Attention!, LoCon Version need install Extension: https://github.com/KohakuBlueleaf/a1111-sd-webui-locon

Update 2023-03-23

add LoRA-v1.1

• traing by 100 images

• resolution = "704,1024"

• noise_offset = 0.05

• lion optimizer

• network_dim / network_alpha = 128

• checkpoint: Anything v4.5 (DO NOT Forget VAE)

Update 2023-03-08

add LoCon-v1.1, smaller and better, only 2.5MB

Update 2023-03-07

add LoCon Version, it is small and great, give it try

Original

better than expected, Lora has great potential.

• checkpoint: abyssorangemix3-aom3

• no trigger words required

• recommond weight: 1.0

• you can use any prompts

• recommond use hires.fix

The weight can be adjusted as desired. Besides the texture and style, the composition has a high degree of repetition, perhaps due to the insufficient number of images?

I will keep researching.

用了50张图训练，输出模型72MB。结果比我预想的要好，看来Lora的潜力很高。

权重应该是可以根据需要随意调整，除了风格和纹理，构图重复度比较高，也许是我图片太少了？

我会保持探索。