在保持画质基本不变的基础上，修复了已知问题（你懂的），经过测试 即使不输入提示词也能改变画风一种基于daioo画风的Lora，请使用rinFlanimeAnima_V14模型作为底模使用，该模型有着优秀的模型品质，推荐使用dpmpp_2m_sde_gpu采样器配合sgm_uniforrm以获得最佳速度和质量，此Lora有着不错的风格效果，但请注意初版训练对于nsfw内容有很大概率打码。通过在正面提示词添加<code>uncensord,</code> 和<code>censored, mosaic_censoring,</code>负面提示词可以消除，V2版本已经修复。模型基于loha格式训练，虽然训练时间更长但画风泛化和文件尺寸上有着优秀的表现。返图请尽量不要上传nsfw内容（个人喜好）。谢谢，接下来请尽情使用吧!While keeping image quality essentially unchanged, we’ve fixed some known issues (you know the ones); testing shows that you can now alter the art style even without entering any prompts.This is a LoRA based on the "daioo" art style. Please use the <code>rinFlanimeAnima_V14</code> model as the base model, it offers excellent quality. For the best balance of speed and quality, I recommend using the <code>dpmpp_2m_sde_gpu</code> sampler combined with <code>sgm_uniform</code> scheduler. While this LoRA delivers a great stylistic effect, please note that, due to the nature of this initial training, there is a high probability that NSFW content will be censored (pixelated).You can attempt to remove this by adding <code>uncensored</code> to your positive prompts, or by including <code>censored</code> and <code>mosaic_censoring</code> in your negative prompts; Fixed in version V2. The model was trained using the LoHA format, which, despite longer training times, provides superior stylistic generalization and a more efficient file size. Please refrain from uploading NSFW content when sharing your generated images (a personal preference). Thank you—enjoy using it!

Daioo Style Anima

在保持画质基本不变的基础上，修复了已知问题（你懂的），经过测试 即使不输入提示词也能改变画风 While keeping image quality essentially unchanged, we’ve fixed some known issues (you know the ones); testing shows that you can now alter the art style even without entering any prompts.

Files with SHA256: 5dfbc15fc8cf11e47f90b7a1d06dd0c491242721001da8d6108b7567cfe9f997