## English Description

This is my complete ANIMA anime character workflow for ComfyUI.

The workflow is organized according to the image-generation process:

1. Resolution, batch size, and global seed settings

2. ANIMA UNet, Qwen CLIP, VAE, and LoRA loading

3. Positive and negative prompts

4. Base sampling, decoding, and image preview

5. Upscale model and target-size calculation

6. Ultimate SD Upscale tiled refinement

7. Local refinement with detectors, SAM, and Detailer

The local refinement section is designed to improve faces, hands, eyes, and other areas that commonly contain generation errors.

Group switches are included, allowing each section to be enabled, disabled, or hidden independently. Generate a base image first, then enable the upscaling and local refinement sections after confirming the composition and style.

这是我制作并整理的一套 ANIMA 二次元人物 ComfyUI 工作流。

工作流按照实际出图流程划分为多个独立区域：

1. 分辨率、批量数量与全局种子设置

2. ANIMA UNet、Qwen CLIP、VAE 与 LoRA 加载

3. 正向提示词与负向提示词

4. 基础采样、解码与图片预览

5. 放大模型与目标尺寸计算

6. Ultimate SD Upscale 分块重绘

7. 检测器、SAM 与 Detailer 局部精修

局部精修部分主要处理脸部、手部和眼睛等容易出错的位置。

工作流带有组开关，可以单独开启、关闭或隐藏每个功能区域。建议先生成基础图片，确认构图与风格后，再开启高清处理和局部精修。