2.0版本使用说明（必看）

• lora强度设在0.8-1.0之间：低于0.5一般不生效，0.5-0.8之间不太明显，超过1.0容易过拟合，超过1.5会直接生成训练的素材图 ：）

• 注意：触发词为tt-marks

• 不需要在提示词的开头单独写“tt-marks”作为触发词，你只需要将你的提示词正文中的“cane marks”调整为“tt-marks”即可。

• 例如“Revealing her buttocks with cane marks”应写作“Revealing her buttocks with tt-marks”，一般这样写就可以生效。或者只需你的提示词里提到了“裸露的臀部”，就会生效，通过调整lora强度来控制藤条痕迹的颜色深浅。

• 注意最好是小写的tt而非TT，否则有可能直接在屁股上出现“T”形痕迹。

• 提示词中最好有一句类似“girl with delicate makeup”这样的脸部简单描述，以及“eyes looking at the camera”这样的描述，否则无论如何你永远只能看见她的屁股，而看不见她的脸。

2.0 version instructions (must read)

1、Set the lora strength between 0.8-1.0

• It is generally not effective if it is lower than 0.5

• It is not very obvious between 0.5-0.8

• It is easy to overfit if it exceeds 1.0

• If it exceeds 1.5, the training material image will be directly generated ：）

2、The trigger word is “tt-marks”

• You don’t need to write "tt-marks" as a trigger word at the beginning of the prompt. You only need to adjust the "cane marks" in the body of your prompt to "tt-marks".

• For example, "Revealing her buttocks with cane marks" should be written as "Revealing her buttocks with tt-marks", it will generally effective.

• Or as long as your prompt word mentions "exposed buttocks", it will take effect, and the color depth of the cane marks can be controlled by adjusting the lora strength.

• It is better to use “tt” not “TT”, otherwise there may be a "T"-shaped mark directly on the buttocks.

3、Others

• It is best to have a simple facial description like "girl with delicate makeup" and a description like "eyes looking at the camera" in the prompt.

• Otherwise, you can only see her buttocks forever.