The frown expression of this model still needs to be strengthened. It is recommended to use Majic and Chillmix as base models.
The recommended LoRa weights are: 1, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 0, 0, 0, 0.
Description
FAQ
Comments (4)
Low quality LoRA
LoRA metadata reveals that the training did not use any regularization images. Because of this it tends to generate asian faces with cheekbones and definetely does not work with character loras.
Because most of my training set comes from TikTok: argenby, this is the source of Asian faces. And Facts prove that it does work with character loras. You can send out the pic you generated, I can give you some suggestions.
and maybe because this is not a lora intented to work with asian ppl ....
The recommended LoRa weights are: 1, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 0, 0, 0, 0.
17 weights, all of them either 1 or 0?
I was expecting something like "Recommended weight is 0.5 - 0.9"
What is this sequence of 17 digits? Where do they need to go?
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.