<h3 id="heading-16">This release has greatly improved color saturation and clarity, while also greatly increasing the number of scenes and fantasy images, bringing more richness.</h3><h3 id="heading-17">Sampler method:DPM 2M Karras</h3><h3 id="heading-18">magnification algorithm: R-ESRGAN_4x Anime6B</h3><hr />Picture size recommendation: 512,768,1024,1200,1280,1920,2048Full Body Support: (Close Range, Mid Range, Long Range, Full Body)High resolution recommended: 8x-NMKD-Superscale,R-ESRGAN_4x Anime6B,4x-UltraSharp(It is recommended to add words such as prompt words (8K) and (Ultra HD) to improve the texture of the picture when high resolution is not turned on.)Recommended sampling times: 25Guidance coefficient of prompt words: 7-10

Wanxiang Animation (SD-1.5 Edition)

wanxiangAnimationSD15_v10.safetensors

v1.0

v4.0

v3.0

v2.0（Enhanced Edition）

Wanxiang Animation (SD-1.5 version)

Recommended trigger words: 2.5D or 3d or realistic texture , add to positive prompts to achieve the desired effect The prompt runs straight ahead Reverse hint words: (no trigger words) This model uses 512-1024-2048 Picture Training The drawing is delicate, no need to open big algorithms, and the old one is clear and beautiful Recommended image size: 512, 768, 1024, 1200, 1280, 1920, 2048 Full body support: (Close View, Medium View, Long View, Full Body) Sampling method: dpm++ 2m karras High resolution recommendations: 8x-nmkd-superscale, r-esrgan_4x+ anime6b,4x-ultrasharp (Recommended when high resolution is not turned on, Add words such as captions (8k) and (Ultra HD) to enhance the image quality.) Recommended number of samples: Default recommendation 25 (Can also be upgraded and adjusted according to PC configuration) Cue guidance coefficient: 7-10 Display map: ((no lora, no amplification, no ad)))

万享1.5 _21.safetensors

Enhanced edition

Reduce damage.

LCM's new model

Files with SHA256: 64cfe07847179a8cf9eb16dea0ba64744b58889431ab362f27734538bf4fdbc5