Nata Lee | Natalie Lee | NataLee

Paste the Lora file in "stable-diffusion-webui\models\Lora\"

Use it with (Automatic1111 - Newest Version):

rename the downloaded file to "MyNataLeeV1"

<lora:MyNataLeeV1:1>

*try to play with the weight like <lora:MyNataLeeV1:0.85>

Trigger:

MyNataLee woman

***Lora quality depends of the used model and prompt/negative prompt

***sometimes you can use inpaint and img2img to fix small errors

________________________________________________________________________________________________________

Tips for Automatic1111:

-> use SD VAE: vae-ft-mse-840000-ema-pruned

-> Set Clip skip: 1

-> Eta noise seed delta: 31337 (to get the same images)

-> try Restore faces On / Off (mostly i turn it off)

• full body shots - Restore face On helps sometimes

• closeups - Restore face Off

-> Sampling steps 50+

-> Use Hires fix! (makes the Lora/face better)

My Hires.fix Settings:

Upscaler: Latent / R-ESRGAN 4x+

Hires steps: 50+

Denoising str: 0,55-0,7

Upscale by: 1,3

________________________________________________________________________________________________________

Tested on Models:

- f222 (https://civarchive.com/models/1188/f222)

- hassanblend1512 (https://civarchive.com/models/1173/hassanblend-1512-and-previous-versions)

- protogenX58 (https://civarchive.com/models/3867/protogen-x58-rebuilt-scifianime-official-release)

- realisticVisionV13 (https://civarchive.com/models/4201/realistic-vision-v13)

- urpmv12 (https://civarchive.com/models/2661/uber-realistic-porn-merge-urpm)

- DreamShaper (https://civarchive.com/models/4384/dreamshaper)

________________________________________________________________________________________________________

Recreate my images:

A: save sample images and paste them in "Automatic1111 --> PNG Info ---> Send to txt2img"

B: press the info icon on the image -> Copy Generation Data -> paste into txt2img prompt ->

click the blue button on the right (read generation parameters...)