Sign In


Updated: Dec 8, 2023
Checkpoint Merge
Dec 8, 2023
Base Model
SD 1.5


πŸ‘Œ This model generates attractive characters with SIMPLE BACKGROUND, and probably, performs well in some other scenes.

🎈 This model is based on Anything v5.

πŸ“• This model merges 3 self-trained styles (LyCORIS) to Anything v5, since the author found the merged style really attractive and its offensive to directly publish the 3 LyCOs separately.

πŸ•ΆThe motivation of merging this model:

  • ✨ The author get bored with the lighting and shadow effects of some models.

  • 🎑 The author pursues clear lines, simple lighting and shadow effects, and, attractive colors.

  • πŸ₯° The author really likes each separate style.

  • πŸ˜‰ For fun.

Explanation of version L0 and L1:

Simply, L1 is merged with slightly larger weights on 2 of the separted styles, which leads to more stylized output.

Supplementary explanation:

Since this model merges LyCO modules, this will lead to VoLaTiLe outputs, overfitting (i.e., on poses, profilings, scenes and watermarks ...), and might perform really poor in some usecases.
This model sometimes performs very well when coupled with some LoRA (As shown in the uploaded sample figures, at least the author deems attractive).

Usage suggestions

CFG: 7.0/8.0 ~ (Sometimes, ~10.0 gives interestingly stylized outputs)

Sampler: Euler a with 20~28 steps (Of course you can also try DPM series, where the author regards less attractive)

Upsampler: ESRGAN anime 6B, with Denoising strength 0.55~0.6

πŸ’©Some nonsense craps, probabaly makes this page cool 😎

  1. Of course at this timepoint, the state-of-the-art-model is NAI3, which INSANELY aligns the prompt inputs and the output image with extremely noticable aesthetic features.

  2. This model has not been sufficiently evaluated, which might produce disappointing outputs in many scenes.

  3. The following is a statement of responsibility.

# Disclaimer for this AI Painting Model

Please read this disclaimer ("disclaimer") carefully before using this AI Painting Model ("the Model") operated by the user ("Us", "We", "Our") on Civitai ("Website").

## Intellectual Property Rights

We would like to remind you that the Model generates artworks based on algorithms and cannot guarantee the originality of the output. Therefore, if you use the Model to create artworks and publicly display or sell them, you may inadvertently infringe on the intellectual property rights of others. We will not be responsible for any infringement caused by the use of the Model.

## Limitations of liability

We will not be held responsible for any losses, damages, or claims that may result from the use of the Model. This includes, but is not limited to, loss of profits or business interruptions. You agree to use the Model at your own risk.

## Accuracy of Information

While we strive to ensure that the information provided by the Model is accurate and up-to-date, we do not guarantee the accuracy, reliability, or completeness of any information produced by the Model. Any reliance on the output of the Model will be at your own risk.

## Changes to the Disclaimer

We reserve the right to modify this disclaimer at any time. When we make changes, we will revise the updated date at the bottom of this page. We encourage users to frequently check this page for any changes to stay informed.

Last Updated: 2023/12/8
  1. Guess what does EHA means :-), for a big LIKE as reward.

  2. Feel free to criticize on the above grammar mistakes.

  3. The author encourages users to fix the image defects by locally inpainting.

Future work

πŸ“… Probably, try some weights' combinitions on A5-Ink version, which seems to produce awesome outputs.