initial commit

python_server/.gitignore

@@ -0,0 +1 @@
+outputs/

python_server/__init__.py

@@ -0,0 +1 @@
+

python_server/config.py

@@ -0,0 +1,223 @@
+"""
+python_server 的统一配置文件。
+
+特点：
+- 使用 Python 而不是 YAML，方便在代码中集中列举所有可用模型，供前端读取。
+- 后端加载模型时，也从这里读取默认值，保证单一信息源。
+"""
+
+from __future__ import annotations
+
+from dataclasses import dataclass, field
+from typing import Literal, TypedDict, List
+
+from model.Depth.zoe_loader import ZoeModelName
+
+
+# -----------------------------
+# 1. 深度模型枚举（给前端展示用）
+# -----------------------------
+
+
+class DepthModelInfo(TypedDict):
+    id: str            # 唯一 ID，如 "zoedepth_n"
+    family: str        # 模型家族，如 "ZoeDepth"
+    name: str          # 展示名，如 "ZoeD_N (NYU+KITT)"
+    description: str   # 简短描述
+    backend: str       # 后端类型，如 "zoedepth", "depth_anything_v2", "midas", "dpt"
+
+
+DEPTH_MODELS: List[DepthModelInfo] = [
+    # ZoeDepth 系列
+    {
+        "id": "zoedepth_n",
+        "family": "ZoeDepth",
+        "name": "ZoeD_N",
+        "description": "ZoeDepth zero-shot 模型，适用室内/室外通用场景。",
+        "backend": "zoedepth",
+    },
+    {
+        "id": "zoedepth_k",
+        "family": "ZoeDepth",
+        "name": "ZoeD_K",
+        "description": "ZoeDepth Kitti 专用版本，针对户外驾驶场景优化。",
+        "backend": "zoedepth",
+    },
+    {
+        "id": "zoedepth_nk",
+        "family": "ZoeDepth",
+        "name": "ZoeD_NK",
+        "description": "ZoeDepth 双头版本（NYU+KITTI），综合室内/室外场景。",
+        "backend": "zoedepth",
+    },
+    # 预留：Depth Anything v2
+    {
+        "id": "depth_anything_v2_s",
+        "family": "Depth Anything V2",
+        "name": "Depth Anything V2 Small",
+        "description": "轻量级 Depth Anything V2 小模型。",
+        "backend": "depth_anything_v2",
+    },
+    # 预留：MiDaS
+    {
+        "id": "midas_dpt_large",
+        "family": "MiDaS",
+        "name": "MiDaS DPT Large",
+        "description": "MiDaS DPT-Large 高质量深度模型。",
+        "backend": "midas",
+    },
+    # 预留：DPT
+    {
+        "id": "dpt_large",
+        "family": "DPT",
+        "name": "DPT Large",
+        "description": "DPT Large 单目深度估计模型。",
+        "backend": "dpt",
+    },
+]
+
+# -----------------------------
+# 1.2 补全模型枚举（给前端展示用）
+# -----------------------------
+
+
+class InpaintModelInfo(TypedDict):
+    id: str
+    family: str
+    name: str
+    description: str
+    backend: str  # "sdxl_inpaint" | "controlnet"
+
+
+INPAINT_MODELS: List[InpaintModelInfo] = [
+    {
+        "id": "sdxl_inpaint",
+        "family": "SDXL",
+        "name": "SDXL Inpainting",
+        "description": "基于 diffusers 的 SDXL 补全管线（需要 prompt + mask）。",
+        "backend": "sdxl_inpaint",
+    },
+    {
+        "id": "controlnet",
+        "family": "ControlNet",
+        "name": "ControlNet (placeholder)",
+        "description": "ControlNet 补全/控制生成（当前统一封装暂未实现）。",
+        "backend": "controlnet",
+    },
+]
+
+
+# -----------------------------
+# 1.3 动画模型枚举（给前端展示用）
+# -----------------------------
+
+
+class AnimationModelInfo(TypedDict):
+    id: str
+    family: str
+    name: str
+    description: str
+    backend: str  # "animatediff"
+
+
+ANIMATION_MODELS: List[AnimationModelInfo] = [
+    {
+        "id": "animatediff",
+        "family": "AnimateDiff",
+        "name": "AnimateDiff (Text-to-Video)",
+        "description": "基于 AnimateDiff 的文生动画能力，输出 GIF 动画。",
+        "backend": "animatediff",
+    },
+]
+
+
+# -----------------------------
+# 2. 后端默认配置（给服务端用）
+# -----------------------------
+
+
+@dataclass
+class DepthConfig:
+    # 深度后端选择：前端不参与选择；只允许在后端配置中切换
+    backend: Literal["zoedepth", "depth_anything_v2", "dpt", "midas"] = "zoedepth"
+    # ZoeDepth 家族默认选择
+    zoe_model: ZoeModelName = "ZoeD_N"
+    # Depth Anything V2 默认 encoder
+    da_v2_encoder: Literal["vits", "vitb", "vitl", "vitg"] = "vitl"
+    # DPT 默认模型类型
+    dpt_model_type: Literal["dpt_large", "dpt_hybrid"] = "dpt_large"
+    # MiDaS 默认模型类型
+    midas_model_type: Literal[
+        "dpt_beit_large_512",
+        "dpt_swin2_large_384",
+        "dpt_swin2_tiny_256",
+        "dpt_levit_224",
+    ] = "dpt_beit_large_512"
+    # 统一的默认运行设备
+    device: str = "cuda"
+
+
+@dataclass
+class InpaintConfig:
+    # 统一补全默认后端
+    backend: Literal["sdxl_inpaint", "controlnet"] = "sdxl_inpaint"
+    # SDXL Inpaint 的基础模型（可写 HuggingFace model id 或本地目录）
+    sdxl_base_model: str = "stabilityai/stable-diffusion-xl-base-1.0"
+    # ControlNet Inpaint 基础模型与 controlnet 权重
+    controlnet_base_model: str = "runwayml/stable-diffusion-inpainting"
+    controlnet_model: str = "lllyasviel/control_v11p_sd15_inpaint"
+    device: str = "cuda"
+
+
+@dataclass
+class AnimationConfig:
+    # 统一动画默认后端
+    backend: Literal["animatediff"] = "animatediff"
+    # AnimateDiff 根目录（相对 python_server/ 或绝对路径）
+    animate_diff_root: str = "model/Animation/AnimateDiff"
+    # 文生图基础模型（HuggingFace model id 或本地目录）
+    pretrained_model_path: str = "runwayml/stable-diffusion-v1-5"
+    # AnimateDiff 推理配置
+    inference_config: str = "configs/inference/inference-v3.yaml"
+    # 运动模块与个性化底模（为空则由脚本按默认处理）
+    motion_module: str = "v3_sd15_mm.ckpt"
+    dreambooth_model: str = "realisticVisionV60B1_v51VAE.safetensors"
+    lora_model: str = ""
+    lora_alpha: float = 0.8
+    # 部分环境 xformers 兼容性差，可手动关闭
+    without_xformers: bool = False
+    device: str = "cuda"
+
+
+@dataclass
+class AppConfig:
+    # 使用 default_factory 避免 dataclass 的可变默认值问题
+    depth: DepthConfig = field(default_factory=DepthConfig)
+    inpaint: InpaintConfig = field(default_factory=InpaintConfig)
+    animation: AnimationConfig = field(default_factory=AnimationConfig)
+
+
+# 后端代码直接 import DEFAULT_CONFIG 即可
+DEFAULT_CONFIG = AppConfig()
+
+
+def list_depth_models() -> List[DepthModelInfo]:
+    """
+    返回所有可用深度模型的元信息，方便前端通过 /models 等接口读取。
+    """
+    return DEPTH_MODELS
+
+
+def list_inpaint_models() -> List[InpaintModelInfo]:
+    """
+    返回所有可用补全模型的元信息，方便前端通过 /models 等接口读取。
+    """
+    return INPAINT_MODELS
+
+
+def list_animation_models() -> List[AnimationModelInfo]:
+    """
+    返回所有可用动画模型的元信息，方便前端通过 /models 等接口读取。
+    """
+    return ANIMATION_MODELS
+

python_server/config_loader.py

@@ -0,0 +1,153 @@
+"""
+兼容层：从 Python 配置模块中构造 zoe_loader 需要的 ZoeConfig。
+
+后端其它代码尽量只依赖这里的函数，而不直接依赖 config.py 的具体结构，
+便于以后扩展。
+"""
+
+from model.Depth.zoe_loader import ZoeConfig
+from model.Depth.depth_anything_v2_loader import DepthAnythingV2Config
+from model.Depth.dpt_loader import DPTConfig
+from model.Depth.midas_loader import MiDaSConfig
+from config import AppConfig, DEFAULT_CONFIG
+
+
+def load_app_config() -> AppConfig:
+    """
+    当前直接返回 DEFAULT_CONFIG。
+    如未来需要支持多环境 / 覆盖配置，可以在这里加逻辑。
+    """
+    return DEFAULT_CONFIG
+
+
+def build_zoe_config_from_app(app_cfg: AppConfig | None = None) -> ZoeConfig:
+    """
+    将 AppConfig.depth 映射为 ZoeConfig，供 zoe_loader 使用。
+    如果未显式传入 app_cfg，则使用全局 DEFAULT_CONFIG。
+    """
+    if app_cfg is None:
+        app_cfg = load_app_config()
+
+    return ZoeConfig(
+        model=app_cfg.depth.zoe_model,
+        device=app_cfg.depth.device,
+    )
+
+
+def build_depth_anything_v2_config_from_app(
+    app_cfg: AppConfig | None = None,
+) -> DepthAnythingV2Config:
+    """
+    将 AppConfig.depth 映射为 DepthAnythingV2Config。
+    """
+    if app_cfg is None:
+        app_cfg = load_app_config()
+
+    return DepthAnythingV2Config(
+        encoder=app_cfg.depth.da_v2_encoder,
+        device=app_cfg.depth.device,
+    )
+
+
+def build_dpt_config_from_app(app_cfg: AppConfig | None = None) -> DPTConfig:
+    if app_cfg is None:
+        app_cfg = load_app_config()
+    return DPTConfig(
+        model_type=app_cfg.depth.dpt_model_type,
+        device=app_cfg.depth.device,
+    )
+
+
+def build_midas_config_from_app(app_cfg: AppConfig | None = None) -> MiDaSConfig:
+    if app_cfg is None:
+        app_cfg = load_app_config()
+    return MiDaSConfig(
+        model_type=app_cfg.depth.midas_model_type,
+        device=app_cfg.depth.device,
+    )
+
+
+def get_depth_backend_from_app(app_cfg: AppConfig | None = None) -> str:
+    if app_cfg is None:
+        app_cfg = load_app_config()
+    return app_cfg.depth.backend
+
+
+def get_inpaint_backend_from_app(app_cfg: AppConfig | None = None) -> str:
+    if app_cfg is None:
+        app_cfg = load_app_config()
+    return app_cfg.inpaint.backend
+
+
+def get_sdxl_base_model_from_app(app_cfg: AppConfig | None = None) -> str:
+    if app_cfg is None:
+        app_cfg = load_app_config()
+    return app_cfg.inpaint.sdxl_base_model
+
+
+def get_controlnet_base_model_from_app(app_cfg: AppConfig | None = None) -> str:
+    if app_cfg is None:
+        app_cfg = load_app_config()
+    return app_cfg.inpaint.controlnet_base_model
+
+
+def get_controlnet_model_from_app(app_cfg: AppConfig | None = None) -> str:
+    if app_cfg is None:
+        app_cfg = load_app_config()
+    return app_cfg.inpaint.controlnet_model
+
+
+def get_animation_backend_from_app(app_cfg: AppConfig | None = None) -> str:
+    if app_cfg is None:
+        app_cfg = load_app_config()
+    return app_cfg.animation.backend
+
+
+def get_animatediff_root_from_app(app_cfg: AppConfig | None = None) -> str:
+    if app_cfg is None:
+        app_cfg = load_app_config()
+    return app_cfg.animation.animate_diff_root
+
+
+def get_animatediff_pretrained_model_from_app(app_cfg: AppConfig | None = None) -> str:
+    if app_cfg is None:
+        app_cfg = load_app_config()
+    return app_cfg.animation.pretrained_model_path
+
+
+def get_animatediff_inference_config_from_app(app_cfg: AppConfig | None = None) -> str:
+    if app_cfg is None:
+        app_cfg = load_app_config()
+    return app_cfg.animation.inference_config
+
+
+def get_animatediff_motion_module_from_app(app_cfg: AppConfig | None = None) -> str:
+    if app_cfg is None:
+        app_cfg = load_app_config()
+    return app_cfg.animation.motion_module
+
+
+def get_animatediff_dreambooth_model_from_app(app_cfg: AppConfig | None = None) -> str:
+    if app_cfg is None:
+        app_cfg = load_app_config()
+    return app_cfg.animation.dreambooth_model
+
+
+def get_animatediff_lora_model_from_app(app_cfg: AppConfig | None = None) -> str:
+    if app_cfg is None:
+        app_cfg = load_app_config()
+    return app_cfg.animation.lora_model
+
+
+def get_animatediff_lora_alpha_from_app(app_cfg: AppConfig | None = None) -> float:
+    if app_cfg is None:
+        app_cfg = load_app_config()
+    return app_cfg.animation.lora_alpha
+
+
+def get_animatediff_without_xformers_from_app(app_cfg: AppConfig | None = None) -> bool:
+    if app_cfg is None:
+        app_cfg = load_app_config()
+    return app_cfg.animation.without_xformers
+
+

python_server/model/Animation/__init__.py

@@ -0,0 +1,12 @@
+from .animation_loader import (
+    AnimationBackend,
+    UnifiedAnimationConfig,
+    build_animation_predictor,
+)
+
+__all__ = [
+    "AnimationBackend",
+    "UnifiedAnimationConfig",
+    "build_animation_predictor",
+]
+

python_server/model/Animation/animation_loader.py

@@ -0,0 +1,268 @@
+from __future__ import annotations
+
+"""
+Unified animation model loading entry.
+
+Current support:
+- AnimateDiff (script-based invocation)
+"""
+
+from dataclasses import dataclass
+from enum import Enum
+from pathlib import Path
+import json
+import os
+import subprocess
+import sys
+import tempfile
+from typing import Callable
+
+from config_loader import (
+    load_app_config,
+    get_animatediff_root_from_app,
+    get_animatediff_pretrained_model_from_app,
+    get_animatediff_inference_config_from_app,
+    get_animatediff_motion_module_from_app,
+    get_animatediff_dreambooth_model_from_app,
+    get_animatediff_lora_model_from_app,
+    get_animatediff_lora_alpha_from_app,
+    get_animatediff_without_xformers_from_app,
+)
+
+
+class AnimationBackend(str, Enum):
+    ANIMATEDIFF = "animatediff"
+
+
+@dataclass
+class UnifiedAnimationConfig:
+    backend: AnimationBackend = AnimationBackend.ANIMATEDIFF
+    # Optional overrides. If None, values come from app config.
+    animate_diff_root: str | None = None
+    pretrained_model_path: str | None = None
+    inference_config: str | None = None
+    motion_module: str | None = None
+    dreambooth_model: str | None = None
+    lora_model: str | None = None
+    lora_alpha: float | None = None
+    without_xformers: bool | None = None
+    controlnet_path: str | None = None
+    controlnet_config: str | None = None
+
+
+def _yaml_string(value: str) -> str:
+    return json.dumps(value, ensure_ascii=False)
+
+
+def _resolve_root(root_cfg: str) -> Path:
+    root = Path(root_cfg)
+    if not root.is_absolute():
+        root = Path(__file__).resolve().parents[2] / root_cfg
+    return root.resolve()
+
+
+def _make_animatediff_predictor(
+    cfg: UnifiedAnimationConfig,
+) -> Callable[..., Path]:
+    app_cfg = load_app_config()
+
+    root = _resolve_root(cfg.animate_diff_root or get_animatediff_root_from_app(app_cfg))
+    script_path = root / "scripts" / "animate.py"
+    samples_dir = root / "samples"
+    samples_dir.mkdir(parents=True, exist_ok=True)
+
+    if not script_path.is_file():
+        raise FileNotFoundError(f"AnimateDiff script not found: {script_path}")
+
+    pretrained_model_path = (
+        cfg.pretrained_model_path or get_animatediff_pretrained_model_from_app(app_cfg)
+    )
+    inference_config = cfg.inference_config or get_animatediff_inference_config_from_app(app_cfg)
+    motion_module = cfg.motion_module or get_animatediff_motion_module_from_app(app_cfg)
+    dreambooth_model = cfg.dreambooth_model
+    if dreambooth_model is None:
+        dreambooth_model = get_animatediff_dreambooth_model_from_app(app_cfg)
+    lora_model = cfg.lora_model
+    if lora_model is None:
+        lora_model = get_animatediff_lora_model_from_app(app_cfg)
+    lora_alpha = cfg.lora_alpha
+    if lora_alpha is None:
+        lora_alpha = get_animatediff_lora_alpha_from_app(app_cfg)
+    without_xformers = cfg.without_xformers
+    if without_xformers is None:
+        without_xformers = get_animatediff_without_xformers_from_app(app_cfg)
+
+    def _predict(
+        prompt: str,
+        negative_prompt: str = "",
+        num_inference_steps: int = 25,
+        guidance_scale: float = 8.0,
+        width: int = 512,
+        height: int = 512,
+        video_length: int = 16,
+        seed: int = -1,
+        control_image_path: str | None = None,
+        output_format: str = "gif",
+    ) -> Path:
+        if output_format not in {"gif", "png_sequence"}:
+            raise ValueError("output_format must be 'gif' or 'png_sequence'")
+        prompt_value = prompt.strip()
+        if not prompt_value:
+            raise ValueError("prompt must not be empty")
+
+        negative_prompt_value = negative_prompt or ""
+
+        motion_module_line = (
+            f'  motion_module: {_yaml_string(motion_module)}\n' if motion_module else ""
+        )
+        dreambooth_line = (
+            f'  dreambooth_path: {_yaml_string(dreambooth_model)}\n' if dreambooth_model else ""
+        )
+        lora_path_line = f'  lora_model_path: {_yaml_string(lora_model)}\n' if lora_model else ""
+        lora_alpha_line = f"  lora_alpha: {float(lora_alpha)}\n" if lora_model else ""
+        controlnet_path_value = cfg.controlnet_path or "v3_sd15_sparsectrl_rgb.ckpt"
+        controlnet_config_value = cfg.controlnet_config or "configs/inference/sparsectrl/image_condition.yaml"
+        control_image_line = ""
+        if control_image_path:
+            control_image = Path(control_image_path).expanduser().resolve()
+            if not control_image.is_file():
+                raise FileNotFoundError(f"control_image_path not found: {control_image}")
+            control_image_line = (
+                f'  controlnet_path: {_yaml_string(controlnet_path_value)}\n'
+                f'  controlnet_config: {_yaml_string(controlnet_config_value)}\n'
+                "  controlnet_images:\n"
+                f'  - {_yaml_string(str(control_image))}\n'
+                "  controlnet_image_indexs:\n"
+                "  - 0\n"
+            )
+
+        config_text = (
+            "- prompt:\n"
+            f"  - {_yaml_string(prompt_value)}\n"
+            "  n_prompt:\n"
+            f"  - {_yaml_string(negative_prompt_value)}\n"
+            f"  steps: {int(num_inference_steps)}\n"
+            f"  guidance_scale: {float(guidance_scale)}\n"
+            f"  W: {int(width)}\n"
+            f"  H: {int(height)}\n"
+            f"  L: {int(video_length)}\n"
+            "  seed:\n"
+            f"  - {int(seed)}\n"
+            f"{motion_module_line}{dreambooth_line}{lora_path_line}{lora_alpha_line}{control_image_line}"
+        )
+
+        before_dirs = {p for p in samples_dir.iterdir() if p.is_dir()}
+        cfg_file = tempfile.NamedTemporaryFile(
+            mode="w",
+            suffix=".yaml",
+            prefix="animatediff_cfg_",
+            dir=str(root),
+            delete=False,
+            encoding="utf-8",
+        )
+        cfg_file_path = Path(cfg_file.name)
+        try:
+            cfg_file.write(config_text)
+            cfg_file.flush()
+            cfg_file.close()
+
+            cmd = [
+                sys.executable,
+                str(script_path),
+                "--pretrained-model-path",
+                pretrained_model_path,
+                "--inference-config",
+                inference_config,
+                "--config",
+                str(cfg_file_path),
+                "--L",
+                str(int(video_length)),
+                "--W",
+                str(int(width)),
+                "--H",
+                str(int(height)),
+            ]
+            if without_xformers:
+                cmd.append("--without-xformers")
+            if output_format == "png_sequence":
+                cmd.append("--save-png-sequence")
+
+            env = dict(os.environ)
+            existing_pythonpath = env.get("PYTHONPATH", "")
+            root_pythonpath = str(root)
+            env["PYTHONPATH"] = (
+                f"{root_pythonpath}:{existing_pythonpath}" if existing_pythonpath else root_pythonpath
+            )
+
+            def _run_once(command: list[str]) -> subprocess.CompletedProcess[str]:
+                return subprocess.run(
+                    command,
+                    cwd=str(root),
+                    check=True,
+                    capture_output=True,
+                    text=True,
+                    env=env,
+                )
+
+            try:
+                proc = _run_once(cmd)
+            except subprocess.CalledProcessError as first_error:
+                stderr_text = first_error.stderr or ""
+                should_retry_without_xformers = (
+                    not without_xformers
+                    and "--without-xformers" not in cmd
+                    and (
+                        "memory_efficient_attention" in stderr_text
+                        or "AcceleratorError" in stderr_text
+                        or "invalid configuration argument" in stderr_text
+                    )
+                )
+                if not should_retry_without_xformers:
+                    raise
+
+                retry_cmd = [*cmd, "--without-xformers"]
+                proc = _run_once(retry_cmd)
+            _ = proc
+        except subprocess.CalledProcessError as e:
+            raise RuntimeError(
+                "AnimateDiff inference failed.\n"
+                f"stdout:\n{e.stdout}\n"
+                f"stderr:\n{e.stderr}"
+            ) from e
+        finally:
+            try:
+                cfg_file_path.unlink(missing_ok=True)
+            except Exception:
+                pass
+
+        after_dirs = [p for p in samples_dir.iterdir() if p.is_dir() and p not in before_dirs]
+        candidates = [p for p in after_dirs if (p / "sample.gif").is_file()]
+        if not candidates:
+            candidates = [p for p in samples_dir.iterdir() if p.is_dir() and (p / "sample.gif").is_file()]
+        if not candidates:
+            raise FileNotFoundError("AnimateDiff finished but sample.gif was not found in samples/")
+
+        latest = sorted(candidates, key=lambda p: p.stat().st_mtime, reverse=True)[0]
+        if output_format == "png_sequence":
+            frames_root = latest / "sample_frames"
+            if not frames_root.is_dir():
+                raise FileNotFoundError("AnimateDiff finished but sample_frames/ was not found in samples/")
+            frame_dirs = sorted([p for p in frames_root.iterdir() if p.is_dir()], key=lambda p: p.stat().st_mtime, reverse=True)
+            if not frame_dirs:
+                raise FileNotFoundError("AnimateDiff finished but no PNG sequence directory was found in sample_frames/")
+            return frame_dirs[0].resolve()
+        return (latest / "sample.gif").resolve()
+
+    return _predict
+
+
+def build_animation_predictor(
+    cfg: UnifiedAnimationConfig | None = None,
+) -> tuple[Callable[..., Path], AnimationBackend]:
+    cfg = cfg or UnifiedAnimationConfig()
+
+    if cfg.backend == AnimationBackend.ANIMATEDIFF:
+        return _make_animatediff_predictor(cfg), AnimationBackend.ANIMATEDIFF
+
+    raise ValueError(f"Unsupported animation backend: {cfg.backend}")
+

python_server/model/Depth/__init__.py

@@ -0,0 +1 @@
+

python_server/model/Depth/depth_anything_v2_loader.py

@@ -0,0 +1,147 @@
+from dataclasses import dataclass
+from typing import Literal, Tuple
+
+import sys
+from pathlib import Path
+
+import numpy as np
+import torch
+import requests
+
+# 确保本地克隆的 Depth Anything V2 仓库在 sys.path 中，
+# 这样其内部的 `from depth_anything_v2...` 导入才能正常工作。
+_THIS_DIR = Path(__file__).resolve().parent
+_DA_REPO_ROOT = _THIS_DIR / "Depth-Anything-V2"
+if _DA_REPO_ROOT.is_dir():
+    da_path = str(_DA_REPO_ROOT)
+    if da_path not in sys.path:
+        sys.path.insert(0, da_path)
+
+from depth_anything_v2.dpt import DepthAnythingV2  # type: ignore[import]
+
+
+EncoderName = Literal["vits", "vitb", "vitl", "vitg"]
+
+
+@dataclass
+class DepthAnythingV2Config:
+    """
+    Depth Anything V2 模型选择配置。
+
+    encoder: "vits" | "vitb" | "vitl" | "vitg"
+    device: "cuda" | "cpu"
+    input_size: 推理时的输入分辨率（短边），参考官方 demo，默认 518。
+    """
+
+    encoder: EncoderName = "vitl"
+    device: str = "cuda"
+    input_size: int = 518
+
+
+_MODEL_CONFIGS = {
+    "vits": {"encoder": "vits", "features": 64, "out_channels": [48, 96, 192, 384]},
+    "vitb": {"encoder": "vitb", "features": 128, "out_channels": [96, 192, 384, 768]},
+    "vitl": {"encoder": "vitl", "features": 256, "out_channels": [256, 512, 1024, 1024]},
+    "vitg": {"encoder": "vitg", "features": 384, "out_channels": [1536, 1536, 1536, 1536]},
+}
+
+
+_DA_V2_WEIGHTS_URLS = {
+    # 官方权重托管在 HuggingFace:
+    # - Small  -> vits
+    # - Base   -> vitb
+    # - Large  -> vitl
+    # - Giant  -> vitg
+    # 如需替换为国内镜像，可直接修改这些 URL。
+    "vits": "https://huggingface.co/depth-anything/Depth-Anything-V2-Small/resolve/main/depth_anything_v2_vits.pth",
+    "vitb": "https://huggingface.co/depth-anything/Depth-Anything-V2-Base/resolve/main/depth_anything_v2_vitb.pth",
+    "vitl": "https://huggingface.co/depth-anything/Depth-Anything-V2-Large/resolve/main/depth_anything_v2_vitl.pth",
+    "vitg": "https://huggingface.co/depth-anything/Depth-Anything-V2-Giant/resolve/main/depth_anything_v2_vitg.pth",
+}
+
+
+def _download_if_missing(encoder: str, ckpt_path: Path) -> None:
+    if ckpt_path.is_file():
+        return
+
+    url = _DA_V2_WEIGHTS_URLS.get(encoder)
+    if not url:
+        raise FileNotFoundError(
+            f"找不到权重文件: {ckpt_path}\n"
+            f"且当前未为 encoder='{encoder}' 配置自动下载 URL，请手动下载到该路径。"
+        )
+
+    ckpt_path.parent.mkdir(parents=True, exist_ok=True)
+    print(f"自动下载 Depth Anything V2 权重 ({encoder}):\n  {url}\n  -> {ckpt_path}")
+
+    resp = requests.get(url, stream=True)
+    resp.raise_for_status()
+
+    total = int(resp.headers.get("content-length", "0") or "0")
+    downloaded = 0
+    chunk_size = 1024 * 1024
+
+    with ckpt_path.open("wb") as f:
+        for chunk in resp.iter_content(chunk_size=chunk_size):
+            if not chunk:
+                continue
+            f.write(chunk)
+            downloaded += len(chunk)
+            if total > 0:
+                done = int(50 * downloaded / total)
+                print("\r[{}{}] {:.1f}%".format("#" * done, "." * (50 - done), downloaded * 100 / total), end="")
+
+    print("\n权重下载完成。")
+
+
+def load_depth_anything_v2_from_config(
+    cfg: DepthAnythingV2Config,
+) -> Tuple[DepthAnythingV2, DepthAnythingV2Config]:
+    """
+    根据配置加载 Depth Anything V2 模型与对应配置。
+
+    说明：
+    - 权重文件路径遵循官方命名约定：
+      checkpoints/depth_anything_v2_{encoder}.pth
+      例如：depth_anything_v2_vitl.pth
+    - 请确保上述权重文件已下载到
+      python_server/model/Depth/Depth-Anything-V2/checkpoints 下。
+    """
+    if cfg.encoder not in _MODEL_CONFIGS:
+        raise ValueError(f"不支持的 encoder: {cfg.encoder}")
+
+    ckpt_path = _DA_REPO_ROOT / "checkpoints" / f"depth_anything_v2_{cfg.encoder}.pth"
+    _download_if_missing(cfg.encoder, ckpt_path)
+
+    model = DepthAnythingV2(**_MODEL_CONFIGS[cfg.encoder])
+
+    state_dict = torch.load(str(ckpt_path), map_location="cpu")
+    model.load_state_dict(state_dict)
+
+    if cfg.device.startswith("cuda") and torch.cuda.is_available():
+        device = "cuda"
+    else:
+        device = "cpu"
+
+    model = model.to(device).eval()
+    cfg = DepthAnythingV2Config(
+        encoder=cfg.encoder,
+        device=device,
+        input_size=cfg.input_size,
+    )
+    return model, cfg
+
+
+def infer_depth_anything_v2(
+    model: DepthAnythingV2,
+    image_bgr: np.ndarray,
+    input_size: int,
+) -> np.ndarray:
+    """
+    对单张 BGR 图像做深度推理，返回 float32 深度图（未归一化）。
+    image_bgr: OpenCV 读取的 BGR 图像 (H, W, 3), uint8
+    """
+    depth = model.infer_image(image_bgr, input_size)
+    depth = np.asarray(depth, dtype="float32")
+    return depth
+

python_server/model/Depth/depth_loader.py

@@ -0,0 +1,148 @@
+from __future__ import annotations
+
+"""
+统一的深度模型加载入口。
+
+当前支持：
+- ZoeDepth（三种：ZoeD_N / ZoeD_K / ZoeD_NK）
+- Depth Anything V2（四种 encoder：vits / vitb / vitl / vitg）
+
+未来如果要加 MiDaS / DPT，只需要在这里再接一层即可。
+"""
+
+from dataclasses import dataclass
+from enum import Enum
+from typing import Callable
+
+import numpy as np
+from PIL import Image
+
+from config_loader import (
+    load_app_config,
+    build_zoe_config_from_app,
+    build_depth_anything_v2_config_from_app,
+    build_dpt_config_from_app,
+    build_midas_config_from_app,
+)
+from .zoe_loader import load_zoe_from_config
+from .depth_anything_v2_loader import (
+    load_depth_anything_v2_from_config,
+    infer_depth_anything_v2,
+)
+from .dpt_loader import load_dpt_from_config, infer_dpt
+from .midas_loader import load_midas_from_config, infer_midas
+
+
+class DepthBackend(str, Enum):
+    """统一的深度模型后端类型。"""
+
+    ZOEDEPTH = "zoedepth"
+    DEPTH_ANYTHING_V2 = "depth_anything_v2"
+    DPT = "dpt"
+    MIDAS = "midas"
+
+
+@dataclass
+class UnifiedDepthConfig:
+    """
+    统一深度配置。
+
+    backend: 使用哪个后端
+    device: 强制设备（可选），不填则使用 config.py 中的设置
+    """
+
+    backend: DepthBackend = DepthBackend.ZOEDEPTH
+    device: str | None = None
+
+
+def _make_zoe_predictor(device_override: str | None = None) -> Callable[[Image.Image], np.ndarray]:
+    app_cfg = load_app_config()
+    zoe_cfg = build_zoe_config_from_app(app_cfg)
+    if device_override is not None:
+        zoe_cfg.device = device_override
+
+    model, _ = load_zoe_from_config(zoe_cfg)
+
+    def _predict(img: Image.Image) -> np.ndarray:
+        depth = model.infer_pil(img.convert("RGB"), output_type="numpy")
+        return np.asarray(depth, dtype="float32").squeeze()
+
+    return _predict
+
+
+def _make_da_v2_predictor(device_override: str | None = None) -> Callable[[Image.Image], np.ndarray]:
+    app_cfg = load_app_config()
+    da_cfg = build_depth_anything_v2_config_from_app(app_cfg)
+    if device_override is not None:
+        da_cfg.device = device_override
+
+    model, da_cfg = load_depth_anything_v2_from_config(da_cfg)
+
+    def _predict(img: Image.Image) -> np.ndarray:
+        # Depth Anything V2 的 infer_image 接收 BGR uint8
+        rgb = np.array(img.convert("RGB"), dtype=np.uint8)
+        bgr = rgb[:, :, ::-1]
+        depth = infer_depth_anything_v2(model, bgr, da_cfg.input_size)
+        return depth.astype("float32").squeeze()
+
+    return _predict
+
+
+def _make_dpt_predictor(device_override: str | None = None) -> Callable[[Image.Image], np.ndarray]:
+    app_cfg = load_app_config()
+    dpt_cfg = build_dpt_config_from_app(app_cfg)
+    if device_override is not None:
+        dpt_cfg.device = device_override
+
+    model, dpt_cfg, transform = load_dpt_from_config(dpt_cfg)
+
+    def _predict(img: Image.Image) -> np.ndarray:
+        bgr = cv2.cvtColor(np.array(img.convert("RGB"), dtype=np.uint8), cv2.COLOR_RGB2BGR)
+        depth = infer_dpt(model, transform, bgr, dpt_cfg.device)
+        return depth.astype("float32").squeeze()
+
+    return _predict
+
+
+def _make_midas_predictor(device_override: str | None = None) -> Callable[[Image.Image], np.ndarray]:
+    app_cfg = load_app_config()
+    midas_cfg = build_midas_config_from_app(app_cfg)
+    if device_override is not None:
+        midas_cfg.device = device_override
+
+    model, midas_cfg, transform, net_w, net_h = load_midas_from_config(midas_cfg)
+
+    def _predict(img: Image.Image) -> np.ndarray:
+        rgb = np.array(img.convert("RGB"), dtype=np.float32) / 255.0
+        depth = infer_midas(model, transform, rgb, net_w, net_h, midas_cfg.device)
+        return depth.astype("float32").squeeze()
+
+    return _predict
+
+
+def build_depth_predictor(
+    cfg: UnifiedDepthConfig | None = None,
+) -> tuple[Callable[[Image.Image], np.ndarray], DepthBackend]:
+    """
+    统一构建深度预测函数。
+
+    返回：
+    - predictor(image: PIL.Image) -> np.ndarray[H, W], float32
+    - 实际使用的 backend 类型
+    """
+    cfg = cfg or UnifiedDepthConfig()
+
+    if cfg.backend == DepthBackend.ZOEDEPTH:
+        return _make_zoe_predictor(cfg.device), DepthBackend.ZOEDEPTH
+
+    if cfg.backend == DepthBackend.DEPTH_ANYTHING_V2:
+        return _make_da_v2_predictor(cfg.device), DepthBackend.DEPTH_ANYTHING_V2
+
+    if cfg.backend == DepthBackend.DPT:
+        return _make_dpt_predictor(cfg.device), DepthBackend.DPT
+
+    if cfg.backend == DepthBackend.MIDAS:
+        return _make_midas_predictor(cfg.device), DepthBackend.MIDAS
+
+    raise ValueError(f"不支持的深度后端: {cfg.backend}")
+

python_server/model/Depth/dpt_loader.py

@@ -0,0 +1,156 @@
+from dataclasses import dataclass
+from typing import Literal, Tuple
+
+import sys
+from pathlib import Path
+
+import numpy as np
+import torch
+import requests
+
+_THIS_DIR = Path(__file__).resolve().parent
+_DPT_REPO_ROOT = _THIS_DIR / "DPT"
+if _DPT_REPO_ROOT.is_dir():
+    dpt_path = str(_DPT_REPO_ROOT)
+    if dpt_path not in sys.path:
+        sys.path.insert(0, dpt_path)
+
+from dpt.models import DPTDepthModel  # type: ignore[import]
+from dpt.transforms import Resize, NormalizeImage, PrepareForNet  # type: ignore[import]
+from torchvision.transforms import Compose
+import cv2
+
+
+DPTModelType = Literal["dpt_large", "dpt_hybrid"]
+
+
+@dataclass
+class DPTConfig:
+    model_type: DPTModelType = "dpt_large"
+    device: str = "cuda"
+
+
+_DPT_WEIGHTS_URLS = {
+    # 官方 DPT 模型权重托管在：
+    # https://github.com/isl-org/DPT#models
+    "dpt_large": "https://github.com/isl-org/DPT/releases/download/1_0/dpt_large-midas-2f21e586.pt",
+    "dpt_hybrid": "https://github.com/isl-org/DPT/releases/download/1_0/dpt_hybrid-midas-501f0c75.pt",
+}
+
+
+def _download_if_missing(model_type: str, ckpt_path: Path) -> None:
+    if ckpt_path.is_file():
+        return
+
+    url = _DPT_WEIGHTS_URLS.get(model_type)
+    if not url:
+        raise FileNotFoundError(
+            f"找不到 DPT 权重文件: {ckpt_path}\n"
+            f"且当前未为 model_type='{model_type}' 配置自动下载 URL，请手动下载到该路径。"
+        )
+
+    ckpt_path.parent.mkdir(parents=True, exist_ok=True)
+    print(f"自动下载 DPT 权重 ({model_type}):\n  {url}\n  -> {ckpt_path}")
+
+    resp = requests.get(url, stream=True)
+    resp.raise_for_status()
+
+    total = int(resp.headers.get("content-length", "0") or "0")
+    downloaded = 0
+    chunk_size = 1024 * 1024
+
+    with ckpt_path.open("wb") as f:
+        for chunk in resp.iter_content(chunk_size=chunk_size):
+            if not chunk:
+                continue
+            f.write(chunk)
+            downloaded += len(chunk)
+            if total > 0:
+                done = int(50 * downloaded / total)
+                print("\r[{}{}] {:.1f}%".format("#" * done, "." * (50 - done), downloaded * 100 / total), end="")
+    print("\nDPT 权重下载完成。")
+
+
+def load_dpt_from_config(cfg: DPTConfig) -> Tuple[DPTDepthModel, DPTConfig, Compose]:
+    """
+    加载 DPT 模型与对应的预处理 transform。
+    """
+    ckpt_name = {
+        "dpt_large": "dpt_large-midas-2f21e586.pt",
+        "dpt_hybrid": "dpt_hybrid-midas-501f0c75.pt",
+    }[cfg.model_type]
+
+    ckpt_path = _DPT_REPO_ROOT / "weights" / ckpt_name
+    _download_if_missing(cfg.model_type, ckpt_path)
+
+    if cfg.model_type == "dpt_large":
+        net_w = net_h = 384
+        model = DPTDepthModel(
+            path=str(ckpt_path),
+            backbone="vitl16_384",
+            non_negative=True,
+            enable_attention_hooks=False,
+        )
+        normalization = NormalizeImage(mean=[0.5, 0.5, 0.5], std=[0.5, 0.5, 0.5])
+    else:
+        net_w = net_h = 384
+        model = DPTDepthModel(
+            path=str(ckpt_path),
+            backbone="vitb_rn50_384",
+            non_negative=True,
+            enable_attention_hooks=False,
+        )
+        normalization = NormalizeImage(mean=[0.5, 0.5, 0.5], std=[0.5, 0.5, 0.5])
+
+    device = "cuda" if cfg.device.startswith("cuda") and torch.cuda.is_available() else "cpu"
+    model.to(device).eval()
+
+    transform = Compose(
+        [
+            Resize(
+                net_w,
+                net_h,
+                resize_target=None,
+                keep_aspect_ratio=True,
+                ensure_multiple_of=32,
+                resize_method="minimal",
+                image_interpolation_method=cv2.INTER_CUBIC,
+            ),
+            normalization,
+            PrepareForNet(),
+        ]
+    )
+
+    cfg = DPTConfig(model_type=cfg.model_type, device=device)
+    return model, cfg, transform
+
+
+def infer_dpt(
+    model: DPTDepthModel,
+    transform: Compose,
+    image_bgr: np.ndarray,
+    device: str,
+) -> np.ndarray:
+    """
+    对单张 BGR 图像做深度推理，返回 float32 深度图（未归一化）。
+    """
+    img = cv2.cvtColor(image_bgr, cv2.COLOR_BGR2RGB)
+    sample = transform({"image": img})["image"]
+
+    input_batch = torch.from_numpy(sample).to(device).unsqueeze(0)
+    with torch.no_grad():
+        prediction = model(input_batch)
+        prediction = (
+            torch.nn.functional.interpolate(
+                prediction.unsqueeze(1),
+                size=img.shape[:2],
+                mode="bicubic",
+                align_corners=False,
+            )
+            .squeeze()
+            .cpu()
+            .numpy()
+        )
+
+    return prediction.astype("float32")
+

python_server/model/Depth/midas_loader.py

@@ -0,0 +1,127 @@
+from dataclasses import dataclass
+from typing import Literal, Tuple
+
+import sys
+from pathlib import Path
+
+import numpy as np
+import torch
+import requests
+
+_THIS_DIR = Path(__file__).resolve().parent
+_MIDAS_REPO_ROOT = _THIS_DIR / "MiDaS"
+if _MIDAS_REPO_ROOT.is_dir():
+    midas_path = str(_MIDAS_REPO_ROOT)
+    if midas_path not in sys.path:
+        sys.path.insert(0, midas_path)
+
+from midas.model_loader import load_model, default_models  # type: ignore[import]
+import utils  # from MiDaS repo
+
+
+MiDaSModelType = Literal[
+    "dpt_beit_large_512",
+    "dpt_swin2_large_384",
+    "dpt_swin2_tiny_256",
+    "dpt_levit_224",
+]
+
+
+@dataclass
+class MiDaSConfig:
+    model_type: MiDaSModelType = "dpt_beit_large_512"
+    device: str = "cuda"
+
+
+_MIDAS_WEIGHTS_URLS = {
+    # 官方权重参见 MiDaS 仓库 README
+    "dpt_beit_large_512": "https://github.com/isl-org/MiDaS/releases/download/v3_1/dpt_beit_large_512.pt",
+    "dpt_swin2_large_384": "https://github.com/isl-org/MiDaS/releases/download/v3_1/dpt_swin2_large_384.pt",
+    "dpt_swin2_tiny_256": "https://github.com/isl-org/MiDaS/releases/download/v3_1/dpt_swin2_tiny_256.pt",
+    "dpt_levit_224": "https://github.com/isl-org/MiDaS/releases/download/v3_1/dpt_levit_224.pt",
+}
+
+
+def _download_if_missing(model_type: str, ckpt_path: Path) -> None:
+    if ckpt_path.is_file():
+        return
+
+    url = _MIDAS_WEIGHTS_URLS.get(model_type)
+    if not url:
+        raise FileNotFoundError(
+            f"找不到 MiDaS 权重文件: {ckpt_path}\n"
+            f"且当前未为 model_type='{model_type}' 配置自动下载 URL，请手动下载到该路径。"
+        )
+
+    ckpt_path.parent.mkdir(parents=True, exist_ok=True)
+    print(f"自动下载 MiDaS 权重 ({model_type}):\n  {url}\n  -> {ckpt_path}")
+
+    resp = requests.get(url, stream=True)
+    resp.raise_for_status()
+
+    total = int(resp.headers.get("content-length", "0") or "0")
+    downloaded = 0
+    chunk_size = 1024 * 1024
+
+    with ckpt_path.open("wb") as f:
+        for chunk in resp.iter_content(chunk_size=chunk_size):
+            if not chunk:
+                continue
+            f.write(chunk)
+            downloaded += len(chunk)
+            if total > 0:
+                done = int(50 * downloaded / total)
+                print("\r[{}{}] {:.1f}%".format("#" * done, "." * (50 - done), downloaded * 100 / total), end="")
+    print("\nMiDaS 权重下载完成。")
+
+
+def load_midas_from_config(
+    cfg: MiDaSConfig,
+) -> Tuple[torch.nn.Module, MiDaSConfig, callable, int, int]:
+    """
+    加载 MiDaS 模型与对应 transform。
+    返回: model, cfg, transform, net_w, net_h
+    """
+    # default_models 中给了默认权重路径名
+    model_info = default_models[cfg.model_type]
+    ckpt_path = _MIDAS_REPO_ROOT / model_info.path
+    _download_if_missing(cfg.model_type, ckpt_path)
+
+    device = "cuda" if cfg.device.startswith("cuda") and torch.cuda.is_available() else "cpu"
+    model, transform, net_w, net_h = load_model(
+        device=torch.device(device),
+        model_path=str(ckpt_path),
+        model_type=cfg.model_type,
+        optimize=False,
+        height=None,
+        square=False,
+    )
+
+    cfg = MiDaSConfig(model_type=cfg.model_type, device=device)
+    return model, cfg, transform, net_w, net_h
+
+
+def infer_midas(
+    model: torch.nn.Module,
+    transform: callable,
+    image_rgb: np.ndarray,
+    net_w: int,
+    net_h: int,
+    device: str,
+) -> np.ndarray:
+    """
+    对单张 RGB 图像做深度推理，返回 float32 深度图（未归一化）。
+    """
+    image = transform({"image": image_rgb})["image"]
+    prediction = utils.process(
+        torch.device(device),
+        model,
+        model_type="dpt",  # 这里具体字符串对 utils.process 的逻辑影响不大，只要不包含 "openvino"
+        image=image,
+        input_size=(net_w, net_h),
+        target_size=image_rgb.shape[1::-1],
+        optimize=False,
+        use_camera=False,
+    )
+    return np.asarray(prediction, dtype="float32").squeeze()
+

python_server/model/Depth/zoe_loader.py

@@ -0,0 +1,74 @@
+from dataclasses import dataclass
+from typing import Literal
+
+import sys
+from pathlib import Path
+
+import torch
+
+# 确保本地克隆的 ZoeDepth 仓库在 sys.path 中，
+# 这样其内部的 `import zoedepth...` 才能正常工作。
+_THIS_DIR = Path(__file__).resolve().parent
+_ZOE_REPO_ROOT = _THIS_DIR / "ZoeDepth"
+if _ZOE_REPO_ROOT.is_dir():
+    zoe_path = str(_ZOE_REPO_ROOT)
+    if zoe_path not in sys.path:
+        sys.path.insert(0, zoe_path)
+
+from zoedepth.models.builder import build_model
+from zoedepth.utils.config import get_config
+
+
+ZoeModelName = Literal["ZoeD_N", "ZoeD_K", "ZoeD_NK"]
+
+
+@dataclass
+class ZoeConfig:
+    """
+    ZoeDepth 模型选择配置。
+
+    model: "ZoeD_N" | "ZoeD_K" | "ZoeD_NK"
+    device: "cuda" | "cpu"
+    """
+
+    model: ZoeModelName = "ZoeD_N"
+    device: str = "cuda"
+
+
+def load_zoe_from_name(name: ZoeModelName, device: str = "cuda"):
+    """
+    手动加载 ZoeDepth 三种模型之一：
+    - "ZoeD_N"
+    - "ZoeD_K"
+    - "ZoeD_NK"
+    """
+    if name == "ZoeD_N":
+        conf = get_config("zoedepth", "infer")
+    elif name == "ZoeD_K":
+        conf = get_config("zoedepth", "infer", config_version="kitti")
+    elif name == "ZoeD_NK":
+        conf = get_config("zoedepth_nk", "infer")
+    else:
+        raise ValueError(f"不支持的 ZoeDepth 模型名称: {name}")
+
+    model = build_model(conf)
+
+    if device.startswith("cuda") and torch.cuda.is_available():
+        model = model.to("cuda")
+    else:
+        model = model.to("cpu")
+
+    model.eval()
+    return model, conf
+
+
+def load_zoe_from_config(config: ZoeConfig):
+    """
+    根据 ZoeConfig 加载模型。
+
+    示例：
+        cfg = ZoeConfig(model="ZoeD_NK", device="cuda")
+        model, conf = load_zoe_from_config(cfg)
+    """
+    return load_zoe_from_name(config.model, config.device)
+

python_server/model/Inpaint/__init__.py

@@ -0,0 +1 @@
+

python_server/model/Inpaint/inpaint_loader.py

@@ -0,0 +1,413 @@
+from __future__ import annotations
+
+"""
+统一的补全（Inpaint）模型加载入口。
+
+当前支持：
+- SDXL Inpaint（diffusers AutoPipelineForInpainting）
+- ControlNet（占位，暂未统一封装）
+"""
+
+from dataclasses import dataclass
+from enum import Enum
+from typing import Callable
+
+import numpy as np
+from PIL import Image
+
+from config_loader import (
+    load_app_config,
+    get_sdxl_base_model_from_app,
+    get_controlnet_base_model_from_app,
+    get_controlnet_model_from_app,
+)
+
+
+class InpaintBackend(str, Enum):
+    SDXL_INPAINT = "sdxl_inpaint"
+    CONTROLNET = "controlnet"
+
+
+@dataclass
+class UnifiedInpaintConfig:
+    backend: InpaintBackend = InpaintBackend.SDXL_INPAINT
+    device: str | None = None
+    # SDXL base model (HF id 或本地目录)，不填则用 config.py 的默认值
+    sdxl_base_model: str | None = None
+
+
+@dataclass
+class UnifiedDrawConfig:
+    """
+    统一绘图配置：
+    - 纯文生图：image=None
+    - 图生图（模仿输入图）：image=某张参考图
+    """
+    device: str | None = None
+    sdxl_base_model: str | None = None
+
+
+def _resolve_device_and_dtype(device: str | None):
+    import torch
+
+    app_cfg = load_app_config()
+    if device is None:
+        device = app_cfg.inpaint.device
+    device = "cuda" if device.startswith("cuda") and torch.cuda.is_available() else "cpu"
+    torch_dtype = torch.float16 if device == "cuda" else torch.float32
+    return device, torch_dtype
+
+
+def _enable_memory_opts(pipe, device: str) -> None:
+    if device == "cuda":
+        try:
+            pipe.enable_attention_slicing()
+        except Exception:
+            pass
+        try:
+            pipe.enable_vae_slicing()
+        except Exception:
+            pass
+        try:
+            pipe.enable_vae_tiling()
+        except Exception:
+            pass
+        try:
+            pipe.enable_model_cpu_offload()
+        except Exception:
+            pass
+
+
+def _align_size(orig_w: int, orig_h: int, max_side: int) -> tuple[int, int]:
+    run_w, run_h = orig_w, orig_h
+    if max_side > 0 and max(orig_w, orig_h) > max_side:
+        scale = max_side / float(max(orig_w, orig_h))
+        run_w = int(round(orig_w * scale))
+        run_h = int(round(orig_h * scale))
+    run_w = max(8, run_w - (run_w % 8))
+    run_h = max(8, run_h - (run_h % 8))
+    return run_w, run_h
+
+
+def _make_sdxl_inpaint_predictor(
+    cfg: UnifiedInpaintConfig,
+) -> Callable[[Image.Image, Image.Image, str, str], Image.Image]:
+    """
+    返回补全函数：
+    - 输入：image(PIL RGB), mask(PIL L/1), prompt, negative_prompt
+    - 输出：PIL RGB 结果图
+    """
+    import torch
+    from diffusers import AutoPipelineForText2Image, AutoPipelineForInpainting
+
+    app_cfg = load_app_config()
+    base_model = cfg.sdxl_base_model or get_sdxl_base_model_from_app(app_cfg)
+
+    device = cfg.device
+    if device is None:
+        device = app_cfg.inpaint.device
+    device = "cuda" if device.startswith("cuda") and torch.cuda.is_available() else "cpu"
+
+    torch_dtype = torch.float16 if device == "cuda" else torch.float32
+
+    pipe_t2i = AutoPipelineForText2Image.from_pretrained(
+        base_model,
+        torch_dtype=torch_dtype,
+        variant="fp16" if device == "cuda" else None,
+        use_safetensors=True,
+    ).to(device)
+    pipe = AutoPipelineForInpainting.from_pipe(pipe_t2i).to(device)
+
+    # 省显存设置（尽量不改变输出语义）
+    # 注意：CPU offload 会明显变慢，但能显著降低显存占用。
+    if device == "cuda":
+        try:
+            pipe.enable_attention_slicing()
+        except Exception:
+            pass
+        try:
+            pipe.enable_vae_slicing()
+        except Exception:
+            pass
+        try:
+            pipe.enable_vae_tiling()
+        except Exception:
+            pass
+        try:
+            pipe.enable_model_cpu_offload()
+        except Exception:
+            pass
+
+    def _predict(
+        image: Image.Image,
+        mask: Image.Image,
+        prompt: str,
+        negative_prompt: str = "",
+        strength: float = 0.8,
+        guidance_scale: float = 7.5,
+        num_inference_steps: int = 30,
+        max_side: int = 1024,
+    ) -> Image.Image:
+        image = image.convert("RGB")
+        # diffusers 要求 mask 为单通道，白色区域为需要重绘
+        mask = mask.convert("L")
+
+        # SDXL / diffusers 通常要求宽高为 8 的倍数；同时为了避免 OOM，
+        # 推理时将图像按比例缩放到不超过 max_side（默认 1024）并对齐到 8 的倍数。
+        # 推理后再 resize 回原始尺寸，保证输出与原图分辨率一致。
+        orig_w, orig_h = image.size
+        run_w, run_h = orig_w, orig_h
+
+        if max(orig_w, orig_h) > max_side:
+            scale = max_side / float(max(orig_w, orig_h))
+            run_w = int(round(orig_w * scale))
+            run_h = int(round(orig_h * scale))
+
+        run_w = max(8, run_w - (run_w % 8))
+        run_h = max(8, run_h - (run_h % 8))
+        if run_w <= 0:
+            run_w = 8
+        if run_h <= 0:
+            run_h = 8
+
+        if (run_w, run_h) != (orig_w, orig_h):
+            image_run = image.resize((run_w, run_h), resample=Image.BICUBIC)
+            mask_run = mask.resize((run_w, run_h), resample=Image.NEAREST)
+        else:
+            image_run = image
+            mask_run = mask
+
+        if device == "cuda":
+            try:
+                torch.cuda.empty_cache()
+            except Exception:
+                pass
+
+        out = pipe(
+            prompt=prompt,
+            negative_prompt=negative_prompt,
+            image=image_run,
+            mask_image=mask_run,
+            strength=strength,
+            guidance_scale=guidance_scale,
+            num_inference_steps=num_inference_steps,
+            width=run_w,
+            height=run_h,
+        ).images[0]
+        out = out.convert("RGB")
+        if out.size != (orig_w, orig_h):
+            out = out.resize((orig_w, orig_h), resample=Image.BICUBIC)
+        return out
+
+    return _predict
+
+
+def _make_controlnet_predictor(_: UnifiedInpaintConfig):
+    import torch
+    from diffusers import ControlNetModel, StableDiffusionControlNetInpaintPipeline
+
+    app_cfg = load_app_config()
+
+    device = _.device
+    if device is None:
+        device = app_cfg.inpaint.device
+    device = "cuda" if device.startswith("cuda") and torch.cuda.is_available() else "cpu"
+    torch_dtype = torch.float16 if device == "cuda" else torch.float32
+
+    base_model = get_controlnet_base_model_from_app(app_cfg)
+    controlnet_id = get_controlnet_model_from_app(app_cfg)
+
+    controlnet = ControlNetModel.from_pretrained(controlnet_id, torch_dtype=torch_dtype)
+    pipe = StableDiffusionControlNetInpaintPipeline.from_pretrained(
+        base_model,
+        controlnet=controlnet,
+        torch_dtype=torch_dtype,
+        safety_checker=None,
+    )
+
+    if device == "cuda":
+        try:
+            pipe.enable_attention_slicing()
+        except Exception:
+            pass
+        try:
+            pipe.enable_vae_slicing()
+        except Exception:
+            pass
+        try:
+            pipe.enable_vae_tiling()
+        except Exception:
+            pass
+        try:
+            pipe.enable_model_cpu_offload()
+        except Exception:
+            pipe.to(device)
+    else:
+        pipe.to(device)
+
+    def _predict(
+        image: Image.Image,
+        mask: Image.Image,
+        prompt: str,
+        negative_prompt: str = "",
+        strength: float = 0.8,
+        guidance_scale: float = 7.5,
+        num_inference_steps: int = 30,
+        controlnet_conditioning_scale: float = 1.0,
+        max_side: int = 768,
+    ) -> Image.Image:
+        import cv2
+        import numpy as np
+
+        image = image.convert("RGB")
+        mask = mask.convert("L")
+
+        orig_w, orig_h = image.size
+        run_w, run_h = orig_w, orig_h
+        if max(orig_w, orig_h) > max_side:
+            scale = max_side / float(max(orig_w, orig_h))
+            run_w = int(round(orig_w * scale))
+            run_h = int(round(orig_h * scale))
+        run_w = max(8, run_w - (run_w % 8))
+        run_h = max(8, run_h - (run_h % 8))
+
+        if (run_w, run_h) != (orig_w, orig_h):
+            image_run = image.resize((run_w, run_h), resample=Image.BICUBIC)
+            mask_run = mask.resize((run_w, run_h), resample=Image.NEAREST)
+        else:
+            image_run = image
+            mask_run = mask
+
+        # control image：使用 canny 边缘作为约束（最通用）
+        rgb = np.array(image_run, dtype=np.uint8)
+        edges = cv2.Canny(rgb, 100, 200)
+        edges3 = np.stack([edges, edges, edges], axis=-1)
+        control_image = Image.fromarray(edges3)
+
+        if device == "cuda":
+            try:
+                torch.cuda.empty_cache()
+            except Exception:
+                pass
+
+        out = pipe(
+            prompt=prompt,
+            negative_prompt=negative_prompt,
+            image=image_run,
+            mask_image=mask_run,
+            control_image=control_image,
+            strength=strength,
+            guidance_scale=guidance_scale,
+            num_inference_steps=num_inference_steps,
+            controlnet_conditioning_scale=controlnet_conditioning_scale,
+            width=run_w,
+            height=run_h,
+        ).images[0]
+
+        out = out.convert("RGB")
+        if out.size != (orig_w, orig_h):
+            out = out.resize((orig_w, orig_h), resample=Image.BICUBIC)
+        return out
+
+    return _predict
+
+
+def build_inpaint_predictor(
+    cfg: UnifiedInpaintConfig | None = None,
+) -> tuple[Callable[..., Image.Image], InpaintBackend]:
+    """
+    统一构建补全预测函数。
+    """
+    cfg = cfg or UnifiedInpaintConfig()
+
+    if cfg.backend == InpaintBackend.SDXL_INPAINT:
+        return _make_sdxl_inpaint_predictor(cfg), InpaintBackend.SDXL_INPAINT
+
+    if cfg.backend == InpaintBackend.CONTROLNET:
+        return _make_controlnet_predictor(cfg), InpaintBackend.CONTROLNET
+
+    raise ValueError(f"不支持的补全后端: {cfg.backend}")
+
+
+def build_draw_predictor(
+    cfg: UnifiedDrawConfig | None = None,
+) -> Callable[..., Image.Image]:
+    """
+    构建统一绘图函数：
+    - 文生图：draw(prompt, image=None, ...)
+    - 图生图：draw(prompt, image=ref_image, strength=0.55, ...)
+    """
+    import torch
+    from diffusers import AutoPipelineForText2Image, AutoPipelineForImage2Image
+
+    cfg = cfg or UnifiedDrawConfig()
+    app_cfg = load_app_config()
+    base_model = cfg.sdxl_base_model or get_sdxl_base_model_from_app(app_cfg)
+    device, torch_dtype = _resolve_device_and_dtype(cfg.device)
+
+    pipe_t2i = AutoPipelineForText2Image.from_pretrained(
+        base_model,
+        torch_dtype=torch_dtype,
+        variant="fp16" if device == "cuda" else None,
+        use_safetensors=True,
+    ).to(device)
+    pipe_i2i = AutoPipelineForImage2Image.from_pipe(pipe_t2i).to(device)
+
+    _enable_memory_opts(pipe_t2i, device)
+    _enable_memory_opts(pipe_i2i, device)
+
+    def _draw(
+        prompt: str,
+        image: Image.Image | None = None,
+        negative_prompt: str = "",
+        strength: float = 0.55,
+        guidance_scale: float = 7.5,
+        num_inference_steps: int = 30,
+        width: int = 1024,
+        height: int = 1024,
+        max_side: int = 1024,
+    ) -> Image.Image:
+        prompt = prompt or ""
+        negative_prompt = negative_prompt or ""
+
+        if device == "cuda":
+            try:
+                torch.cuda.empty_cache()
+            except Exception:
+                pass
+
+        if image is None:
+            run_w, run_h = _align_size(width, height, max_side=max_side)
+            out = pipe_t2i(
+                prompt=prompt,
+                negative_prompt=negative_prompt,
+                guidance_scale=guidance_scale,
+                num_inference_steps=num_inference_steps,
+                width=run_w,
+                height=run_h,
+            ).images[0]
+            return out.convert("RGB")
+
+        image = image.convert("RGB")
+        orig_w, orig_h = image.size
+        run_w, run_h = _align_size(orig_w, orig_h, max_side=max_side)
+        if (run_w, run_h) != (orig_w, orig_h):
+            image_run = image.resize((run_w, run_h), resample=Image.BICUBIC)
+        else:
+            image_run = image
+
+        out = pipe_i2i(
+            prompt=prompt,
+            negative_prompt=negative_prompt,
+            image=image_run,
+            strength=strength,
+            guidance_scale=guidance_scale,
+            num_inference_steps=num_inference_steps,
+        ).images[0].convert("RGB")
+
+        if out.size != (orig_w, orig_h):
+            out = out.resize((orig_w, orig_h), resample=Image.BICUBIC)
+        return out
+
+    return _draw
+

python_server/model/Seg/__init__.py

@@ -0,0 +1 @@
+

python_server/model/Seg/mask2former_loader.py

@@ -0,0 +1,59 @@
+from __future__ import annotations
+
+from dataclasses import dataclass
+from typing import Callable, Literal, Tuple
+
+import numpy as np
+from PIL import Image
+
+
+@dataclass
+class Mask2FormerHFConfig:
+    """
+    使用 HuggingFace transformers 版本的 Mask2Former 语义分割。
+
+    model_id: HuggingFace 模型 id（默认 ADE20K semantic）
+    device: "cuda" | "cpu"
+    """
+
+    model_id: str = "facebook/mask2former-swin-large-ade-semantic"
+    device: str = "cuda"
+
+
+def build_mask2former_hf_predictor(
+    cfg: Mask2FormerHFConfig | None = None,
+) -> Tuple[Callable[[np.ndarray], np.ndarray], Mask2FormerHFConfig]:
+    """
+    返回 predictor(image_rgb_uint8) -> label_map(int32)。
+    """
+    cfg = cfg or Mask2FormerHFConfig()
+
+    import torch
+    from transformers import AutoImageProcessor, Mask2FormerForUniversalSegmentation
+
+    device = "cuda" if cfg.device.startswith("cuda") and torch.cuda.is_available() else "cpu"
+
+    processor = AutoImageProcessor.from_pretrained(cfg.model_id)
+    model = Mask2FormerForUniversalSegmentation.from_pretrained(cfg.model_id)
+    model.to(device).eval()
+
+    cfg = Mask2FormerHFConfig(model_id=cfg.model_id, device=device)
+
+    @torch.no_grad()
+    def _predict(image_rgb: np.ndarray) -> np.ndarray:
+        if image_rgb.dtype != np.uint8:
+            image_rgb_u8 = image_rgb.astype("uint8")
+        else:
+            image_rgb_u8 = image_rgb
+
+        pil = Image.fromarray(image_rgb_u8, mode="RGB")
+        inputs = processor(images=pil, return_tensors="pt").to(device)
+        outputs = model(**inputs)
+
+        # post-process to original size
+        target_sizes = [(pil.height, pil.width)]
+        seg = processor.post_process_semantic_segmentation(outputs, target_sizes=target_sizes)[0]
+        return seg.detach().to("cpu").numpy().astype("int32")
+
+    return _predict, cfg
+

python_server/model/Seg/seg_loader.py

@@ -0,0 +1,168 @@
+from __future__ import annotations
+
+"""
+统一的分割模型加载入口。
+
+当前支持：
+- SAM (segment-anything)
+- Mask2Former（使用 HuggingFace transformers 的语义分割实现）
+"""
+
+from dataclasses import dataclass
+from enum import Enum
+from typing import Callable
+
+import sys
+from pathlib import Path
+
+import numpy as np
+
+_THIS_DIR = Path(__file__).resolve().parent
+
+
+class SegBackend(str, Enum):
+    SAM = "sam"
+    MASK2FORMER = "mask2former"
+
+
+@dataclass
+class UnifiedSegConfig:
+    backend: SegBackend = SegBackend.SAM
+
+
+# -----------------------------
+# SAM (Segment Anything)
+# -----------------------------
+
+
+def _ensure_sam_on_path() -> Path:
+    sam_root = _THIS_DIR / "segment-anything"
+    if not sam_root.is_dir():
+        raise FileNotFoundError(f"未找到 segment-anything 仓库目录: {sam_root}")
+    sam_path = str(sam_root)
+    if sam_path not in sys.path:
+        sys.path.insert(0, sam_path)
+    return sam_root
+
+
+def _download_sam_checkpoint_if_needed(sam_root: Path) -> Path:
+    import requests
+
+    ckpt_dir = sam_root / "checkpoints"
+    ckpt_dir.mkdir(parents=True, exist_ok=True)
+    ckpt_path = ckpt_dir / "sam_vit_h_4b8939.pth"
+
+    if ckpt_path.is_file():
+        return ckpt_path
+
+    url = (
+        "https://dl.fbaipublicfiles.com/segment_anything/"
+        "sam_vit_h_4b8939.pth"
+    )
+    print(f"自动下载 SAM 权重:\n  {url}\n  -> {ckpt_path}")
+
+    resp = requests.get(url, stream=True)
+    resp.raise_for_status()
+
+    total = int(resp.headers.get("content-length", "0") or "0")
+    downloaded = 0
+    chunk_size = 1024 * 1024
+
+    with ckpt_path.open("wb") as f:
+        for chunk in resp.iter_content(chunk_size=chunk_size):
+            if not chunk:
+                continue
+            f.write(chunk)
+            downloaded += len(chunk)
+            if total > 0:
+                done = int(50 * downloaded / total)
+                print(
+                    "\r[{}{}] {:.1f}%".format(
+                        "#" * done,
+                        "." * (50 - done),
+                        downloaded * 100 / total,
+                    ),
+                    end="",
+                )
+    print("\nSAM 权重下载完成。")
+    return ckpt_path
+
+
+def _make_sam_predictor() -> Callable[[np.ndarray], np.ndarray]:
+    """
+    返回一个分割函数：
+    - 输入：RGB uint8 图像 (H, W, 3)
+    - 输出：语义标签图 (H, W)，每个目标一个 int id（从 1 开始）
+    """
+    sam_root = _ensure_sam_on_path()
+    ckpt_path = _download_sam_checkpoint_if_needed(sam_root)
+
+    from segment_anything import sam_model_registry, SamAutomaticMaskGenerator  # type: ignore[import]
+    import torch
+
+    device = "cuda" if torch.cuda.is_available() else "cpu"
+
+    sam = sam_model_registry["vit_h"](
+        checkpoint=str(ckpt_path),
+    ).to(device)
+
+    mask_generator = SamAutomaticMaskGenerator(sam)
+
+    def _predict(image_rgb: np.ndarray) -> np.ndarray:
+        if image_rgb.dtype != np.uint8:
+            image_rgb_u8 = image_rgb.astype("uint8")
+        else:
+            image_rgb_u8 = image_rgb
+
+        masks = mask_generator.generate(image_rgb_u8)
+        h, w, _ = image_rgb_u8.shape
+        label_map = np.zeros((h, w), dtype="int32")
+
+        for idx, m in enumerate(masks, start=1):
+            seg = m.get("segmentation")
+            if seg is None:
+                continue
+            label_map[seg.astype(bool)] = idx
+
+        return label_map
+
+    return _predict
+
+
+# -----------------------------
+# Mask2Former (占位)
+# -----------------------------
+
+
+def _make_mask2former_predictor() -> Callable[[np.ndarray], np.ndarray]:
+    from .mask2former_loader import build_mask2former_hf_predictor
+
+    predictor, _ = build_mask2former_hf_predictor()
+    return predictor
+
+
+# -----------------------------
+# 统一构建函数
+# -----------------------------
+
+
+def build_seg_predictor(
+    cfg: UnifiedSegConfig | None = None,
+) -> tuple[Callable[[np.ndarray], np.ndarray], SegBackend]:
+    """
+    统一构建分割预测函数。
+
+    返回：
+    - predictor(image_rgb: np.ndarray[H, W, 3], uint8) -> np.ndarray[H, W], int32
+    - 实际使用的 backend
+    """
+    cfg = cfg or UnifiedSegConfig()
+
+    if cfg.backend == SegBackend.SAM:
+        return _make_sam_predictor(), SegBackend.SAM
+
+    if cfg.backend == SegBackend.MASK2FORMER:
+        return _make_mask2former_predictor(), SegBackend.MASK2FORMER
+
+    raise ValueError(f"不支持的分割后端: {cfg.backend}")
+

python_server/model/__init__.py

@@ -0,0 +1 @@
+

python_server/server.py

@@ -0,0 +1,407 @@
+from __future__ import annotations
+
+import base64
+import datetime
+import io
+import os
+import shutil
+from dataclasses import asdict
+from pathlib import Path
+from typing import Any, Dict, Optional
+
+import numpy as np
+from fastapi import FastAPI, HTTPException
+from fastapi.responses import Response
+from pydantic import BaseModel, Field
+from PIL import Image, ImageDraw
+
+from config_loader import load_app_config, get_depth_backend_from_app
+from model.Depth.depth_loader import UnifiedDepthConfig, DepthBackend, build_depth_predictor
+
+from model.Seg.seg_loader import UnifiedSegConfig, SegBackend, build_seg_predictor
+from model.Inpaint.inpaint_loader import UnifiedInpaintConfig, InpaintBackend, build_inpaint_predictor
+from model.Animation.animation_loader import (
+    UnifiedAnimationConfig,
+    AnimationBackend,
+    build_animation_predictor,
+)
+
+
+APP_ROOT = Path(__file__).resolve().parent
+OUTPUT_DIR = APP_ROOT / "outputs"
+OUTPUT_DIR.mkdir(parents=True, exist_ok=True)
+
+app = FastAPI(title="HFUT Model Server", version="0.1.0")
+
+
+class ImageInput(BaseModel):
+    image_b64: str = Field(..., description="PNG/JPG 编码后的 base64（不含 data: 前缀）")
+    model_name: Optional[str] = Field(None, description="模型 key（来自 /models）")
+
+
+class DepthRequest(ImageInput):
+    pass
+
+
+class SegmentRequest(ImageInput):
+    pass
+
+
+class InpaintRequest(ImageInput):
+    prompt: Optional[str] = Field("", description="补全 prompt")
+    strength: float = Field(0.8, ge=0.0, le=1.0)
+    negative_prompt: Optional[str] = Field("", description="负向 prompt")
+    # 可选 mask（白色区域为重绘）
+    mask_b64: Optional[str] = Field(None, description="mask PNG base64（可选）")
+    # 推理缩放上限（避免 OOM）
+    max_side: int = Field(1024, ge=128, le=2048)
+
+
+class AnimateRequest(BaseModel):
+    model_name: Optional[str] = Field(None, description="模型 key（来自 /models）")
+    prompt: str = Field(..., description="文本提示词")
+    negative_prompt: Optional[str] = Field("", description="负向提示词")
+    num_inference_steps: int = Field(25, ge=1, le=200)
+    guidance_scale: float = Field(8.0, ge=0.0, le=30.0)
+    width: int = Field(512, ge=128, le=2048)
+    height: int = Field(512, ge=128, le=2048)
+    video_length: int = Field(16, ge=1, le=128)
+    seed: int = Field(-1, description="-1 表示随机种子")
+
+
+def _b64_to_pil_image(b64: str) -> Image.Image:
+    raw = base64.b64decode(b64)
+    return Image.open(io.BytesIO(raw))
+
+
+def _pil_image_to_png_b64(img: Image.Image) -> str:
+    buf = io.BytesIO()
+    img.save(buf, format="PNG")
+    return base64.b64encode(buf.getvalue()).decode("ascii")
+
+
+def _depth_to_png16_b64(depth: np.ndarray) -> str:
+    depth = np.asarray(depth, dtype=np.float32)
+    dmin = float(depth.min())
+    dmax = float(depth.max())
+    if dmax > dmin:
+        norm = (depth - dmin) / (dmax - dmin)
+    else:
+        norm = np.zeros_like(depth, dtype=np.float32)
+    u16 = (norm * 65535.0).clip(0, 65535).astype(np.uint16)
+    img = Image.fromarray(u16, mode="I;16")
+    return _pil_image_to_png_b64(img)
+
+def _depth_to_png16_bytes(depth: np.ndarray) -> bytes:
+    depth = np.asarray(depth, dtype=np.float32)
+    dmin = float(depth.min())
+    dmax = float(depth.max())
+    if dmax > dmin:
+        norm = (depth - dmin) / (dmax - dmin)
+    else:
+        norm = np.zeros_like(depth, dtype=np.float32)
+    # 前后端约定：最远=255，最近=0（8-bit）
+    u8 = ((1.0 - norm) * 255.0).clip(0, 255).astype(np.uint8)
+    img = Image.fromarray(u8, mode="L")
+    buf = io.BytesIO()
+    img.save(buf, format="PNG")
+    return buf.getvalue()
+
+
+def _default_half_mask(img: Image.Image) -> Image.Image:
+    w, h = img.size
+    mask = Image.new("L", (w, h), 0)
+    draw = ImageDraw.Draw(mask)
+    draw.rectangle([w // 2, 0, w, h], fill=255)
+    return mask
+
+
+# -----------------------------
+# /models（给前端/GUI 使用）
+# -----------------------------
+
+
+@app.get("/models")
+def get_models() -> Dict[str, Any]:
+    """
+    返回一个兼容 Qt 前端的 schema：
+    {
+      "models": {
+        "depth":   { "key": { "name": "..."} ... },
+        "segment": { ... },
+        "inpaint": { "key": { "name": "...", "params": [...] } ... }
+      }
+    }
+    """
+    return {
+        "models": {
+            "depth": {
+                # 兼容旧配置默认值
+                "midas": {"name": "MiDaS (default)"},
+                "zoedepth_n": {"name": "ZoeDepth (ZoeD_N)"},
+                "zoedepth_k": {"name": "ZoeDepth (ZoeD_K)"},
+                "zoedepth_nk": {"name": "ZoeDepth (ZoeD_NK)"},
+                "depth_anything_v2_vits": {"name": "Depth Anything V2 (vits)"},
+                "depth_anything_v2_vitb": {"name": "Depth Anything V2 (vitb)"},
+                "depth_anything_v2_vitl": {"name": "Depth Anything V2 (vitl)"},
+                "depth_anything_v2_vitg": {"name": "Depth Anything V2 (vitg)"},
+                "dpt_large": {"name": "DPT (large)"},
+                "dpt_hybrid": {"name": "DPT (hybrid)"},
+                "midas_dpt_beit_large_512": {"name": "MiDaS (dpt_beit_large_512)"},
+                "midas_dpt_swin2_large_384": {"name": "MiDaS (dpt_swin2_large_384)"},
+                "midas_dpt_swin2_tiny_256": {"name": "MiDaS (dpt_swin2_tiny_256)"},
+                "midas_dpt_levit_224": {"name": "MiDaS (dpt_levit_224)"},
+            },
+            "segment": {
+                "sam": {"name": "SAM (vit_h)"},
+                # 兼容旧配置默认值
+                "mask2former_debug": {"name": "SAM (compat mask2former_debug)"},
+                "mask2former": {"name": "Mask2Former (not implemented)"},
+            },
+            "inpaint": {
+                # 兼容旧配置默认值：copy 表示不做补全
+                "copy": {"name": "Copy (no-op)", "params": []},
+                "sdxl_inpaint": {
+                    "name": "SDXL Inpaint",
+                    "params": [
+                        {"id": "prompt", "label": "提示词", "optional": True},
+                    ],
+                },
+                "controlnet": {
+                    "name": "ControlNet Inpaint (canny)",
+                    "params": [
+                        {"id": "prompt", "label": "提示词", "optional": True},
+                    ],
+                },
+            },
+            "animation": {
+                "animatediff": {
+                    "name": "AnimateDiff (Text-to-Video)",
+                    "params": [
+                        {"id": "prompt", "label": "提示词", "optional": False},
+                        {"id": "negative_prompt", "label": "负向提示词", "optional": True},
+                        {"id": "num_inference_steps", "label": "采样步数", "optional": True},
+                        {"id": "guidance_scale", "label": "CFG Scale", "optional": True},
+                        {"id": "width", "label": "宽度", "optional": True},
+                        {"id": "height", "label": "高度", "optional": True},
+                        {"id": "video_length", "label": "帧数", "optional": True},
+                        {"id": "seed", "label": "随机种子", "optional": True},
+                    ],
+                },
+            },
+        }
+    }
+
+
+# -----------------------------
+# Depth
+# -----------------------------
+
+
+_depth_predictor = None
+_depth_backend: DepthBackend | None = None
+
+
+def _ensure_depth_predictor() -> None:
+    global _depth_predictor, _depth_backend
+    if _depth_predictor is not None and _depth_backend is not None:
+        return
+
+    app_cfg = load_app_config()
+    backend_str = get_depth_backend_from_app(app_cfg)
+    try:
+        backend = DepthBackend(backend_str)
+    except Exception as e:
+        raise ValueError(f"config.py 中 depth.backend 不合法: {backend_str}") from e
+
+    _depth_predictor, _depth_backend = build_depth_predictor(UnifiedDepthConfig(backend=backend))
+
+
+@app.post("/depth")
+def depth(req: DepthRequest):
+    """
+    计算深度并直接返回二进制 PNG（16-bit 灰度）。
+
+    约束：
+    - 前端不传/不选模型；模型选择写死在后端 config.py
+    - 成功：HTTP 200 + Content-Type: image/png
+    - 失败：HTTP 500，detail 为错误信息
+    """
+    try:
+        _ensure_depth_predictor()
+        pil = _b64_to_pil_image(req.image_b64).convert("RGB")
+        depth_arr = _depth_predictor(pil)  # type: ignore[misc]
+        png_bytes = _depth_to_png16_bytes(np.asarray(depth_arr))
+        return Response(content=png_bytes, media_type="image/png")
+    except Exception as e:
+        raise HTTPException(status_code=500, detail=str(e))
+
+
+# -----------------------------
+# Segment
+# -----------------------------
+
+
+_seg_cache: Dict[str, Any] = {}
+
+
+def _get_seg_predictor(model_name: str):
+    if model_name in _seg_cache:
+        return _seg_cache[model_name]
+
+    # 兼容旧默认 key
+    if model_name == "mask2former_debug":
+        model_name = "sam"
+
+    if model_name == "sam":
+        pred, _ = build_seg_predictor(UnifiedSegConfig(backend=SegBackend.SAM))
+        _seg_cache[model_name] = pred
+        return pred
+
+    if model_name == "mask2former":
+        pred, _ = build_seg_predictor(UnifiedSegConfig(backend=SegBackend.MASK2FORMER))
+        _seg_cache[model_name] = pred
+        return pred
+
+    raise ValueError(f"未知 segment model_name: {model_name}")
+
+
+@app.post("/segment")
+def segment(req: SegmentRequest) -> Dict[str, Any]:
+    try:
+        model_name = req.model_name or "sam"
+        pil = _b64_to_pil_image(req.image_b64).convert("RGB")
+        rgb = np.array(pil, dtype=np.uint8)
+
+        predictor = _get_seg_predictor(model_name)
+        label_map = predictor(rgb).astype(np.int32)
+
+        out_dir = OUTPUT_DIR / "segment"
+        out_dir.mkdir(parents=True, exist_ok=True)
+        out_path = out_dir / f"{model_name}_label.png"
+        # 保存为 8-bit 灰度（若 label 超过 255 会截断；当前 SAM 通常不会太大）
+        Image.fromarray(np.clip(label_map, 0, 255).astype(np.uint8), mode="L").save(out_path)
+
+        return {"success": True, "label_path": str(out_path)}
+    except Exception as e:
+        return {"success": False, "error": str(e)}
+
+
+# -----------------------------
+# Inpaint
+# -----------------------------
+
+
+_inpaint_cache: Dict[str, Any] = {}
+
+
+def _get_inpaint_predictor(model_name: str):
+    if model_name in _inpaint_cache:
+        return _inpaint_cache[model_name]
+
+    if model_name == "copy":
+        def _copy(image: Image.Image, *_args, **_kwargs) -> Image.Image:
+            return image.convert("RGB")
+        _inpaint_cache[model_name] = _copy
+        return _copy
+
+    if model_name == "sdxl_inpaint":
+        pred, _ = build_inpaint_predictor(UnifiedInpaintConfig(backend=InpaintBackend.SDXL_INPAINT))
+        _inpaint_cache[model_name] = pred
+        return pred
+
+    if model_name == "controlnet":
+        pred, _ = build_inpaint_predictor(UnifiedInpaintConfig(backend=InpaintBackend.CONTROLNET))
+        _inpaint_cache[model_name] = pred
+        return pred
+
+    raise ValueError(f"未知 inpaint model_name: {model_name}")
+
+
+@app.post("/inpaint")
+def inpaint(req: InpaintRequest) -> Dict[str, Any]:
+    try:
+        model_name = req.model_name or "sdxl_inpaint"
+        pil = _b64_to_pil_image(req.image_b64).convert("RGB")
+
+        if req.mask_b64:
+            mask = _b64_to_pil_image(req.mask_b64).convert("L")
+        else:
+            mask = _default_half_mask(pil)
+
+        predictor = _get_inpaint_predictor(model_name)
+        out = predictor(
+            pil,
+            mask,
+            req.prompt or "",
+            req.negative_prompt or "",
+            strength=req.strength,
+            max_side=req.max_side,
+        )
+
+        out_dir = OUTPUT_DIR / "inpaint"
+        out_dir.mkdir(parents=True, exist_ok=True)
+        out_path = out_dir / f"{model_name}_inpaint.png"
+        out.save(out_path)
+
+        return {"success": True, "output_path": str(out_path)}
+    except Exception as e:
+        return {"success": False, "error": str(e)}
+
+
+_animation_cache: Dict[str, Any] = {}
+
+
+def _get_animation_predictor(model_name: str):
+    if model_name in _animation_cache:
+        return _animation_cache[model_name]
+
+    if model_name == "animatediff":
+        pred, _ = build_animation_predictor(
+            UnifiedAnimationConfig(backend=AnimationBackend.ANIMATEDIFF)
+        )
+        _animation_cache[model_name] = pred
+        return pred
+
+    raise ValueError(f"未知 animation model_name: {model_name}")
+
+
+@app.post("/animate")
+def animate(req: AnimateRequest) -> Dict[str, Any]:
+    try:
+        model_name = req.model_name or "animatediff"
+        predictor = _get_animation_predictor(model_name)
+
+        result_path = predictor(
+            prompt=req.prompt,
+            negative_prompt=req.negative_prompt or "",
+            num_inference_steps=req.num_inference_steps,
+            guidance_scale=req.guidance_scale,
+            width=req.width,
+            height=req.height,
+            video_length=req.video_length,
+            seed=req.seed,
+        )
+
+        out_dir = OUTPUT_DIR / "animation"
+        out_dir.mkdir(parents=True, exist_ok=True)
+        ts = datetime.datetime.now().strftime("%Y%m%d_%H%M%S")
+        out_path = out_dir / f"{model_name}_{ts}.gif"
+        shutil.copy2(result_path, out_path)
+
+        return {"success": True, "output_path": str(out_path)}
+    except Exception as e:
+        return {"success": False, "error": str(e)}
+
+
+@app.get("/health")
+def health() -> Dict[str, str]:
+    return {"status": "ok"}
+
+
+if __name__ == "__main__":
+    import uvicorn
+
+    port = int(os.environ.get("MODEL_SERVER_PORT", "8000"))
+    uvicorn.run(app, host="0.0.0.0", port=port)
+

python_server/test_animation.py

@@ -0,0 +1,77 @@
+
+from __future__ import annotations
+
+from pathlib import Path
+import shutil
+
+from model.Animation.animation_loader import (
+    build_animation_predictor,
+    UnifiedAnimationConfig,
+    AnimationBackend,
+)
+
+
+# -----------------------------
+# 配置区（按需修改）
+# -----------------------------
+OUTPUT_DIR = "outputs/test_animation"
+ANIMATION_BACKEND = AnimationBackend.ANIMATEDIFF
+OUTPUT_FORMAT = "png_sequence"  # "gif" | "png_sequence"
+
+PROMPT = "a cinematic mountain landscape, camera slowly pans left"
+NEGATIVE_PROMPT = "blurry, low quality"
+NUM_INFERENCE_STEPS = 25
+GUIDANCE_SCALE = 8.0
+WIDTH = 512
+HEIGHT = 512
+VIDEO_LENGTH = 16
+SEED = -1
+CONTROL_IMAGE_PATH = "path/to/your_image.png"
+
+
+def main() -> None:
+    base_dir = Path(__file__).resolve().parent
+    out_dir = base_dir / OUTPUT_DIR
+    out_dir.mkdir(parents=True, exist_ok=True)
+
+    predictor, used_backend = build_animation_predictor(
+        UnifiedAnimationConfig(backend=ANIMATION_BACKEND)
+    )
+
+    if CONTROL_IMAGE_PATH.strip() in {"", "path/to/your_image.png"}:
+        raise ValueError("请先设置 CONTROL_IMAGE_PATH 为你的输入图片路径（png/jpg）。")
+
+    control_image = (base_dir / CONTROL_IMAGE_PATH).resolve()
+    if not control_image.is_file():
+        raise FileNotFoundError(f"control image not found: {control_image}")
+
+    result_path = predictor(
+        prompt=PROMPT,
+        negative_prompt=NEGATIVE_PROMPT,
+        num_inference_steps=NUM_INFERENCE_STEPS,
+        guidance_scale=GUIDANCE_SCALE,
+        width=WIDTH,
+        height=HEIGHT,
+        video_length=VIDEO_LENGTH,
+        seed=SEED,
+        control_image_path=str(control_image),
+        output_format=OUTPUT_FORMAT,
+    )
+
+    source = Path(result_path)
+    if OUTPUT_FORMAT == "png_sequence":
+        out_seq_dir = out_dir / f"{used_backend.value}_frames"
+        if out_seq_dir.exists():
+            shutil.rmtree(out_seq_dir)
+        shutil.copytree(source, out_seq_dir)
+        print(f"[Animation] backend={used_backend.value}, saved={out_seq_dir}")
+        return
+
+    out_path = out_dir / f"{used_backend.value}.gif"
+    out_path.write_bytes(source.read_bytes())
+    print(f"[Animation] backend={used_backend.value}, saved={out_path}")
+
+
+if __name__ == "__main__":
+    main()
+

python_server/test_depth.py

@@ -0,0 +1,101 @@
+from __future__ import annotations
+
+from pathlib import Path
+import numpy as np
+import cv2
+from PIL import Image
+
+# 解除大图限制
+Image.MAX_IMAGE_PIXELS = None 
+
+from model.Depth.depth_loader import build_depth_predictor, UnifiedDepthConfig, DepthBackend
+
+# ================= 配置区 =================
+INPUT_IMAGE = "/home/dwh/Documents/毕业设计/dwh/数据集/Up the River During Qingming (detail) - Court painters.jpg"
+OUTPUT_DIR = "outputs/test_depth_v4"
+DEPTH_BACKEND = DepthBackend.DEPTH_ANYTHING_V2
+
+# 边缘捕捉参数
+# 增大这个值会让边缘更细，减小会让边缘更粗（捕捉更多微弱信息）
+EDGE_TOP_PERCENTILE = 96.0  # 选取梯度最强的前 7.0% 的像素
+# 局部增强的灵敏度，建议在 2.0 - 5.0 之间
+CLAHE_CLIP_LIMIT = 3.0 
+# 形态学核大小
+MORPH_SIZE = 5
+# =========================================
+
+def _extract_robust_edges(depth_norm: np.ndarray) -> np.ndarray:
+    """
+    通过局部增强和 Sobel 梯度提取闭合边缘
+    """
+    # 1. 转换为 8 位灰度
+    depth_u8 = (depth_norm * 255).astype(np.uint8)
+    
+    # 2. 【核心步骤】CLAHE 局部自适应对比度增强
+    # 这会强行放大古画中细微的建筑/人物深度差
+    clahe = cv2.createCLAHE(clipLimit=CLAHE_CLIP_LIMIT, tileGridSize=(16, 16))
+    enhanced_depth = clahe.apply(depth_u8)
+    
+    # 3. 高斯模糊：减少数字化噪声
+    blurred = cv2.GaussianBlur(enhanced_depth, (5, 5), 0)
+    
+    # 4. Sobel 算子计算梯度强度
+    grad_x = cv2.Sobel(blurred, cv2.CV_64F, 1, 0, ksize=3)
+    grad_y = cv2.Sobel(blurred, cv2.CV_64F, 0, 1, ksize=3)
+    grad_mag = np.sqrt(grad_x**2 + grad_y**2)
+    
+    # 5. 【统计学阈值】不再死守固定数值，而是选 Top X%
+    threshold = np.percentile(grad_mag, EDGE_TOP_PERCENTILE)
+    binary_edges = (grad_mag >= threshold).astype(np.uint8) * 255
+    
+    # 6. 形态学闭合：桥接裂缝，让线条连起来
+    kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (MORPH_SIZE, MORPH_SIZE))
+    closed = cv2.morphologyEx(binary_edges, cv2.MORPH_CLOSE, kernel)
+    
+    # 7. 再次轻微膨胀，为 SAM 提供更好的引导范围
+    final_mask = cv2.dilate(closed, kernel, iterations=1)
+    
+    return final_mask
+
+def main() -> None:
+    base_dir = Path(__file__).resolve().parent
+    img_path = Path(INPUT_IMAGE)
+    out_dir = base_dir / OUTPUT_DIR
+    out_dir.mkdir(parents=True, exist_ok=True)
+
+    # 1. 初始化深度模型
+    predictor, used_backend = build_depth_predictor(UnifiedDepthConfig(backend=DEPTH_BACKEND))
+    
+    # 2. 加载图像
+    print(f"[Loading] 正在处理: {img_path.name}")
+    img_pil = Image.open(img_path).convert("RGB")
+    w_orig, h_orig = img_pil.size
+    
+    # 3. 深度预测
+    print(f"[Depth] 正在进行深度估计 (Large Image)...")
+    depth = np.asarray(predictor(img_pil), dtype=np.float32).squeeze()
+    
+    # 4. 归一化与保存
+    dmin, dmax = depth.min(), depth.max()
+    depth_norm = (depth - dmin) / (dmax - dmin + 1e-8)
+    depth_u16 = (depth_norm * 65535.0).astype(np.uint16)
+    Image.fromarray(depth_u16).save(out_dir / f"{img_path.stem}.depth.png")
+
+    # 5. 提取强鲁棒性边缘
+    print(f"[Edge] 正在应用 CLAHE + Sobel 增强算法提取边缘...")
+    edge_mask = _extract_robust_edges(depth_norm)
+    
+    # 6. 导出
+    mask_path = out_dir / f"{img_path.stem}.edge_mask_robust.png"
+    Image.fromarray(edge_mask).save(mask_path)
+
+    edge_ratio = float((edge_mask > 0).sum()) / float(edge_mask.size)
+    print("-" * 30)
+    print(f"提取完成！")
+    print(f"边缘密度: {edge_ratio:.2%} (目标通常应在 1% ~ 8% 之间)")
+    print(f"如果 Mask 依然太黑，请调低 EDGE_TOP_PERCENTILE (如 90.0)")
+    print(f"如果 Mask 太乱，请调高 EDGE_TOP_PERCENTILE (如 96.0)")
+    print("-" * 30)
+
+if __name__ == "__main__":
+    main()

python_server/test_inpaint.py

@@ -0,0 +1,284 @@
+
+
+from __future__ import annotations
+
+from datetime import datetime
+from pathlib import Path
+
+import numpy as np
+from PIL import Image, ImageDraw
+
+from model.Inpaint.inpaint_loader import (
+    build_inpaint_predictor,
+    UnifiedInpaintConfig,
+    InpaintBackend,
+    build_draw_predictor,
+    UnifiedDrawConfig,
+)
+
+
+# -----------------------------
+# 配置区（按需修改）
+# -----------------------------
+# 任务模式：
+# - "inpaint": 补全
+# - "draw": 绘图（文生图 / 图生图）
+TASK_MODE = "draw"
+
+# 指向你的输入图像，例如 image_0.png
+INPUT_IMAGE = "/home/dwh/code/hfut-bishe/python_server/outputs/test_seg_v2/Up the River During Qingming (detail) - Court painters.objects/Up the River During Qingming (detail) - Court painters.obj_06.png"
+INPUT_MASK = ""                                  # 为空表示不使用手工 mask
+OUTPUT_DIR = "outputs/test_inpaint_v2"
+MASK_RECT = (256, 0, 512, 512)                  # x1, y1, x2, y2 (如果不使用 AUTO_MASK)
+USE_AUTO_MASK = True                             # True: 自动从图像推断补全区域
+AUTO_MASK_BLACK_THRESHOLD = 6                    # 自动mask时，接近黑色像素阈值
+AUTO_MASK_DILATE_ITER = 4                        # 增加膨胀，确保树枝边缘被覆盖
+FALLBACK_TO_RECT_MASK = False                    # 自动mask为空时是否回退到矩形mask
+
+# 透明输出控制
+PRESERVE_TRANSPARENCY = True                     # 保持输出透明
+EXPAND_ALPHA_WITH_MASK = True                    # 将补全区域 alpha 设为不透明
+NON_BLACK_ALPHA_THRESHOLD = 6                    # 无 alpha 输入时，用近黑判定透明背景
+
+INPAINT_BACKEND = InpaintBackend.SDXL_INPAINT   # 推荐使用 SDXL_INPAINT 获得更好效果
+
+# -----------------------------
+# 关键 Prompt 修改：只生成树
+# -----------------------------
+# 细化 PROMPT，强调树的种类、叶子和风格，使其与原图融合
+PROMPT = (
+    "A highly detailed traditional Chinese ink brush painting. "
+    "Restore and complete the existing trees naturally. "
+    "Extend the complex tree branches and dense, varied green and teal leaves. "
+    "Add more harmonious foliage and intricate bark texture to match the style and flow of the original image_0.png trees. "
+    "Focus solely on vegetation. "
+    "Maintain the light beige background color and texture."
+)
+
+# 使用 NEGATIVE_PROMPT 显式排除不想要的内容
+NEGATIVE_PROMPT = (
+    "buildings, architecture, houses, pavilions, temples, windows, doors, "
+    "people, figures, persons, characters, figures, clothing, faces, hands, "
+    "text, writing, characters, words, letters, signatures, seals, stamps, "
+    "calligraphy, objects, artifacts, boxes, baskets, tools, "
+    "extra branches crossing unnaturally, bad composition, watermark, signature"
+)
+
+STRENGTH = 0.8                                   # 保持较高强度以进行生成
+GUIDANCE_SCALE = 8.0                             # 稍微增加，更严格遵循 prompt
+NUM_INFERENCE_STEPS = 35                         # 稍微增加，提升细节
+MAX_SIDE = 1024
+CONTROLNET_SCALE = 1.0
+
+# -----------------------------
+# 绘图（draw）参数
+# -----------------------------
+# DRAW_INPUT_IMAGE 为空时：文生图
+# DRAW_INPUT_IMAGE 不为空时：图生图（按输入图进行模仿/重绘）
+DRAW_INPUT_IMAGE = ""
+DRAW_PROMPT = """
+Chinese ink wash painting, vast snowy river under a pale sky,
+a small lonely boat at the horizon where water meets sky,
+an old fisherman wearing a straw hat sits at the stern, fishing quietly,
+gentle snowfall, misty atmosphere, distant mountains barely visible,
+minimalist composition, large empty space, soft brush strokes,
+calm, cold, and silent mood, poetic and serene
+"""
+
+DRAW_NEGATIVE_PROMPT = """
+blurry, low quality, many people, bright colors, modern elements, crowded, noisy
+"""
+DRAW_STRENGTH = 0.55                            # 仅图生图使用，越大越偏向重绘
+DRAW_GUIDANCE_SCALE = 9
+DRAW_STEPS = 64
+DRAW_WIDTH = 2560
+DRAW_HEIGHT = 1440
+DRAW_MAX_SIDE = 2560
+
+
+def _dilate(mask: np.ndarray, iterations: int = 1) -> np.ndarray:
+    out = mask.astype(bool)
+    for _ in range(max(0, iterations)):
+        p = np.pad(out, ((1, 1), (1, 1)), mode="constant", constant_values=False)
+        out = (
+            p[:-2, :-2] | p[:-2, 1:-1] | p[:-2, 2:]
+            | p[1:-1, :-2] | p[1:-1, 1:-1] | p[1:-1, 2:]
+            | p[2:, :-2] | p[2:, 1:-1] | p[2:, 2:]
+        )
+    return out
+
+
+def _make_rect_mask(img_size: tuple[int, int]) -> Image.Image:
+    w, h = img_size
+    x1, y1, x2, y2 = MASK_RECT
+    x1 = max(0, min(x1, w - 1))
+    x2 = max(0, min(x2, w - 1))
+    y1 = max(0, min(y1, h - 1))
+    y2 = max(0, min(y2, h - 1))
+    if x2 < x1:
+        x1, x2 = x2, x1
+    if y2 < y1:
+        y1, y2 = y2, y1
+
+    mask = Image.new("L", (w, h), 0)
+    draw = ImageDraw.Draw(mask)
+    draw.rectangle([x1, y1, x2, y2], fill=255)
+    return mask
+
+
+def _auto_mask_from_image(img_path: Path) -> Image.Image:
+    """
+    自动推断缺失区域：
+    1) 若输入带 alpha，透明区域作为 mask
+    2) 否则将“接近黑色”区域作为候选缺失区域
+    """
+    raw = Image.open(img_path)
+    arr = np.asarray(raw)
+
+    if arr.ndim == 3 and arr.shape[2] == 4:
+        alpha = arr[:, :, 3]
+        mask_bool = alpha < 250
+    else:
+        rgb = np.asarray(raw.convert("RGB"), dtype=np.uint8)
+        # 抠图后透明区域常被写成黑色，优先把近黑区域视作缺失
+        dark = np.all(rgb <= AUTO_MASK_BLACK_THRESHOLD, axis=-1)
+        mask_bool = dark
+
+    mask_bool = _dilate(mask_bool, AUTO_MASK_DILATE_ITER)
+    return Image.fromarray((mask_bool.astype(np.uint8) * 255), mode="L")
+
+
+def _build_alpha_from_input(img_path: Path, img_rgb: Image.Image) -> np.ndarray:
+    """
+    生成输出 alpha：
+    - 输入若有 alpha，优先沿用
+    - 输入若无 alpha，则把接近黑色区域视作透明背景
+    """
+    raw = Image.open(img_path)
+    arr = np.asarray(raw)
+    if arr.ndim == 3 and arr.shape[2] == 4:
+        return arr[:, :, 3].astype(np.uint8)
+
+    rgb = np.asarray(img_rgb.convert("RGB"), dtype=np.uint8)
+    non_black = np.any(rgb > NON_BLACK_ALPHA_THRESHOLD, axis=-1)
+    return (non_black.astype(np.uint8) * 255)
+
+
+def _load_or_make_mask(base_dir: Path, img_path: Path, img_rgb: Image.Image) -> Image.Image:
+    if INPUT_MASK:
+        raw_mask_path = Path(INPUT_MASK)
+        mask_path = raw_mask_path if raw_mask_path.is_absolute() else (base_dir / raw_mask_path)
+        if mask_path.is_file():
+            return Image.open(mask_path).convert("L")
+
+    if USE_AUTO_MASK:
+        auto = _auto_mask_from_image(img_path)
+        auto_arr = np.asarray(auto, dtype=np.uint8)
+        if (auto_arr > 0).any():
+            return auto
+        if not FALLBACK_TO_RECT_MASK:
+            raise ValueError("自动mask为空，请检查输入图像是否存在透明/黑色缺失区域。")
+
+    return _make_rect_mask(img_rgb.size)
+
+
+def _resolve_path(base_dir: Path, p: str) -> Path:
+    raw = Path(p)
+    return raw if raw.is_absolute() else (base_dir / raw)
+
+
+def run_inpaint_test(base_dir: Path, out_dir: Path) -> None:
+    img_path = _resolve_path(base_dir, INPUT_IMAGE)
+    if not img_path.is_file():
+        raise FileNotFoundError(f"找不到输入图像，请修改 INPUT_IMAGE: {img_path}")
+
+    predictor, used_backend = build_inpaint_predictor(
+        UnifiedInpaintConfig(backend=INPAINT_BACKEND)
+    )
+
+    img = Image.open(img_path).convert("RGB")
+    mask = _load_or_make_mask(base_dir, img_path, img)
+    mask_out = out_dir / f"{img_path.stem}.mask_used.png"
+    mask.save(mask_out)
+
+    kwargs = dict(
+        strength=STRENGTH,
+        guidance_scale=GUIDANCE_SCALE,
+        num_inference_steps=NUM_INFERENCE_STEPS,
+        max_side=MAX_SIDE,
+    )
+    if used_backend == InpaintBackend.CONTROLNET:
+        kwargs["controlnet_conditioning_scale"] = CONTROLNET_SCALE
+
+    out = predictor(img, mask, PROMPT, NEGATIVE_PROMPT, **kwargs)
+    out_path = out_dir / f"{img_path.stem}.{used_backend.value}.inpaint.png"
+    if PRESERVE_TRANSPARENCY:
+        alpha = _build_alpha_from_input(img_path, img)
+        mask_u8 = np.asarray(mask, dtype=np.uint8)
+        if EXPAND_ALPHA_WITH_MASK:
+            alpha = np.maximum(alpha, mask_u8)
+
+        out_rgb = np.asarray(out.convert("RGB"), dtype=np.uint8)
+        out_rgba = np.concatenate([out_rgb, alpha[..., None]], axis=-1)
+        Image.fromarray(out_rgba, mode="RGBA").save(out_path)
+    else:
+        out.save(out_path)
+
+    ratio = float((np.asarray(mask, dtype=np.uint8) > 0).sum()) / float(mask.size[0] * mask.size[1])
+    print(f"[Inpaint] backend={used_backend.value}, saved={out_path}")
+    print(f"[Mask] saved={mask_out}, ratio={ratio:.4f}")
+
+
+def run_draw_test(base_dir: Path, out_dir: Path) -> None:
+    """
+    绘图测试：
+    - 文生图：DRAW_INPUT_IMAGE=""
+    - 图生图：DRAW_INPUT_IMAGE 指向参考图
+    """
+    draw_predictor = build_draw_predictor(UnifiedDrawConfig())
+    ref_image: Image.Image | None = None
+    mode = "text2img"
+
+    if DRAW_INPUT_IMAGE:
+        ref_path = _resolve_path(base_dir, DRAW_INPUT_IMAGE)
+        if not ref_path.is_file():
+            raise FileNotFoundError(f"找不到参考图，请修改 DRAW_INPUT_IMAGE: {ref_path}")
+        ref_image = Image.open(ref_path).convert("RGB")
+        mode = "img2img"
+
+    out = draw_predictor(
+        prompt=DRAW_PROMPT,
+        image=ref_image,
+        negative_prompt=DRAW_NEGATIVE_PROMPT,
+        strength=DRAW_STRENGTH,
+        guidance_scale=DRAW_GUIDANCE_SCALE,
+        num_inference_steps=DRAW_STEPS,
+        width=DRAW_WIDTH,
+        height=DRAW_HEIGHT,
+        max_side=DRAW_MAX_SIDE,
+    )
+
+    ts = datetime.now().strftime("%Y%m%d_%H%M%S")
+    out_path = out_dir / f"draw_{mode}_{ts}.png"
+    out.save(out_path)
+    print(f"[Draw] mode={mode}, saved={out_path}")
+
+
+def main() -> None:
+    base_dir = Path(__file__).resolve().parent
+    out_dir = base_dir / OUTPUT_DIR
+    out_dir.mkdir(parents=True, exist_ok=True)
+
+    if TASK_MODE == "inpaint":
+        run_inpaint_test(base_dir, out_dir)
+        return
+    if TASK_MODE == "draw":
+        run_draw_test(base_dir, out_dir)
+        return
+
+    raise ValueError(f"不支持的 TASK_MODE: {TASK_MODE}（可选: inpaint / draw）")
+
+
+if __name__ == "__main__":
+    main()
+

python_server/test_seg.py

@@ -0,0 +1,163 @@
+from __future__ import annotations
+
+import os
+from pathlib import Path
+
+import numpy as np
+import cv2
+from PIL import Image, ImageDraw
+from scipy.ndimage import label as nd_label
+
+from model.Seg.seg_loader import SegBackend, _ensure_sam_on_path, _download_sam_checkpoint_if_needed
+
+# ================= 配置区 =================
+INPUT_IMAGE = "/home/dwh/Documents/毕业设计/dwh/数据集/Up the River During Qingming (detail) - Court painters.jpg"
+INPUT_MASK = "/home/dwh/code/hfut-bishe/python_server/outputs/test_depth/Up the River During Qingming (detail) - Court painters.depth_anything_v2.edge_mask.png"
+OUTPUT_DIR = "outputs/test_seg_v2"
+SEG_BACKEND = SegBackend.SAM
+
+# 目标筛选参数
+TARGET_MIN_AREA = 1000       # 过滤太小的碎片
+TARGET_MAX_OBJECTS = 20      # 最多提取多少个物体
+SAM_MAX_SIDE = 2048          # SAM 推理时的长边限制
+
+# 视觉效果
+MASK_ALPHA = 0.4
+BOUNDARY_COLOR = np.array([0, 255, 0], dtype=np.uint8) # 边界绿色
+TARGET_FILL_COLOR = np.array([255, 230, 0], dtype=np.uint8)
+SAVE_OBJECT_PNG = True
+# =========================================
+
+def _resize_long_side(arr: np.ndarray, max_side: int, is_mask: bool = False) -> tuple[np.ndarray, tuple[int, int], tuple[int, int]]:
+    h, w = arr.shape[:2]
+    if max_side <= 0 or max(h, w) <= max_side:
+        return arr, (h, w), (h, w)
+    scale = float(max_side) / float(max(h, w))
+    run_w, run_h = max(1, int(round(w * scale))), max(1, int(round(h * scale)))
+    resample = Image.NEAREST if is_mask else Image.BICUBIC
+    pil = Image.fromarray(arr)
+    out = pil.resize((run_w, run_h), resample=resample)
+    return np.asarray(out), (h, w), (run_h, run_w)
+
+def _get_prompts_from_mask(edge_mask: np.ndarray, max_components: int = 20):
+    """
+    分析边缘 Mask 的连通域，为每个独立的边缘簇提取一个引导点
+    """
+    # 确保是布尔类型
+    mask_bool = edge_mask > 127
+    # 连通域标记
+    labeled_array, num_features = nd_label(mask_bool)
+    
+    prompts = []
+    component_info = []
+    for i in range(1, num_features + 1):
+        coords = np.argwhere(labeled_array == i)
+        area = len(coords)
+        if area < 100: continue # 过滤噪声
+        # 取几何中心作为引导点
+        center_y, center_x = np.median(coords, axis=0).astype(int)
+        component_info.append(((center_x, center_y), area))
+    
+    # 按面积排序，优先处理大面积线条覆盖的物体
+    component_info.sort(key=lambda x: x[1], reverse=True)
+    
+    for pt, _ in component_info[:max_components]:
+        prompts.append({
+            "point_coords": np.array([pt], dtype=np.float32),
+            "point_labels": np.array([1], dtype=np.int32)
+        })
+    return prompts
+
+def _save_object_pngs(rgb: np.ndarray, label_map: np.ndarray, targets: list[int], out_dir: Path, stem: str):
+    obj_dir = out_dir / f"{stem}.objects"
+    obj_dir.mkdir(parents=True, exist_ok=True)
+    for idx, lb in enumerate(targets, start=1):
+        mask = (label_map == lb)
+        ys, xs = np.where(mask)
+        if len(ys) == 0: continue
+        y1, y2, x1, x2 = ys.min(), ys.max(), xs.min(), xs.max()
+        rgb_crop = rgb[y1:y2+1, x1:x2+1]
+        alpha = (mask[y1:y2+1, x1:x2+1].astype(np.uint8) * 255)[..., None]
+        rgba = np.concatenate([rgb_crop, alpha], axis=-1)
+        Image.fromarray(rgba).save(obj_dir / f"{stem}.obj_{idx:02d}.png")
+
+def main():
+    # 1. 加载资源
+    img_path, mask_path = Path(INPUT_IMAGE), Path(INPUT_MASK)
+    out_dir = Path(__file__).parent / OUTPUT_DIR
+    out_dir.mkdir(parents=True, exist_ok=True)
+    
+    rgb_orig = np.asarray(Image.open(img_path).convert("RGB"))
+    mask_orig = np.asarray(Image.open(mask_path).convert("L"))
+    
+    # 2. 准备推理尺寸 (缩放以节省显存)
+    rgb_run, orig_hw, run_hw = _resize_long_side(rgb_orig, SAM_MAX_SIDE)
+    mask_run, _, _ = _resize_long_side(mask_orig, SAM_MAX_SIDE, is_mask=True)
+    
+    # 3. 初始化 SAM（直接使用 SamPredictor，避免 wrapper 接口不支持 point/bbox prompt）
+    import torch
+
+    _ensure_sam_on_path()
+    sam_root = Path(__file__).resolve().parent / "model" / "Seg" / "segment-anything"
+    ckpt_path = _download_sam_checkpoint_if_needed(sam_root)
+    from segment_anything import sam_model_registry, SamPredictor  # type: ignore[import]
+
+    device = "cuda" if torch.cuda.is_available() else "cpu"
+    sam = sam_model_registry["vit_h"](checkpoint=str(ckpt_path)).to(device)
+    predictor = SamPredictor(sam)
+    
+    print(f"[SAM] 正在处理图像: {img_path.name}，推理尺寸: {run_hw}")
+    
+    # 4. 提取引导点
+    prompts = _get_prompts_from_mask(mask_run, max_components=TARGET_MAX_OBJECTS)
+    print(f"[SAM] 从 Mask 提取了 {len(prompts)} 个引导点")
+
+    # 5. 执行推理
+    final_label_map_run = np.zeros(run_hw, dtype=np.int32)
+    predictor.set_image(rgb_run)
+
+    for idx, p in enumerate(prompts, start=1):
+        # multimask_output=True 可以获得更稳定的结果
+        masks, scores, _ = predictor.predict(
+            point_coords=p["point_coords"],
+            point_labels=p["point_labels"],
+            multimask_output=True
+        )
+        # 挑选得分最高的 mask
+        best_mask = masks[np.argmax(scores)]
+
+        # 只有面积足够才保留
+        if np.sum(best_mask) > TARGET_MIN_AREA * (run_hw[0] / orig_hw[0]):
+            # 将新 mask 覆盖到 label_map 上，后续的覆盖前面的
+            final_label_map_run[best_mask > 0] = idx
+
+    # 6. 后处理与映射回原图
+    # 映射回原图尺寸 (Nearest 保证 label ID 不会产生小数)
+    label_map = np.asarray(Image.fromarray(final_label_map_run).resize((orig_hw[1], orig_hw[0]), Image.NEAREST))
+    
+    # 7. 导出与可视化
+    unique_labels = [l for l in np.unique(label_map) if l > 0]
+    
+    # 绘制可视化图
+    marked_img = rgb_orig.copy()
+    draw = ImageDraw.Draw(Image.fromarray(marked_img)) # 这里只是为了画框方便
+    
+    # 混合颜色显示
+    overlay = rgb_orig.astype(np.float32)
+    for lb in unique_labels:
+        m = (label_map == lb)
+        overlay[m] = overlay[m] * (1-MASK_ALPHA) + TARGET_FILL_COLOR * MASK_ALPHA
+        # 简单边缘处理
+        contours, _ = cv2.findContours(m.astype(np.uint8), cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
+        cv2.drawContours(marked_img, contours, -1, (0, 255, 0), 2)
+
+    final_vis = Image.fromarray(cv2.addWeighted(marked_img, 0.7, overlay.astype(np.uint8), 0.3, 0))
+    final_vis.save(out_dir / f"{img_path.stem}.sam_guided_result.png")
+    
+    if SAVE_OBJECT_PNG:
+        _save_object_pngs(rgb_orig, label_map, unique_labels, out_dir, img_path.stem)
+        
+    print(f"[Done] 分割完成。提取了 {len(unique_labels)} 个物体。结果保存在: {out_dir}")
+
+if __name__ == "__main__":
+    main()