添加单张图片变换

CLAUDE.md

@@ -7,61 +7,79 @@ This file provides guidance to Claude Code (claude.ai/code) when working with co
 Each service runs independently and must be started separately:
 
 ```bash
-# Shirt-only try-on service (port 8888)
+# Single-image edit service (port 12221)
+python change1/service1.py
+
+# Shirt-only try-on service (port 12222)
 python change2/service2.py
 
-# Shirt + pants try-on service (port 8889)
+# Shirt + pants try-on service (port 12223)
 python change3/service3.py
 
 # Gateway router (port 28888) — requires Flask
 python change_app.py
 ```
 
-ComfyUI must be running at `http://127.0.0.1:8188` before starting the services.
+ComfyUI must be reachable at the URL configured in each service (currently `http://112.126.94.241:28188`, with `:38188` as backup).
 
 ## Dependencies
 
 ```bash
-pip install fastapi uvicorn httpx pydantic flask requests
+pip install fastapi uvicorn httpx pydantic flask requests oss2 pillow
 ```
 
 ## Architecture
 
-This is a three-tier AI virtual try-on system built around ComfyUI.
+This is an AI virtual try-on / image-edit system built around ComfyUI, fronted by three FastAPI workers and a Flask gateway.
 
 ```
 Client
   └─► change_app.py (Flask, port 28888)            ← Gateway router
-        ├─► change2/service2.py (FastAPI, port 8888)  ← shirt-only
+        ├─► change1/service1.py (FastAPI, 12221)   ← single image (model only)
-        └─► change3/service3.py (FastAPI, port 8889)  ← shirt + pants
+        ├─► change2/service2.py (FastAPI, 12222)   ← model + shirt
-              └─► ComfyUI (port 8188)
+        └─► change3/service3.py (FastAPI, 12223)   ← model + shirt + pants
+              └─► ComfyUI
 ```
 
-**Gateway** (`change_app.py`): Accepts `POST /change_cloth_base64` with `human_img`, `cloth_img`, and optional `kuzi_img` (pants). Routes to `service3` when `kuzi_img` is present, otherwise to `service2`. Both backends are called at `/do_change_cloth`.
+**Gateway** (`change_app.py`): Accepts `POST /change_cloth_base64` with `human_img` (required) and optional `cloth_img` / `kuzi_img` / `change_desc`. Routing rules:
+- `kuzi_img` present → `service3` (12223)
+- `cloth_img` present (no `kuzi_img`) → `service2` (12222)
+- only `human_img` → `service1` (12221)
 
-**service2** (`change2/service2.py`): Handles shirt-only try-on. Exposes `POST /try-on` with `model_image` + `shirt_image`. Uses ComfyUI workflow `change2_1_ex.json`.
+All backends expose `POST /try-on` and return `{ "result_url": "<oss url>" }`. The gateway wraps that into `{ "ret": 0, "state": 0, "msg": "success", "data": <oss url> }`.
 
-**service3** (`change3/service3.py`): Handles shirt + pants try-on. Exposes `POST /try-on` with `model_image` + `shirt_image` + `pants_image`. Uses ComfyUI workflow `change2_2_0308.json`.
+**service1** (`change1/service1.py`): Single-image edit (e.g. background replacement, full-body generation from a reference). Exposes `POST /try-on` with `model_image` + optional `change_desc`. Uses ComfyUI workflow `change1.json`.
+
+**service2** (`change2/service2.py`): Shirt-only try-on. Exposes `POST /try-on` with `model_image` + `shirt_image` + optional `change_desc`. Uses ComfyUI workflow `change2_1_ex.json`.
+
+**service3** (`change3/service3.py`): Shirt + pants try-on. Exposes `POST /try-on` with `model_image` + `shirt_image` + `pants_image` + optional `change_desc`. Uses ComfyUI workflow `change2_2_0308.json`.
 
 ### ComfyUI Workflow Integration
 
-Both services dynamically modify a JSON workflow template before submission:
+Each service dynamically modifies a JSON workflow template before submission:
 
-| Node ID | Purpose | service2 | service3 |
+| Node ID | Purpose | service1 | service2 | service3 |
-|---------|---------|----------|----------|
+|---------|---------|----------|----------|----------|
-| `"11"` | Model/person image | ✓ | ✓ |
+| `"11"` | Model/person image (LoadImage) | ✓ | ✓ | ✓ |
-| `"10"` | Shirt image | ✓ | ✓ |
+| `"10"` | Shirt image (LoadImage) | — | ✓ | ✓ |
-| `"4"`  | Pants image | — | ✓ |
+| `"4"`  | Pants image (LoadImage) | — | — | ✓ |
-| `"33"` | Output image | ✓ | ✓ |
+| `"31"` | Text prompt (PrimitiveStringMultiline, overridden when `change_desc` is non-empty) | ✓ | ✓ | ✓ |
+| `"33"` | Output image (SaveImage) | ✓ | ✓ | ✓ |
 
 The request flow within each service:
-1. Decode base64 images
+1. Pick a healthy ComfyUI host (primary, fallback to backup)
-2. Upload images to ComfyUI (`/upload/image`) with UUID-prefixed filenames to avoid cache collisions
+2. Decode base64 images
-3. Inject filenames into workflow node inputs
+3. Upload images to ComfyUI (`/upload/image`) with UUID-prefixed filenames to avoid cache collisions
-4. Submit workflow to ComfyUI queue (`/prompt`)
+4. Inject filenames into workflow node inputs; if `change_desc` is non-empty, replace node `31` inputs with `{"value": change_desc}`
-5. Poll `/history/{prompt_id}` until `status.completed` is true (max 300s timeout, 2s poll interval)
+5. Submit workflow to ComfyUI queue (`/prompt`)
-6. Download result from `/view` and return as base64
+6. Poll `/history/{prompt_id}` until `status.completed` is true (max 300s timeout, 2s poll interval)
+7. Download the result from `/view`, push it to Aliyun OSS, and return the signed URL
 
 ### Frontend
 
-Each service has its own test UI at `/` (served from `<service_dir>/static/index.html`). The `change2` UI accepts 2 images; `change3` accepts 3 images. Both POST directly to the service's `/try-on` endpoint.
+Each service has its own test UI at `/` (served from `<service_dir>/static/index.html`):
+- `change1` UI: 1 upload card (model) + prompt panel
+- `change2` UI: 2 upload cards (model, shirt) + prompt panel
+- `change3` UI: 3 upload cards (model, shirt, pants) + prompt panel
+
+All three POST directly to the service's own `/try-on` endpoint.

change1/change1.json

@@ -0,0 +1,400 @@
+{
+  "1": {
+    "inputs": {
+      "samples": [
+        "22",
+        0
+      ],
+      "vae": [
+        "13",
+        0
+      ]
+    },
+    "class_type": "VAEDecode",
+    "_meta": {
+      "title": "VAE解码"
+    }
+  },
+  "2": {
+    "inputs": {
+      "to_ref": true,
+      "ref_main_image": false,
+      "ref_longest_edge": 1024,
+      "ref_crop": "center",
+      "ref_upscale": "lanczos",
+      "to_vl": true,
+      "vl_resize": true,
+      "vl_target_size": 384,
+      "vl_crop": "center",
+      "vl_upscale": "bicubic",
+      "image": [
+        "11",
+        0
+      ],
+      "configs": [
+        "8",
+        0
+      ]
+    },
+    "class_type": "QwenEditConfigPreparer",
+    "_meta": {
+      "title": "Qwen Edit Config Preparer"
+    }
+  },
+  "6": {
+    "inputs": {
+      "conditioning": [
+        "15",
+        0
+      ]
+    },
+    "class_type": "ConditioningZeroOut",
+    "_meta": {
+      "title": "条件零化"
+    }
+  },
+  "7": {
+    "inputs": {
+      "custom_output": [
+        "15",
+        2
+      ]
+    },
+    "class_type": "QwenEditOutputExtractor",
+    "_meta": {
+      "title": "Qwen Edit Output Extractor"
+    }
+  },
+  "8": {
+    "inputs": {
+      "to_ref": true,
+      "ref_main_image": true,
+      "ref_longest_edge": [
+        "9",
+        0
+      ],
+      "ref_crop": "pad",
+      "ref_upscale": "lanczos",
+      "to_vl": true,
+      "vl_resize": true,
+      "vl_target_size": 384,
+      "vl_crop": "center",
+      "vl_upscale": "bicubic",
+      "image": [
+        "11",
+        0
+      ]
+    },
+    "class_type": "QwenEditConfigPreparer",
+    "_meta": {
+      "title": "Qwen Edit Config Preparer"
+    }
+  },
+  "9": {
+    "inputs": {
+      "max_size": 2048,
+      "image": [
+        "11",
+        0
+      ]
+    },
+    "class_type": "QwenEditAdaptiveLongestEdge",
+    "_meta": {
+      "title": "Qwen Edit Adaptive Longest Edge"
+    }
+  },
+  "11": {
+    "inputs": {
+      "image": "cq5dam.web.hE7E3DA.1800.1800 - 2024-04-29T161212.795.jpg"
+    },
+    "class_type": "LoadImage",
+    "_meta": {
+      "title": "加载图像"
+    }
+  },
+  "13": {
+    "inputs": {
+      "vae_name": "qwen_image_vae.safetensors"
+    },
+    "class_type": "VAELoader",
+    "_meta": {
+      "title": "加载VAE"
+    }
+  },
+  "14": {
+    "inputs": {
+      "prompt": "",
+      "return_full_refs_cond": true,
+      "instruction": "Describe the key features of the input image (color, shape, size, texture, objects, background), then explain how the user's text instruction should alter or modify the image. Generate a new image that meets the user's requirements while maintaining consistency with the original input where appropriate.",
+      "clip": [
+        "18",
+        0
+      ],
+      "vae": [
+        "13",
+        0
+      ],
+      "configs": [
+        "2",
+        0
+      ]
+    },
+    "class_type": "TextEncodeQwenImageEditPlusCustom_lrzjason",
+    "_meta": {
+      "title": "TextEncodeQwenImageEditPlusCustom lrzjason"
+    }
+  },
+  "15": {
+    "inputs": {
+      "prompt": [
+        "31",
+        0
+      ],
+      "return_full_refs_cond": true,
+      "instruction": "Describe the key features of the input image (color, shape, size, texture, objects, background), then explain how the user's text instruction should alter or modify the image. Generate a new image that meets the user's requirements while maintaining consistency with the original input where appropriate.",
+      "clip": [
+        "18",
+        0
+      ],
+      "vae": [
+        "13",
+        0
+      ],
+      "configs": [
+        "2",
+        0
+      ]
+    },
+    "class_type": "TextEncodeQwenImageEditPlusCustom_lrzjason",
+    "_meta": {
+      "title": "TextEncodeQwenImageEditPlusCustom lrzjason"
+    }
+  },
+  "16": {
+    "inputs": {
+      "gguf_name": "FireRed-Image-Edit-1.1-transformer-q4_k_m.gguf"
+    },
+    "class_type": "LoaderGGUF",
+    "_meta": {
+      "title": "GGUF Loader"
+    }
+  },
+  "17": {
+    "inputs": {
+      "lora_name": "FireRed-Image-Edit-1.0-Lightning-8steps-v1.0.safetensors",
+      "strength_model": 1,
+      "model": [
+        "16",
+        0
+      ]
+    },
+    "class_type": "LoraLoaderModelOnly",
+    "_meta": {
+      "title": "LoRA加载器（仅模型）"
+    }
+  },
+  "18": {
+    "inputs": {
+      "clip_name": "qwen_2.5_vl_7b_fp8_scaled.safetensors",
+      "type": "qwen_image",
+      "device": "default"
+    },
+    "class_type": "ClipLoaderGGUF",
+    "_meta": {
+      "title": "GGUF CLIP Loader"
+    }
+  },
+  "19": {
+    "inputs": {
+      "image": [
+        "1",
+        0
+      ],
+      "pad_info": [
+        "7",
+        0
+      ]
+    },
+    "class_type": "CropWithPadInfo",
+    "_meta": {
+      "title": "Crop With Pad Info"
+    }
+  },
+  "21": {
+    "inputs": {
+      "samples1": [
+        "15",
+        1
+      ],
+      "samples2": [
+        "14",
+        1
+      ]
+    },
+    "class_type": "LatentAdd",
+    "_meta": {
+      "title": "Latent相加"
+    }
+  },
+  "22": {
+    "inputs": {
+      "seed": 814498275732948,
+      "steps": 8,
+      "cfg": 1,
+      "sampler_name": "euler",
+      "scheduler": "beta",
+      "denoise": 1,
+      "model": [
+        "17",
+        0
+      ],
+      "positive": [
+        "15",
+        0
+      ],
+      "negative": [
+        "14",
+        0
+      ],
+      "latent_image": [
+        "28",
+        0
+      ]
+    },
+    "class_type": "KSampler",
+    "_meta": {
+      "title": "K采样器"
+    }
+  },
+  "24": {
+    "inputs": {
+      "upscale_method": "lanczos",
+      "scale_by": [
+        "19",
+        1
+      ],
+      "image": [
+        "19",
+        0
+      ]
+    },
+    "class_type": "ImageScaleBy",
+    "_meta": {
+      "title": "缩放图像（比例）"
+    }
+  },
+  "28": {
+    "inputs": {
+      "upscale_method": "nearest-exact",
+      "width": [
+        "29",
+        0
+      ],
+      "height": [
+        "30",
+        0
+      ],
+      "crop": "disabled",
+      "samples": [
+        "21",
+        0
+      ]
+    },
+    "class_type": "LatentUpscale",
+    "_meta": {
+      "title": "缩放Latent"
+    }
+  },
+  "29": {
+    "inputs": {
+      "Number": "768"
+    },
+    "class_type": "Int",
+    "_meta": {
+      "title": "宽度"
+    }
+  },
+  "30": {
+    "inputs": {
+      "Number": "1024"
+    },
+    "class_type": "Int",
+    "_meta": {
+      "title": "高度"
+    }
+  },
+  "31": {
+    "inputs": {
+      "value": "图1的背景改成纯色深灰色，生成人物全身站立图。"
+    },
+    "class_type": "PrimitiveStringMultiline",
+    "_meta": {
+      "title": "字符串（多行）"
+    }
+  },
+  "32": {
+    "inputs": {
+      "upscale_method": "nearest-exact",
+      "width": [
+        "29",
+        0
+      ],
+      "height": [
+        "30",
+        0
+      ],
+      "crop": "center",
+      "image": [
+        "24",
+        0
+      ]
+    },
+    "class_type": "ImageScale",
+    "_meta": {
+      "title": "缩放图像"
+    }
+  },
+  "33": {
+    "inputs": {
+      "filename_prefix": "ComfyUI",
+      "images": [
+        "32",
+        0
+      ]
+    },
+    "class_type": "SaveImage",
+    "_meta": {
+      "title": "保存图像"
+    }
+  },
+  "36": {
+    "inputs": {
+      "rgthree_comparer": {
+        "images": [
+          {
+            "name": "A",
+            "selected": true,
+            "url": "/api/view?filename=rgthree.compare._temp_pfeet_00001_.png&type=temp&subfolder=&rand=0.983916958744797"
+          },
+          {
+            "name": "B",
+            "selected": true,
+            "url": "/api/view?filename=rgthree.compare._temp_pfeet_00002_.png&type=temp&subfolder=&rand=0.7859180062182538"
+          }
+        ]
+      },
+      "image_a": [
+        "32",
+        0
+      ],
+      "image_b": [
+        "11",
+        0
+      ]
+    },
+    "class_type": "Image Comparer (rgthree)",
+    "_meta": {
+      "title": "Image Comparer (rgthree)"
+    }
+  }
+}

change1/service1.py

@@ -0,0 +1,261 @@
+#!/usr/bin/env python3
+"""
+ComfyUI 单图编辑服务
+- 接受 1 个 base64 图片（模特/原图）
+- 上传到 ComfyUI，运行工作流 change1.json
+- 返回生成结果 OSS URL
+"""
+
+import asyncio
+import base64
+import io
+import json
+import os
+import tempfile
+import time
+import uuid
+from pathlib import Path
+
+import httpx
+import oss2
+from fastapi import FastAPI, HTTPException
+from fastapi.middleware.cors import CORSMiddleware
+from fastapi.responses import FileResponse
+from fastapi.staticfiles import StaticFiles
+from pydantic import BaseModel
+
+# COMFYUI_URL = "http://127.0.0.1:8188"
+COMFYUI_URL = "http://112.126.94.241:28188"
+COMFYUI_URL_BACKUP = "http://112.126.94.241:38188"
+WORKFLOW_PATH = Path(__file__).parent / "change1.json"
+
+NODE_MODEL = "11"   # 输入图（LoadImage）
+NODE_OUTPUT = "33"  # 输出（SaveImage）
+NODE_PROMPT = "31"  # 文本提示（PrimitiveStringMultiline）
+
+app = FastAPI(title="ComfyUI 单图编辑服务")
+
+app.add_middleware(
+    CORSMiddleware,
+    allow_origins=["*"],
+    allow_methods=["*"],
+    allow_headers=["*"],
+)
+
+
+def upload_to_oss(image_path, object_name=None):
+    access_key_id = 'LTAI5tGp1sLzedqxihcNC1eb'
+    access_key_secret = 'IFZE1b8YYreCP6zfA6GaZ9uBT678qO'
+    endpoint = 'oss-cn-beijing.aliyuncs.com'
+    bucket_name = 'xiangsilian'
+
+    auth = oss2.Auth(access_key_id, access_key_secret)
+    bucket = oss2.Bucket(auth, endpoint, bucket_name)
+
+    if object_name is None:
+        object_name = image_path.split('/')[-1]
+
+    try:
+        bucket.put_object_from_file(object_name, image_path)
+        url = bucket.sign_url('GET', object_name, 3600 * 24 * 365 * 10)
+        print(f"文件上传成功，URL: {url}")
+        return url
+    except Exception as e:
+        print(f"上传失败: {str(e)}")
+        return None
+
+
+class TryOnRequest(BaseModel):
+    model_image: str   # base64 输入图片
+    change_desc: str = ""  # 编辑描述，可空；非空时替换节点 31 的 inputs
+
+
+class TryOnResponse(BaseModel):
+    result_url: str  # OSS 图片 URL
+
+
+def decode_base64_image(b64_str: str) -> bytes:
+    """解码 base64 图片，自动处理 data:image/... 前缀"""
+    if "," in b64_str:
+        b64_str = b64_str.split(",", 1)[1]
+    return base64.b64decode(b64_str)
+
+
+async def check_comfyui_alive(base_url: str, timeout: float = 3.0) -> bool:
+    """检查指定的 ComfyUI 服务器是否可用"""
+    try:
+        async with httpx.AsyncClient(timeout=timeout) as client:
+            resp = await client.get(f"{base_url}/system_stats")
+            return resp.status_code == 200
+    except Exception as e:
+        print(f"ComfyUI 健康检查失败 {base_url}: {e}")
+        return False
+
+
+async def pick_comfyui_url() -> str:
+    """优先返回正式服务器，不可用时切换到备份服务器"""
+    if await check_comfyui_alive(COMFYUI_URL):
+        print(f"使用正式服务器: {COMFYUI_URL}")
+        return COMFYUI_URL
+    print(f"正式服务器不可用，尝试切换到备份服务器: {COMFYUI_URL_BACKUP}")
+    if await check_comfyui_alive(COMFYUI_URL_BACKUP):
+        print(f"使用备份服务器: {COMFYUI_URL_BACKUP}")
+        return COMFYUI_URL_BACKUP
+    raise HTTPException(status_code=503, detail="正式与备份 ComfyUI 服务器均不可用")
+
+
+async def upload_image(client: httpx.AsyncClient, base_url: str, image_bytes: bytes, filename: str) -> str:
+    """上传图片到 ComfyUI，返回服务器文件名"""
+    files = {
+        "image": (filename, io.BytesIO(image_bytes), "image/jpeg"),
+    }
+    data = {"overwrite": "true"}
+    resp = await client.post(f"{base_url}/upload/image", files=files, data=data)
+    resp.raise_for_status()
+    result = resp.json()
+    return result["name"]
+
+
+async def queue_prompt(client: httpx.AsyncClient, base_url: str, workflow: dict) -> str:
+    """提交工作流到队列，返回 prompt_id"""
+    payload = {"prompt": workflow, "client_id": str(uuid.uuid4())}
+    resp = await client.post(f"{base_url}/prompt", json=payload)
+    resp.raise_for_status()
+    return resp.json()["prompt_id"]
+
+
+async def wait_for_result(client: httpx.AsyncClient, base_url: str, prompt_id: str, timeout: int = 300) -> dict:
+    """轮询历史记录，等待任务完成，返回输出节点数据"""
+    deadline = time.time() + timeout
+    while time.time() < deadline:
+        resp = await client.get(f"{base_url}/history/{prompt_id}")
+        resp.raise_for_status()
+        history = resp.json()
+        if prompt_id in history:
+            entry = history[prompt_id]
+            status = entry.get("status", {})
+            if status.get("completed"):
+                return entry.get("outputs", {})
+            if status.get("status_str") == "error":
+                messages = status.get("messages", [])
+                raise HTTPException(status_code=500, detail=f"ComfyUI 工作流执行出错: {messages}")
+        await asyncio.sleep(2)
+    raise HTTPException(status_code=504, detail="等待 ComfyUI 超时（300秒）")
+
+
+async def fetch_image_bytes(client: httpx.AsyncClient, base_url: str, filename: str, subfolder: str = "", type_: str = "output") -> bytes:
+    """从 ComfyUI 下载图片，返回原始字节"""
+    params = {"filename": filename, "subfolder": subfolder, "type": type_}
+    resp = await client.get(f"{base_url}/view", params=params)
+    resp.raise_for_status()
+    return resp.content
+
+
+@app.post("/try-on", response_model=TryOnResponse)
+async def try_on(req: TryOnRequest):
+    """
+    单图编辑接口
+    - model_image: 输入图片 base64
+    - change_desc: 可选的编辑描述（非空时覆盖工作流节点 31）
+    返回 result_url: 生成结果的 OSS URL
+    """
+    print(f"接收到单图编辑请求，正在处理...")
+
+    comfyui_url = await pick_comfyui_url()
+
+    workflow = json.loads(WORKFLOW_PATH.read_text(encoding="utf-8"))
+
+    desc = (req.change_desc or "").strip()
+    if desc:
+        workflow[NODE_PROMPT]["inputs"] = {"value": desc}
+
+    async with httpx.AsyncClient(timeout=60.0) as client:
+        try:
+            model_bytes = decode_base64_image(req.model_image)
+        except Exception as e:
+            raise HTTPException(status_code=400, detail=f"图片解码失败: {e}")
+
+        uid = uuid.uuid4().hex[:8]
+        model_fname = f"model_{uid}.jpg"
+
+        try:
+            model_name = await upload_image(client, comfyui_url, model_bytes, model_fname)
+        except httpx.HTTPError as e:
+            raise HTTPException(status_code=502, detail=f"上传图片到 ComfyUI 失败: {e}")
+
+        workflow[NODE_MODEL]["inputs"]["image"] = model_name
+
+        try:
+            prompt_id = await queue_prompt(client, comfyui_url, workflow)
+        except httpx.HTTPError as e:
+            raise HTTPException(status_code=502, detail=f"提交工作流失败: {e}")
+
+        async with httpx.AsyncClient(timeout=30.0) as poll_client:
+            outputs = await wait_for_result(poll_client, comfyui_url, prompt_id, timeout=300)
+
+        node_output = outputs.get(NODE_OUTPUT)
+        if not node_output:
+            raise HTTPException(status_code=500, detail=f"工作流未返回节点 {NODE_OUTPUT} 的输出")
+
+        images = node_output.get("images", [])
+        if not images:
+            raise HTTPException(status_code=500, detail="输出节点没有图片")
+
+        img_info = images[0]
+        filename = img_info["filename"]
+        subfolder = img_info.get("subfolder", "")
+        type_ = img_info.get("type", "output")
+
+        async with httpx.AsyncClient(timeout=30.0) as dl_client:
+            result_bytes = await fetch_image_bytes(dl_client, comfyui_url, filename, subfolder, type_)
+
+    suffix = Path(filename).suffix or ".png"
+    object_name = f"tryon/{uuid.uuid4().hex}{suffix}"
+    with tempfile.NamedTemporaryFile(suffix=suffix, delete=False) as tmp:
+        tmp.write(result_bytes)
+        tmp_path = tmp.name
+
+    try:
+        result_url = upload_to_oss(tmp_path, object_name)
+    finally:
+        os.unlink(tmp_path)
+
+    if not result_url:
+        raise HTTPException(status_code=500, detail="上传结果图片到 OSS 失败")
+
+    return TryOnResponse(result_url=result_url)
+
+
+@app.get("/health")
+async def health():
+    """健康检查"""
+    primary_ok = await check_comfyui_alive(COMFYUI_URL)
+    backup_ok = await check_comfyui_alive(COMFYUI_URL_BACKUP)
+    if primary_ok:
+        active = COMFYUI_URL
+    elif backup_ok:
+        active = COMFYUI_URL_BACKUP
+    else:
+        active = None
+    return {
+        "status": "ok" if active else "degraded",
+        "comfyui_active": active,
+        "comfyui_primary": {"url": COMFYUI_URL, "alive": primary_ok},
+        "comfyui_backup": {"url": COMFYUI_URL_BACKUP, "alive": backup_ok},
+    }
+
+
+static_dir = Path(__file__).parent / "static"
+static_dir.mkdir(exist_ok=True)
+print(f"静态文件目录: {static_dir}")
+app.mount("/static", StaticFiles(directory=str(static_dir)), name="static")
+
+
+@app.get("/")
+async def index():
+    return FileResponse(str(static_dir / "index.html"))
+
+
+if __name__ == "__main__":
+    import uvicorn
+    uvicorn.run(app, host="0.0.0.0", port=12221, log_level="info")

change1/static/index.html

@@ -0,0 +1,390 @@
+<!DOCTYPE html>
+<html lang="zh-CN">
+<head>
+  <meta charset="UTF-8" />
+  <meta name="viewport" content="width=device-width, initial-scale=1.0" />
+  <title>AI 换装系统</title>
+  <style>
+    *, *::before, *::after { box-sizing: border-box; margin: 0; padding: 0; }
+
+    body {
+      font-family: "PingFang SC", "Microsoft YaHei", sans-serif;
+      background: #0f0f13;
+      color: #e0e0e0;
+      min-height: 100vh;
+      display: flex;
+      flex-direction: column;
+      align-items: center;
+      padding: 32px 16px 64px;
+    }
+
+    h1 {
+      font-size: 2rem;
+      font-weight: 700;
+      background: linear-gradient(135deg, #a78bfa, #60a5fa, #34d399);
+      -webkit-background-clip: text;
+      -webkit-text-fill-color: transparent;
+      margin-bottom: 8px;
+      text-align: center;
+    }
+
+    .subtitle {
+      font-size: 0.9rem;
+      color: #888;
+      margin-bottom: 40px;
+      text-align: center;
+    }
+
+    .upload-grid {
+      display: grid;
+      grid-template-columns: 1fr;
+      gap: 20px;
+      width: 100%;
+      max-width: 460px;
+      margin-bottom: 32px;
+    }
+
+    .upload-card {
+      background: #1a1a24;
+      border: 2px dashed #333;
+      border-radius: 16px;
+      padding: 20px;
+      display: flex;
+      flex-direction: column;
+      align-items: center;
+      gap: 12px;
+      cursor: pointer;
+      transition: border-color 0.2s, background 0.2s;
+      position: relative;
+      min-height: 280px;
+    }
+
+    .upload-card:hover {
+      border-color: #7c3aed;
+      background: #1e1e2e;
+    }
+
+    .upload-card.has-image {
+      border-style: solid;
+      border-color: #6d28d9;
+    }
+
+    .upload-card input[type="file"] {
+      position: absolute;
+      inset: 0;
+      opacity: 0;
+      cursor: pointer;
+      width: 100%;
+      height: 100%;
+    }
+
+    .upload-icon {
+      font-size: 2.5rem;
+      line-height: 1;
+    }
+
+    .upload-label {
+      font-size: 1rem;
+      font-weight: 600;
+      color: #c4b5fd;
+    }
+
+    .upload-hint {
+      font-size: 0.78rem;
+      color: #666;
+      text-align: center;
+    }
+
+    .preview-img {
+      width: 100%;
+      max-height: 200px;
+      object-fit: contain;
+      border-radius: 10px;
+      display: none;
+    }
+
+    .preview-img.visible {
+      display: block;
+    }
+
+    .btn-run {
+      padding: 14px 56px;
+      font-size: 1.05rem;
+      font-weight: 700;
+      border: none;
+      border-radius: 50px;
+      background: linear-gradient(135deg, #7c3aed, #2563eb);
+      color: #fff;
+      cursor: pointer;
+      transition: opacity 0.2s, transform 0.1s;
+      letter-spacing: 0.5px;
+    }
+
+    .btn-run:hover:not(:disabled) {
+      opacity: 0.9;
+      transform: translateY(-1px);
+    }
+
+    .btn-run:disabled {
+      opacity: 0.45;
+      cursor: not-allowed;
+    }
+
+    .status-bar {
+      margin-top: 20px;
+      font-size: 0.9rem;
+      color: #888;
+      min-height: 24px;
+      text-align: center;
+    }
+
+    .status-bar.running { color: #60a5fa; }
+    .status-bar.success { color: #34d399; }
+    .status-bar.error   { color: #f87171; }
+
+    .spinner {
+      display: inline-block;
+      width: 14px;
+      height: 14px;
+      border: 2px solid #60a5fa44;
+      border-top-color: #60a5fa;
+      border-radius: 50%;
+      animation: spin 0.8s linear infinite;
+      vertical-align: middle;
+      margin-right: 6px;
+    }
+
+    @keyframes spin { to { transform: rotate(360deg); } }
+
+    .result-section {
+      margin-top: 40px;
+      width: 100%;
+      max-width: 500px;
+      display: none;
+      flex-direction: column;
+      align-items: center;
+      gap: 16px;
+    }
+
+    .result-section.visible { display: flex; }
+
+    .result-title {
+      font-size: 1.1rem;
+      font-weight: 600;
+      color: #a78bfa;
+    }
+
+    .result-img {
+      width: 100%;
+      border-radius: 16px;
+      box-shadow: 0 0 40px #7c3aed44;
+    }
+
+    .btn-download {
+      padding: 10px 36px;
+      border: 2px solid #7c3aed;
+      border-radius: 50px;
+      background: transparent;
+      color: #a78bfa;
+      font-size: 0.9rem;
+      font-weight: 600;
+      cursor: pointer;
+      transition: background 0.2s;
+    }
+
+    .btn-download:hover {
+      background: #7c3aed22;
+    }
+
+    .prompt-panel {
+      width: 100%;
+      max-width: 460px;
+      margin-bottom: 24px;
+      background: #1a1a24;
+      border-radius: 16px;
+      padding: 20px;
+      border: 1px solid #2a2a3a;
+    }
+    .prompt-panel .prompt-title {
+      font-size: 0.85rem;
+      font-weight: 600;
+      color: #a78bfa;
+      margin-bottom: 12px;
+    }
+    .radio-row {
+      display: flex;
+      align-items: flex-start;
+      gap: 10px;
+      margin-bottom: 10px;
+      cursor: pointer;
+      font-size: 0.9rem;
+      color: #ccc;
+    }
+    .radio-row input { margin-top: 3px; accent-color: #7c3aed; flex-shrink: 0; }
+    .change-desc-input {
+      width: 100%;
+      margin-top: 4px;
+      padding: 12px;
+      border-radius: 10px;
+      border: 1px solid #333;
+      background: #12121a;
+      color: #e0e0e0;
+      font-family: inherit;
+      font-size: 0.88rem;
+      resize: vertical;
+      min-height: 88px;
+    }
+    .change-desc-input:disabled {
+      opacity: 0.45;
+      cursor: not-allowed;
+    }
+
+    @media (max-width: 640px) {
+      h1 { font-size: 1.5rem; }
+    }
+  </style>
+</head>
+<body>
+  <h1>AI 换装系统</h1>
+  <p class="subtitle">上传一张图片，AI 自动按提示词进行编辑</p>
+
+  <div class="upload-grid">
+    <!-- 模特/原图 -->
+    <div class="upload-card" id="card-model">
+      <input type="file" accept="image/*" id="input-model" />
+      <div class="upload-icon">🧍</div>
+      <div class="upload-label">模特图片</div>
+      <div class="upload-hint">节点 11 · 人物全身照</div>
+      <img class="preview-img" id="preview-model" alt="模特预览" />
+    </div>
+  </div>
+
+  <div class="prompt-panel" id="prompt-panel">
+    <div class="prompt-title">提示词（change_desc）</div>
+    <label class="radio-row">
+      <input type="radio" name="prompt_mode" id="prompt-default" value="default" checked />
+      <span>使用默认（传空，不覆盖工作流节点 31）</span>
+    </label>
+    <label class="radio-row">
+      <input type="radio" name="prompt_mode" id="prompt-custom" value="custom" />
+      <span>自定义提示词</span>
+    </label>
+    <textarea id="change-desc" class="change-desc-input" placeholder="输入编辑描述…" rows="4" disabled></textarea>
+  </div>
+
+  <button class="btn-run" id="btn-run" disabled>开始生成</button>
+  <div class="status-bar" id="status-bar"></div>
+
+  <div class="result-section" id="result-section">
+    <div class="result-title">生成结果</div>
+    <img class="result-img" id="result-img" alt="生成结果" />
+    <button class="btn-download" id="btn-download">下载图片</button>
+  </div>
+
+  <script>
+    const images = { model: null };
+
+    function setupUpload(inputId, previewId, cardId, key) {
+      const input = document.getElementById(inputId);
+      const preview = document.getElementById(previewId);
+      const card = document.getElementById(cardId);
+
+      input.addEventListener('change', (e) => {
+        const file = e.target.files[0];
+        if (!file) return;
+        const reader = new FileReader();
+        reader.onload = (ev) => {
+          images[key] = ev.target.result; // data:image/...;base64,...
+          preview.src = ev.target.result;
+          preview.classList.add('visible');
+          card.classList.add('has-image');
+          updateRunButton();
+        };
+        reader.readAsDataURL(file);
+      });
+    }
+
+    setupUpload('input-model', 'preview-model', 'card-model', 'model');
+
+    const promptDefault = document.getElementById('prompt-default');
+    const promptCustom = document.getElementById('prompt-custom');
+    const changeDescEl = document.getElementById('change-desc');
+    function syncPromptUi() {
+      changeDescEl.disabled = !promptCustom.checked;
+    }
+    promptDefault.addEventListener('change', syncPromptUi);
+    promptCustom.addEventListener('change', syncPromptUi);
+
+    function getChangeDesc() {
+      return promptCustom.checked ? changeDescEl.value.trim() : '';
+    }
+
+    function updateRunButton() {
+      const btn = document.getElementById('btn-run');
+      btn.disabled = !images.model;
+    }
+
+    function setStatus(msg, cls = '') {
+      const el = document.getElementById('status-bar');
+      el.className = 'status-bar ' + cls;
+      el.innerHTML = msg;
+    }
+
+    function elapsed(start) {
+      const s = Math.round((Date.now() - start) / 1000);
+      return `${s}秒`;
+    }
+
+    document.getElementById('btn-run').addEventListener('click', async () => {
+      const btn = document.getElementById('btn-run');
+      const resultSection = document.getElementById('result-section');
+      const resultImg = document.getElementById('result-img');
+
+      btn.disabled = true;
+      resultSection.classList.remove('visible');
+
+      const start = Date.now();
+      let timer = setInterval(() => {
+        setStatus(`<span class="spinner"></span>正在生成中，请稍候… 已等待 ${elapsed(start)}`, 'running');
+      }, 1000);
+      setStatus('<span class="spinner"></span>正在上传图片并提交工作流…', 'running');
+
+      try {
+        const resp = await fetch('/try-on', {
+          method: 'POST',
+          headers: { 'Content-Type': 'application/json' },
+          body: JSON.stringify({
+            model_image: images.model,
+            change_desc: getChangeDesc(),
+          }),
+        });
+
+        clearInterval(timer);
+
+        if (!resp.ok) {
+          const err = await resp.json().catch(() => ({ detail: resp.statusText }));
+          throw new Error(err.detail || '请求失败');
+        }
+
+        const data = await resp.json();
+        resultImg.src = data.result_url;
+        resultSection.classList.add('visible');
+        setStatus(`✅ 生成完成！耗时 ${elapsed(start)}`, 'success');
+
+        document.getElementById('btn-download').onclick = () => {
+          const a = document.createElement('a');
+          a.href = data.result_url;
+          a.download = `tryon_${Date.now()}.png`;
+          a.target = '_blank';
+          a.click();
+        };
+      } catch (err) {
+        clearInterval(timer);
+        setStatus(`❌ 错误：${err.message}`, 'error');
+      } finally {
+        updateRunButton();
+      }
+    });
+  </script>
+</body>
+</html>

change_app.py

@@ -26,8 +26,8 @@ def change_cloth_base64():
     human_img = data.get('human_img')
     cloth_img = data.get('cloth_img')
 
-    if not human_img or not cloth_img:
+    if not human_img:
-        return jsonify({"ret": -1, "state": -1, "msg": "Both human_img and cloth_img are required"}), 400
+        return jsonify({"ret": -1, "state": -1, "msg": "human_img is required"}), 400
 
     kuzi_img = data.get('kuzi_img')
 
@@ -46,13 +46,19 @@ def change_cloth_base64():
                 'change_desc': change_desc,
             }
             response = requests.post('http://127.0.0.1:12223/try-on', json=payload, timeout=360)
-        else:
+        elif cloth_img:
             payload = {
                 'model_image': human_img,
                 'shirt_image': cloth_img,
                 'change_desc': change_desc,
             }
             response = requests.post('http://127.0.0.1:12222/try-on', json=payload, timeout=360)
+        else:
+            payload = {
+                'model_image': human_img,
+                'change_desc': change_desc,
+            }
+            response = requests.post('http://127.0.0.1:12221/try-on', json=payload, timeout=360)
 
         response.raise_for_status()
         result = response.json()

start_services.sh

@@ -1,6 +1,6 @@
 #!/bin/bash
-# 启动本机换装网关与 service2 / service3。
+# 启动本机换装网关与 service1 / service2 / service3。
-# ComfyUI 在其它机器运行；service2/service3 内已配置远端 COMFYUI_URL。
+# ComfyUI 在其它机器运行；service1/service2/service3 内已配置远端 COMFYUI_URL。
 
 ROOT="/home/ubuntu/change_cloth_server"
 LOG_DIR="/home/ubuntu/log"
@@ -9,6 +9,7 @@ mkdir -p "$LOG_DIR"
 echo "[$(date '+%Y-%m-%d %H:%M:%S')] 开始停止旧进程..."
 
 # 按命令行中的脚本绝对路径结束旧实例（与下方 nohup 路径一致）
+pkill -f "${ROOT}/change1/service1.py" 2>/dev/null && echo "  已停止 service1" || echo "  service1 未在运行"
 pkill -f "${ROOT}/change2/service2.py" 2>/dev/null && echo "  已停止 service2" || echo "  service2 未在运行"
 pkill -f "${ROOT}/change3/service3.py" 2>/dev/null && echo "  已停止 service3" || echo "  service3 未在运行"
 pkill -f "${ROOT}/change_app.py" 2>/dev/null && echo "  已停止 change_app" || echo "  change_app 未在运行"
@@ -19,6 +20,9 @@ echo "[$(date '+%Y-%m-%d %H:%M:%S')] 开始启动服务..."
 
 export PATH="/home/ubuntu/.local/bin:$PATH"
 
+nohup python3 "${ROOT}/change1/service1.py" >> "$LOG_DIR/service1.log" 2>&1 &
+echo "  service1 已启动 (PID: $!)"
+
 nohup python3 "${ROOT}/change2/service2.py" >> "$LOG_DIR/service2.log" 2>&1 &
 echo "  service2 已启动 (PID: $!)"
 

static/index.html

@@ -249,7 +249,7 @@
 </head>
 <body>
   <h1>AI 换装测试</h1>
-  <p class="subtitle">上传模特和衣服图片，调用 /change_cloth_base64 接口测试</p>
+  <p class="subtitle">上传图片调用 /change_cloth_base64 接口；只传模特→单图编辑，加上衣→换装，再加裤子→换装+换裤</p>
 
   <div class="upload-grid">
     <!-- 模特 -->
@@ -262,13 +262,14 @@
       <img class="preview-img" id="preview-human" alt="模特预览" />
     </div>
 
-    <!-- 上衣 -->
+    <!-- 上衣（可选） -->
-    <div class="upload-card" id="card-cloth">
+    <div class="upload-card optional" id="card-cloth">
       <button class="btn-clear" id="clear-cloth" title="清除">✕</button>
       <input type="file" accept="image/*" id="input-cloth" />
       <div class="upload-icon">👕</div>
       <div class="upload-label">上衣图片</div>
-      <div class="upload-hint">cloth_img · 必填</div>
+      <div class="upload-hint">cloth_img · 可选</div>
+      <span class="badge-optional">无则走 service1，有则走 service2</span>
       <img class="preview-img" id="preview-cloth" alt="上衣预览" />
     </div>
 
@@ -279,7 +280,7 @@
       <div class="upload-icon">👖</div>
       <div class="upload-label">裤子图片</div>
       <div class="upload-hint">kuzi_img · 可选</div>
-      <span class="badge-optional">有则走 8888 服务，无则走 8889</span>
+      <span class="badge-optional">填了则走 service3</span>
       <img class="preview-img" id="preview-kuzi" alt="裤子预览" />
     </div>
   </div>
@@ -365,7 +366,7 @@
     }
 
     function updateRunButton() {
-      document.getElementById('btn-run').disabled = !(images.human && images.cloth);
+      document.getElementById('btn-run').disabled = !images.human;
     }
 
     function setStatus(msg, cls = '') {
@@ -397,9 +398,9 @@
 
       const body = {
         human_img: images.human,
-        cloth_img: images.cloth,
         change_desc: getChangeDesc(),
       };
+      if (images.cloth) body.cloth_img = images.cloth;
       if (images.kuzi)  body.kuzi_img  = images.kuzi;
 
       try {