training
Personal feeling effect is better, welcome to try the experience
FAQ
Comments (3)
I really like V1 even if this lora does not follow the prompts or only vaguely.
V2 on the other hand gives horrible results and at worst distorted, at best humm strange or even off topic...
Just see what I posted in the V2 section ( https://civitai.com/images/40778431 ) where we see the same prompt and the same settings with just a change of lora version... tested based (only for the test) on the prompt visible here:
https://civitai.com/images/36472952
We will have to correct this for the 3rd version that I will test with pleasure.
Thank you for liking and testing carefully😇
I have a look, the main reason for this problem is that I changed a trainer training (it seems that the trainer is very badðŸ˜), and then did not carry out a very rigorous test to lead to the current effect.
v1 I chose the one with a strong style in the prompt word compliance and strong style, later I will replace the model in v2 with a model with a better prompt word compliance in v1, hope you have fun🥰
@xjie Courage! In the meantime I continue to test my prompts with a test of V1 (I'm addicted ;-)). For V2, I notice that there is often a sculpture (Greco-Roman) of a man's face in my results while the prompt does not mention sculpture, portrait or man.