• 版本 [version]
  • 下载 49
  • 文件大小 18691
  • 文件计数 1
  • 创建日期 2026年1月7日
  • 最后更新 2026年9月21日

train_lora

这是一份开箱即用的人像调节工作流(SD 1.5、ControlNet),拖入 ComfyUI 即可运行,全图 20 节点 / 23 连线。核心流程 CLIPTextEncode ×2、PreviewImage ×2、LoadImage ×2,输出图像,默认输出 1328×800,参数已调好,替换输入图即可复现效果。主模型 ghostmix_v20Bakedvae.safetensors、control_v11p_sd15_openpose.pth、control_v11p_sd15_lineart.pth。已挂载 LoRA:LCM_LoRA_Weights_SD15.safetensors。下载后直接导入即可使用,建议显存 4GB 起。

  • 主模型:ghostmix_v20Bakedvae.safetensors、control_v11p_sd15_openpose.pth、control_v11p_sd15_lineart.pth
  • LoRA / 适配器:LCM_LoRA_Weights_SD15.safetensors
  • 核心节点:CLIPTextEncode ×2、PreviewImage ×2、LoadImage ×2
  • 输出形式:输出图像 · 1328×800
  • 规模:20 节点 / 23 连线
  • 建议显存:4GB 起
  • 适用场景:人像精修、磨皮、修手修脸、去 AI 味与表情姿态微调。

工作流节点图

train_lora 工作流节点图

图较大可横向滚动;点击图片查看高清矢量原图(浏览器中可自由缩放)。

资源头图

train_lora

使用说明

下载工作流文件后在 ComfyUI 中直接拖入画布(或菜单 Load 导入)即可运行;缺失的自定义节点用 ComfyUI Manager 一键安装,模型按文件名放入对应的 models 目录。

源码预览

{
    "last_node_id": 33,
    "last_link_id": 56,
    "nodes": [
        {
            "id": 6,
            "type": "CLIPTextEncode",
            "pos": [
                399,
                178
            ],
            "size": [
                422.84503173828125,
                164.31304931640625
            ],
            "flags": [],
            "order": 13,
            "mode": 0,
            "inputs": [
                {
                    "name": "clip",
                    "type": "CLIP",
                    "link": 19,
                    "label": "clip",
                    "slot_index": 0
                }
            ],
            "outputs": [
                {
                    "name": "CONDITIONING",
                    "type": "CONDITIONING",
                    "links": [
                        32
                    ],
                    "slot_index": 0,
                    "label": "CONDITIONING"
                }
            ],
            "properties": {
                "Node name for S&R": "CLIPTextEncode"
            },
            "widgets_values": [
                "(a character sheet of a woman from different angles with a grey background:1.4) , auburn hair, eyes open, cinematic lighting, Hyperrealism, depth of field, photography, ultra highres, photorealistic, 8k, hyperrealism, studio lighting, photography"
            ]
        },
        {
            "id": 5,
            "type": "EmptyLatentImage",
            "pos": [
                500,
                623
            ],
            "size": [
                315,
                106
            ],
            "flags": [],
            "order": 0,
            "mode": 0,
            "outputs": [
                {
                    "name": "LATENT",
                    "type": "LATENT",
                    "links": [
                        34
                    ],
                    "slot_index": 0,
                    "label": "LATENT"
                }
            ],
            "properties": {
                "Node name for S&R": "EmptyLatentImage"
            },
            "widgets_values": [
                1328,
                800,
                1
            ]
        },
        {
            "id": 27,
            "type": "LineArtPreprocessor",
            "pos": [
                490,
                830
            ],
            "size": [
                315,
                82
            ],
            "flags": [],
            "order": 9,
            "mode": 0,
            "inputs": [
                {
                    "name": "image",
                    "type": "IMAGE",
                    "link": 38,
                    "label": "image"
                }
            ],
            "outputs": [
                {
                    "name": "IMAGE",
                    "type": "IMAGE",
                    "links": [
                        39,
                        40
                    ],
                    "shape": 3,
                    "label": "IMAGE",
                    "slot_index": 0
                }
            ],
            "properties": {
                "Node name for S&R": "LineArtPreprocessor"
            },
            "widgets_values": [
                "disable",
                1024
            ]
        },
        {
            "id": 7,
            "type": "CLIPTextEncode",
            "pos": [
                411,
                385
            ],
            "size": [
                425.27801513671875,
                180.6060791015625
            ],
            "flags": [],
            "order": 14,
            "mode": 0,
            "inputs": [
                {
                    "name": "clip",
                    "type": "CLIP",
                    "link": 20,
                    "labe
… (完整内容请下载文件查看)

代码较长,以上仅前 3800 字符 —— 点击下方按钮下载完整 JSON。

发表回复

你的电子邮件地址不会被公开 * 为必填字段

提交评论