Unlike the previous Turbo + LCM version, this is a derivative of Animagine XL V3. We made slight adjustments to the hand effects, fine-tuned the style and colors, and added some content. If your hard drive space is limited, please confirm whether you really need to download this model, considering its size of 6.46 GB. Don't waste valuable hard drive space; you can continue to use the original Animagine XL V3. The usage is the same as AXL 3, including everything from AXL 3. It inherit CLIP from Animagine, so most of anime game characters are built in: <a target="_blank" rel="ugc" href="https://huggingface.co/spaces/Linaqruf/animagine-xl/blob/main/wildcard/character.txt">https://huggingface.co/spaces/Linaqruf/animagine-xl/blob/main/wildcard/character.txt</a>Be aware, since this model is a descendant of Animagine V3, it follows Fair <a rel="ugc" href="https://freedevproject.org/faipl-1.0-sd/">AI Public License 1.0-SD.</a>

Colorful Anime XL 彩璃二次元 XL

Turbo + LCM Extreme Edition Version

colorfulAnimeXLXL_turboLCM.safetensors

Turbo + LCM

v2.0

Colorful Anime XL Glass Anime XL

Turbo + LCM Version Introduction &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; This is a Turbo+LCM extreme speed version specially designed based on the Anime Cute XL two-dimensional animation large model of the colored glass series, which can greatly accelerate the production speed of XL large models. As we all know, the iteration speed in this community is calculated in hours. Just as LCM was introduced a few days ago, XL Turbo immediately followed. With the support of XL Turbo and LCM, this extreme speed version can generate about 8 images at 4x2 in approximately 10 seconds locally using 4090 testing at 832x1280 non-HD resolution, with astonishing speed and only a slight decrease in image quality. &nbsp; &nbsp; The model has embedded sdxl_vae, allowing image generation without the need to add VAE for the convenience of beginners. &nbsp; &nbsp; &nbsp; &nbsp; Instructions for Using Turbo+LCM Version: &nbsp; &nbsp; Sampler: Euler a , &nbsp; Iteration Steps: 6-8 &nbsp;, CFG Scale for Prompt Word Relevance: 2 &nbsp; &nbsp; Recommended to use 4x-UltraSharp for high-definition restoration, adetailer for facial and hand retouching, sample image size of 832x1280 (size of most training set images), feel free to experiment with others or try with reference images.

Files with SHA256: 4d4b5b35a6aafa4864775ae9c62b25919a50af5cad114a4dcc4b48177181228d