



















This is a mix of models with really good backgrounds and composition. It can also do NSFW
Check the main post's or gallery's images for prompt examples and read the section below for more information.
If you run into issues with the model, feel free to leave a comment or message me using the on-site chat.
↓ !! CLICK "Show More" FOR MY RECOMMENDED USAGE !! ↓
My Generation settings:

CLIP SKIP: 2
Denoising strength: ~ 0.2 - 0.4
CFG scale: 7 ~
VAE: kl-f8-anime2
Use DANBOORU tags for prompting
*with my settings, might differ if using a different Sampler or Upscaler
* Latent (nearest-exact) requires denoising strength at around 0.5+

Negative prompt important
Embeddings: Optional, but highly recommended. Use one or two of these from below.
(EasyNegative:0.8), -> higher contrast / shadows / rounder eyes and face
(negative_hand-neg:0.8), -> minimal changes, ~better hands
(by bad-artist:0.8), | (by bad-artist-anime:0.8), -> similar to EasyNegative, but weaker
*INSERT NEGATIVE EMBEDDINGS*
(worst quality, low quality:1.2),
---OPTIONAL---
(blurry, bokeh, depth of field),
(watermark, username, artist name, signature, text, monochrome, greyscale),Notes: I haven't messed around too much with background / composition tags that can create awesome images. So if you know some, I would love to see them used with this checkpoint.
Ignore the "detailed hair" on the no humans images with the flower, frog and bridge 😅

Click on the 🔔 or Follow me if you want to receive a notification when I update or release new models.
Check out my profile.
I have other models that might also pique your interest :)
1. 转载模型仅供学习与交流分享,其版权及最终解释权归原作者。
2. 模型原作者如需认领模型,请通过官方渠道联系海艺AI工作人员进行认证。我们致力于保护每一位创作者的权益。 点击去认领
