Verified: 2 years ago
SafeTensor
Type | |
Stats | 83 0 |
Reviews | (3) |
Published | Aug 7, 2023 |
Base Model | |
Training | Steps: 7,500 Epochs: 10 |
Usage Tips | Clip Skip: 2 |
Trigger Words | Liam Gallagher |
Hash | AutoV2 2E20FE35FF |
This is my first LoRA trained ever :)
It is trained as a tribute to the rock band Oasis' lead singer Liam Gallagher. I'm also training the one of Noel Gallagher, Oasis' lead guitarist and another famous Britpop rockstar.
This LoRA is ONLY compatible with SDXL 1.0 base model.
!!! DO NOT USE with SDXL 1.0 renifer model !!!
Following recommendation are all intended to use in A1111's SD webui ONLY
Txt 2 Img recommended parameters, using SDXL 1.0 base model:
Sampling method: DPM++ models
Sampling steps: 30-36
Size: 1024 x 1024
GFC scale: 6-8
To further refine the output, you can use SDXL 1.0 refiner model in Img 2 Img.
Remember to set the Denoising strength lower than 0.3
The ideal score would be between 0.15-0.25, otherwise someone else will appear
Currently A1111 SD webui does not support using SDXL 1.0 base + refiner models together in one inference