Sign In

Scarlett Vivienne

34

203

16

Verified:

SafeTensor

Type

LoRA

Stats

203

0

Reviews

Published

Jan 17, 2024

Base Model

SDXL 1.0

Training

Steps: 4,400
Epochs: 10

Usage Tips

Clip Skip: 2
Strength: 0.6

Trigger Words

amber heard woman

Hash

AutoV2
8826E7AC2E
  1. Check us out on TikTok @stableloras and Instagram @stableloras and tell us what you want to see next!

  2. Our new model (Scarlett Vivienne) is loosely inspired by the stunning features of celebrity actress Amber Heard, known to Stable Diffusion. By using the keyword "amber heard woman" and applying our LoRA at a weight of 0.6, the software produces the familiar beauty which is then uniquely enhanced by our specialized LoRA. This approach ensures an underlying consistency with a sprinkle of distinctive flair, making each creation intimately familiar.

  3. For more creative results, up the CFG scale from 7 up to 12. This will adhere more to your prompt. Don't forget to add the usual negative prompts too.

  4. Scarlett Vivienne was trained on a very high resolution set of nude images, ranging from headshots to full body. Which means she is capable of producing NSFW images.

  5. For those finer details that breathe life into a portrait, we employ the After Detailer on the face and hands, ensuring each feature is rendered with the utmost clarity and realism. Our rendering process works great at 800x1024 and retains the model's pristine details without the need for using the HiRes upscaler.

  6. Innovating further, we utilize the 'extras' tab in Stable Diffusion for a new upscaling technique. By employing the ESRGAN_x4 upscaler, Scarlett Vivienne emerges in high resolution while maintaining the integrity of her features—free from the artifacts commonly introduced by traditional HiRes upscaling methods.

  7. Scarlett Vivienne is not just a model; she is a digital masterpiece waiting to grace your project with her poised and tranquil presence.