The strength of Lora is recommended to be 0.7-1.0.
Since all images in the training set have a 3:2 aspect ratio, the composition tends to look more natural when generating images with the same ratio.
Description
FAQ
Comments (1)
A great work! This LoRA produces a person in a cargo space of a sedan or a hatchback nicely. It is not easy to do so only by a prompt; I know it because I have tried so many times myself. Recommended for scenes such as a consensual BDSM plays, roleplay kidnapping, or a childish mischief.
The problem is, as noted in the model description, that it often generates a strange shape of a car in portrait format (like this: Image posted by amich). I hope the flaw will be fixed in the future version. Also, it is great if it becomes responsive to the prompt text for the body type of the car, e.g., sedan, convertible, hatchback, van, etc.
Anyway, thank you for the great gift, @asasp!





