Type | |
Stats | 468 594 |
Reviews | (55) |
Published | Apr 26, 2024 |
Base Model | |
Training | Steps: 2,736 Epochs: 16 |
Usage Tips | Clip Skip: 2 |
Trigger Words | mesugaki v2 open_mouth evil_smile tongue_out |
Hash | AutoV2 BD70D94BBE |
This version(2.0) optimizes the prompt as a whole, and achieves a good effect under the style that does not affect the checkpoint too much. It is recommended to add the trigger word mesugaki v2 and use emoticons as you need, such as open mouth, evil smile, tongue out, etc.
This model can be used to create the image of a cute girl. Of course, you can create the images you want to use. This LoRA model can run in both realistic and anime styles, but the effect may not be obvious and may cause interference in realistic models.
Recommended models for using this LoRA:
Models that perform better with this LoRA can be found by comparing the XY images listed above.
It is recommended to set the weight to between 0.5 and 0.75 for the best results.
The euler dpm++ Sampling Method works well for all models, while euler a may be faster for generation.
In a RTX4090 environment, it is recommended to generate a 768*512 image in 2.1s.
Enjoy it and have fun~!