Sign In

Ao Yem V3 _ SDXL

11
159
4
Verified:
SafeTensor
Type
LoRA
Stats
158
0
Reviews
Published
Nov 14, 2023
Base Model
SDXL 1.0
Training
Steps: 20
Epochs: 30
Trigger Words
Vietnamese girl wear aoyem
<lora:10_aoyem_v1:1>
mung headband
Hash
AutoV2
8E8DC9C0DE
default creator card background decoration
hts2007's Avatar
hts2007

- Trained with 68 images that above 1024px.

- Base on XXMix_9realisticSDXL, so it will best result when generate with that model.

- Recommend sampling method is:

  • DPM++ 3M SDE Exponential: best result but always fit with trained image (background, style, pose) but less issue with hand.

  • DPM2 a Karras: quite same quality with 3M SDE Exponential, but less fit the background about 20-30%. Has issue with finger.

  • DPM++ 2M Karras: the quality less than 3M SDE Exponential about 20-30%. But not fit with background, the clothes look like 80%,

- CFG: 6 - 9

- Must use with Hires fix. Recommend: RealESRGAN_x4plus, ESRGAN_4x