Version 2.0 Content

This model is extracted LoRa from other sdxl base models and fused to help newcomer adapt to sdxl faster and experience the fun of sdxl. Compared to the original sdxl, this model is easier to use, more compatible, and widely applicable. The model is built-in with VAE and does not require additional VAE model.

-----------------------------------------------------------------------------

The model is open source, but authors who integrate my model are advised to acknowledge the original model to avoid unnecessary misunderstandings.

New Version Update

Version 2.0 has optimized Asian facial features to avoid the inherent impression of Asian faces in SDXL. The new version also adds other elements and fixes issues encountered in version 1.0, adding details. Most scenarios do not require enabling AD facial repair, unless for special wide-angle scenes. In such cases, use ADetailer or Face Editor for assistance.

New version recommended CFG 3.5~0.6 for realism, recommended 3.5~3.8. Other parameters not mentioned should follow the previous version.

-----------------------------------------------------------------------------

Suggested Image Output Parameters

No need to mention the starting promptsbest quality, masterpiece, realistic,start or simple 1girl, are all acceptable without limitations.

Generally, you can run without enabling ad repair, but it is recommended to have ad repair suggestions enabled.

Sampling Method (sampler):

dpm++ 3m sde exponential     45 steps or more

dpm++ 3m sde karras        45 steps or more

dpm++ 2m sde karras   45 steps or more   (Use this for Liebu online)

Prompt Word Guidance Factor (cfg scale): 3.5~6.0

Portrait Resolution: 768x1280, 1024x1024

High-Resolution Repair (hires.fix):: 4x-ultrasharp, high iteration steps 10~15, redraw range 0.3~ 0.5

clip skip: 2

Negative Guidance Words:: text, username, logo, (low quality, worst quality:1.4), bad anatomy, inaccurate limb:1.2, bad composition, inaccurate eyes, extra digit, fewer digits, (extra arms:1.2), watermark, multiple moles, mole on body

-----------------------------------------------------------------------------

Here are some random thoughts

It must be emphasized again that this base model is the result of the integration of multiple models and thus has many uncertainties. Of course, this model does not limit what you understand. Speaking of uncertainties, although the integrated model has many uncertainties, it is due to these uncertainties that unexpected results may arise. You may run into unexpected things through this model or encounter pleasing scenarios. Of course, there may also be negative results, etc. This model is my first attempt at integrating models in sdxl. Due to the uncertainties of integrated models, there may be many problems that I have not tested. If you encounter any issues, please provide feedback promptly and I will iterate in future versions.

-----------------------------------------------------------------------------

Model Usage Declaration

1. You may not host this model and its derivative versions (such as integrated model versions) on websites/apps intended to earn revenue or donations. If you need to use this model and its derivative versions for commercial purposes (such as generative services, selling images, publishing images publicly on short video platforms or publications, etc.), please contact me through the site.

2. You may not directly sell images generated by this model unless you have made sufficient manual modifications to the image so that it can be fully determined as your personal work in the legal sense. If you violate this, any legal consequences will be your personal responsibility, as I am not liable.

3. You cannot use this model to intentionally produce or share illegal or harmful content for dissemination; please adhere to moral standards and use this model for positive and constructive purposes.