Enhance device management and model handling in SnapForm

- Updated frame extraction logic to allow for single frame extraction.
- Integrated model readiness checks for BEN2 and DA3 models in relevant operators.
- Added directory name attribute to model specifications for better organization.
- Improved device selection logic for production and debug tasks.
- Refactored preferences to expose device settings and ensure they reflect available hardware.
- Enhanced runtime device detection and management, including support for CUDA and MPS.
- Updated tests to cover new device management features and model directory handling.
- Removed deprecated download model tests to streamline codebase.

README.md

@@ -18,8 +18,8 @@ SnapForm 是一个 Blender 扩展，可以把一张照片、一段视频或一
 - Linux：`SnapForm.v0.10.10.linux-x64.zip`
 2. 打开 Blender，进入 `Preferences > Extensions`
 3. 点击 `Install from Disk`，选择下载的压缩包
-4. 在扩展偏好设置中确认 `uv` 路径与 Torch 后端（保持默认即可）；如果下载经常超时，
-   可以在 `PyPI Index` 里填一个国内镜像，例如 `https://pypi.tuna.tsinghua.edu.cn/simple`
+4. 在扩展偏好设置中确认 Storage Root 与 Device（保持默认即可）；Device 默认取本机
+   最快的 CUDA/MPS，否则 CPU。下载超时会自动重试国内镜像，无需手动配置
 5. 打开 3D 视图右侧的 `SnapForm` 面板，点击 `Install Runtime`
 
 DA3、BEN2 源码包和 uv 已内置在扩展里，安装时直接从本地解包，不需要访问 GitHub。首次使用仍会自动

docs/technical.md

@@ -50,18 +50,21 @@
    `linux-x64` 四个平台的 Extension ZIP（如 `SnapForm.v0.10.10.macos-arm64.zip`）。
    每个包只含对应平台的 uv 二进制，并内置 DA3、BEN2 源码包。
 2. 在 Blender 的 `Preferences > Extensions` 中选择 `Install from Disk`。
-3. 在 Extension Preferences 中确认 `uv` 路径和 Torch Backend。
+3. 在 Extension Preferences 中确认 Storage Root 与 Device。Device 是正式功能
+   （Image Empty 右键菜单）使用的推理设备，默认取本机最快的 CUDA/MPS，否则 CPU，且只列出
+   本机支持的设备。
 4. 在 `Host` 点击 `Install Runtime`。首次安装会下载 Python、PyTorch、FastAPI、Uvicorn；
    DA3、BEN2 源码包与 uv 已内置在扩展中，安装时从 `~/.snapform/sources/` 解包使用，
    并默认下载 `DA3 Small`、`DA3 Base` 和 `BEN2 Base` 模型。安装按 8 个阶段依次推进，每段进度
     从 0 到 100%：解包内置 uv、创建 Python Runtime、下载 PyTorch、安装 xFormers（macOS 跳过，
     内置 DA3 源码包为官方原版并已移除 xformers 硬依赖，自动回退纯 PyTorch 实现）、安装 AI 依赖，
-    然后是三个模型的下载。进度文字统一为 `阶段/总数 阶段名: 状态`，
+    然后是三个模型的下载。安装完成后进入 `验证 Runtime` 阶段，会导入 Torch 与全部 AI 依赖；
+    macOS 首次导入 Torch 会初始化 Metal 后端（编译 shader 缓存），该阶段可能持续较久，状态栏
+    会显示已用时间而不是卡住。进度文字统一为 `阶段/总数 阶段名: 状态`，
     模型下载阶段会实时显示下载速度，例如 `6/8 da3 small: downloading 14.2MB/s`。
     Runtime 就绪后即使模型下载中断也不影响 Host 使用，可稍后在 `Models` 面板单独补下。
-    安装不做启动网络预检：若网络无法稳定访问 pypi.org，可在扩展偏好设置的 `PyPI Index`
-    中指定镜像源（如清华、阿里），该地址会通过 `--pypi-index` 传给 bootstrap 并优先使用；
-    未指定时，pip 安装失败会自动重试一次清华镜像，模型下载由 worker 自行探测回退到 HF 镜像。
+    安装不做启动网络预检：pip 安装失败会自动依次重试清华、阿里镜像，模型下载由 worker
+    自行探测回退到 HF 镜像；需要手动指定镜像时可通过 `PIP_INDEX_URL` 环境变量覆盖。
 5. 在 3D View 按 `N`，打开 `SnapForm` 页签。
 6. 如需其他模型，展开 `Host > Models`，找到模型并点击该行的 `Download`。
 7. 展开默认收起的 `Debug`，选择 `Export Format` 和本次执行使用的模型。
@@ -85,7 +88,7 @@
 状态和原始内容。BEN2 的 `foreground.png` 也会保留在对应 job 目录中。
 
 Image Empty 引用视频或序列帧时，Remove Background 直接执行，默认取 Empty 的 Image Frames
-数值（未设置时为完整帧数），起始帧跟随 Empty 的 Offset；若 Offset 加 Frames 超出视频长度，
+数值（未设置时为完整帧数，显式设为 1 时只处理 1 帧），起始帧跟随 Empty 的 Offset；若 Offset 加 Frames 超出视频长度，
 会把窗口整体收进视频内（保证实际处理 Frames 帧数），Offset 过小时从第 1 帧开始。输出
 `foreground.0001.png` 等透明序列帧（帧号从 0001 开始，与 Blender 场景帧一一对应），并把
 源帧以 `frame.0001.jpg` 等命名复制到 job 目录根下供材质颜色导入。替换 Empty 的 Image 为
@@ -115,7 +118,7 @@ Cyclic 沿用原 Empty 的设置。序列帧无法被 Blender 整体 Pack，因
     `img.foreground` 与 `img.depth`，不会生成 `img.foreground.depth`。Original 仍用原图作为
     位移来源，不改变原有深度逻辑。
     Image Empty 引用视频或序列帧时，AI Generated 会按 Empty 的 Image Frames 数值逐帧估计
-    深度（未设置时为完整帧数），起始帧跟随 Empty 的 Offset，与 Remove Background 范围一致；
+    深度（未设置时为完整帧数，显式设为 1 时只处理 1 帧），起始帧跟随 Empty 的 Offset，与 Remove Background 范围一致；
     Offset 超出视频长度时同样把窗口收进视频内。材质节点与置换贴图的 Start 保持与原 Empty
     一致；Original 模式（无 Remove Background）时，单帧会复制原图改名为 `.depth` 供位移使用，
     视频或序列帧则抽帧保存为 job 目录根下的 `frame.0001.jpg` 等，并作为 `.depth` 序列使用。
@@ -307,8 +310,12 @@ Windows 上代理若只在系统设置里而未写入环境变量，安装程序
   并核对覆盖文件是否仍适用
 - `addict` 2.4.0（补充 DA3 上游包元数据暂未声明的运行依赖）
 
-默认 CUDA Backend 为 CUDA 12.8，也可以选择 CPU。CPU 可以用于验证，但实际推理
-通常很慢。
+安装 Runtime 时会自动选择 PyTorch 后端：检测到 NVIDIA 驱动则安装 CUDA 12.8 版本，否则安装
+CPU 版；macOS 始终安装 CPU 版 PyTorch（MPS 包含其中）。CPU 可以用于验证，但实际推理通常很慢。
+Extension Preferences 的 `Device` 决定正式功能的推理设备，默认取本机最快的 CUDA/MPS，否则
+CPU；Debug 面板的 `Debug Device` 只影响 Debug 任务。两处设备选项都只列出本机支持的设备：
+Windows 不显示 MPS，机器上没有可用 NVIDIA 驱动时不显示 CUDA。Runtime 安装完成时会检测真实
+可用的设备并写入 `snapform_ready.json`，设备列表以该检测结果为准。
 CUDA 12.8 的 PyTorch wheel 自带 CUDA 运行库，无需安装 CUDA Toolkit；只需要
 NVIDIA 驱动 ≥ 570（数据中心显卡可用更老驱动）。驱动不满足或没有 NVIDIA 显卡时，
 设备会自动回退到 CPU（macOS 为 MPS），只有显式选择 CUDA 才会报错。
@@ -352,6 +359,9 @@ Runtime、模型和任务结果默认保存在用户主目录的 `~/.snapform`
 └── host.lock       # Host 运行时持有
 ```
 
+`models/` 下每个模型一个文件夹，命名统一带前缀：DA3 为 `da3_*`（如 `da3_small`、`da3_base`），
+BEN2 为 `ben2_*`（如 `ben2_base`）。旧版无前缀的 DA3 文件夹会在首次使用时自动改名为新名称。
+
 不同 Blender 版本安装相同 Extension 后会共用这个目录。也可以在 Extension
 Preferences 的 `Storage Root` 中修改共享位置。不要把模型写入 Extension 安装目录，
 因为该目录可能只读，并且升级时会被替换。不要同时从多个 Blender 实例执行
@@ -360,7 +370,7 @@ Preferences 的 `Storage Root` 中修改共享位置。不要把模型写入 Ext
 ## 开发
 
 ```bash
-uv run python -m unittest discover -s tests -v
+uv run pytest tests -v
 uv run python pack.py
 ```
 

pyproject.toml

@@ -7,5 +7,8 @@ requires-python = ">=3.11"
 license = "GPL-3.0-or-later"
 dependencies = []
 
+[dependency-groups]
+dev = ["pytest"]
+
 [tool.uv]
 package = false

src/snapform/ai_host.py

@@ -147,7 +147,7 @@ class ModelCache:
                 Path(self.ben2_key[0]).name if self.ben2_key is not None else ""
             ),
             "da3": (
-                f"da3_{Path(self.da3_key[0]).name}"
+                Path(self.da3_key[0]).name
                 if self.da3_key is not None
                 else ""
             ),

src/snapform/bootstrap_runtime.py

@@ -50,11 +50,15 @@ PLAIN_SLICE_SPAN = 0.02
 RAMP_RATE = 0.02
 CREEP_CAP = 0.60
 CHECK_CODE = (
-    "import torch; "
+    "import json; import torch; "
     "from depth_anything_3.api import DepthAnything3; "
     "from ben2 import BEN_Base; "
     "import fastapi, uvicorn; "
-    "print(torch.__version__, torch.cuda.is_available())"
+    "mps = getattr(torch.backends, 'mps', None); "
+    "print('SNAPFORM_DEVICES=' + json.dumps({"
+    "'torch': torch.__version__, "
+    "'cuda': torch.cuda.is_available(), "
+    "'mps': bool(mps is not None and mps.is_available())}))"
 )
 
 
@@ -373,7 +377,7 @@ def install_stages(args, runtime_dir, python, uv):
             "--repository",
             model.repository,
             "--model-dir",
-            models_dir / model.key.lower(),
+            models_dir / model.directory_name,
             "--status",
             args.status,
         ]
@@ -405,6 +409,70 @@ def verify_command(python):
     return [python, "-c", CHECK_CODE]
 
 
+def run_verify(
+    python,
+    status_path,
+    start=VERIFY_SLICE_START,
+    end=VERIFY_SLICE_END,
+    clock=time.monotonic,
+):
+    """Run the runtime import check with heartbeat progress; return the device report."""
+    command = verify_command(python)
+    print("Running:", subprocess.list2cmdline([str(value) for value in command]))
+    process = subprocess.Popen(
+        [str(value) for value in command],
+        stdout=subprocess.PIPE,
+        stderr=subprocess.STDOUT,
+        text=True,
+        encoding="utf-8",
+        errors="replace",
+    )
+    output_queue = queue.Queue()
+
+    def read_output():
+        for raw_line in process.stdout:
+            output_queue.put(raw_line)
+        output_queue.put(None)
+
+    threading.Thread(target=read_output, daemon=True).start()
+
+    report = {}
+    started = clock()
+    last_heartbeat = 0.0
+    while True:
+        try:
+            raw_line = output_queue.get(timeout=0.2)
+            empty = False
+        except queue.Empty:
+            raw_line = None
+            empty = True
+        if raw_line is None and not empty:
+            break
+        if empty:
+            elapsed = clock() - started
+            if elapsed - last_heartbeat >= 5.0:
+                last_heartbeat = elapsed
+                creep = min(elapsed / 60.0, CREEP_CAP)
+                progress = start + creep * (end - start)
+                write_status(
+                    status_path,
+                    progress,
+                    f"verifying runtime {format_elapsed(elapsed)}",
+                )
+            continue
+        line = raw_line.rstrip()
+        print(line)
+        if line.startswith("SNAPFORM_DEVICES="):
+            try:
+                report = json.loads(line[len("SNAPFORM_DEVICES=") :])
+            except json.JSONDecodeError:
+                report = {}
+    return_code = process.wait()
+    if return_code != 0:
+        raise subprocess.CalledProcessError(return_code, command)
+    return report
+
+
 def uv_download_target():
     machine = platform.machine().lower()
     if sys.platform == "win32":
@@ -643,9 +711,15 @@ def install(args):
             installed_packages.extend(stage_installed)
         runtime_offset = stage_end
     write_status(args.status, VERIFY_SLICE_START, "verifying runtime")
-    run(verify_command(python))
+    device_report = run_verify(python, args.status)
     write_status(args.status, VERIFY_SLICE_END, "Runtime is ready")
 
+    devices = []
+    if device_report.get("cuda"):
+        devices.append("cuda")
+    if device_report.get("mps"):
+        devices.append("mps")
+    devices.append("cpu")
     runtime_dir.mkdir(parents=True, exist_ok=True)
     ready_path = runtime_dir / "snapform_ready.json"
     ready_path.write_text(
@@ -658,6 +732,7 @@ def install(args):
                 "fastapi": FASTAPI_VERSION,
                 "uvicorn": UVICORN_VERSION,
                 "backend": args.backend,
+                "devices": devices,
                 "da3_commit": args.da3_commit,
                 "ben2_commit": args.ben2_commit,
                 "packages": sorted(set(installed_packages)),

src/snapform/image_depth.py

@@ -97,7 +97,7 @@ def extract_frame_range(source_object):
     )
     frame_count = (
         min(image_frames, total_frames)
-        if image_frames > 1
+        if image_frames >= 1
         else total_frames
     )
     start_frame = min(1 + frame_offset, total_frames - frame_count + 1)
@@ -1400,12 +1400,14 @@ class RemoveImageBackground(bpy.types.Operator):
     @classmethod
     def poll(cls, context):
         from .operators import active_job
-        from .runtime import host_busy
+        from .runtime import host_busy, model_ready
+        from .models import BEN2_MODEL
 
         return (
             is_image_empty(context.object)
             and active_job() is None
             and not host_busy()
+            and model_ready(BEN2_MODEL.key)
         )
 
     def execute(self, context):
@@ -1511,12 +1513,14 @@ class ConvertImageToDepthPlane(bpy.types.Operator):
     @classmethod
     def poll(cls, context):
         from .operators import active_job
-        from .runtime import host_busy
+        from .runtime import host_busy, model_ready
+        from .models import MODELS
 
         return (
             is_image_empty(context.object)
             and active_job() is None
             and not host_busy()
+            and any(model_ready(model_key) for model_key in MODELS)
         )
 
     def invoke(self, context, _event):
@@ -1535,6 +1539,9 @@ class ConvertImageToDepthPlane(bpy.types.Operator):
         return context.window_manager.invoke_props_dialog(self, width=340)
 
     def draw(self, context):
+        from .models import BEN2_MODEL
+        from .runtime import model_ready
+
         layout = self.layout
         layout.prop(self, "snapform_depth_source")
         if self.snapform_depth_source == "ESTIMATE":
@@ -1563,7 +1570,9 @@ class ConvertImageToDepthPlane(bpy.types.Operator):
         if settings_body is not None:
             settings_body.prop(self, "snapform_displacement_via")
             settings_body.prop(self, "snapform_subdivision_level")
-            settings_body.prop(self, "snapform_remove_background")
+            remove_background_row = settings_body.row()
+            remove_background_row.enabled = model_ready(BEN2_MODEL.key)
+            remove_background_row.prop(self, "snapform_remove_background")
             settings_body.prop(self, "snapform_unshaded")
         layout.prop(self, "snapform_delete_original")
 

src/snapform/models.py

@@ -5,6 +5,7 @@ from dataclasses import dataclass
 class ModelSpec:
     key: str
     label: str
+    directory_name: str
     repository: str
     license_name: str
     description: str
@@ -16,6 +17,7 @@ MODELS = {
     "SMALL": ModelSpec(
         key="SMALL",
         label="DA3 Small",
+        directory_name="da3_small",
         repository="depth-anything/DA3-SMALL",
         license_name="Apache 2.0",
         description="Fastest any-view model with Apache 2.0 weights",
@@ -24,6 +26,7 @@ MODELS = {
     "BASE": ModelSpec(
         key="BASE",
         label="DA3 Base",
+        directory_name="da3_base",
         repository="depth-anything/DA3-BASE",
         license_name="Apache 2.0",
         description="Balanced any-view model with Apache 2.0 weights",
@@ -32,6 +35,7 @@ MODELS = {
     "LARGE_1_1": ModelSpec(
         key="LARGE_1_1",
         label="DA3 Large 1.1",
+        directory_name="da3_large_1_1",
         repository="depth-anything/DA3-LARGE-1.1",
         license_name="CC BY-NC 4.0",
         description="Higher quality non-commercial model with larger VRAM use",
@@ -40,6 +44,7 @@ MODELS = {
     "GIANT_1_1": ModelSpec(
         key="GIANT_1_1",
         label="DA3 Giant 1.1",
+        directory_name="da3_giant_1_1",
         repository="depth-anything/DA3-GIANT-1.1",
         license_name="CC BY-NC 4.0",
         description="Non-commercial Giant model with Gaussian head",
@@ -48,6 +53,7 @@ MODELS = {
     "NESTED_GIANT_LARGE_1_1": ModelSpec(
         key="NESTED_GIANT_LARGE_1_1",
         label="DA3 Nested Giant Large 1.1",
+        directory_name="da3_nested_giant_large_1_1",
         repository="depth-anything/DA3NESTED-GIANT-LARGE-1.1",
         license_name="CC BY-NC 4.0",
         description="Non-commercial nested model with metric depth and Gaussian head",
@@ -58,6 +64,7 @@ MODELS = {
 BEN2_MODEL = ModelSpec(
     key="BEN2_BASE",
     label="BEN2 Base",
+    directory_name="ben2_base",
     repository="PramaLLC/BEN2",
     license_name="MIT",
     description="Fast high-quality background removal model",
@@ -142,3 +149,8 @@ def get_downloadable_model(model_key):
         return DOWNLOADABLE_MODELS[model_key]
     except KeyError as error:
         raise ValueError(f"Unknown model: {model_key}") from error
+
+
+def model_directory_name(model_key):
+    """Return the model's folder name inside the shared models directory."""
+    return get_downloadable_model(model_key).directory_name

src/snapform/operators.py

@@ -337,6 +337,7 @@ class RunJob(bpy.types.Operator):
                     paths["result"],
                     BEN2_MODEL.key,
                     start_frame=start_frame,
+                    device=self._task_device(settings),
                 ),
                 paths,
             )
@@ -398,10 +399,21 @@ class RunJob(bpy.types.Operator):
                     self.job_type == IMAGE_DEPTH_JOB
                     and remove_background
                 ),
+                device=self._task_device(settings),
             ),
             paths,
         )
 
+    def _task_device(self, settings):
+        """Debug panels use the scene device; production tasks use Preferences."""
+        if self.job_type in {
+            IMAGE_DEPTH_JOB,
+            IMAGE_BACKGROUND_JOB,
+            ORIGINAL_BACKGROUND_JOB,
+        }:
+            return runtime.production_device()
+        return settings.device
+
     def _require_online_access(self):
         if not bpy.app.online_access:
             raise RuntimeError(

src/snapform/preferences.py

@@ -1,9 +1,9 @@
-import shutil
-import sys
 from pathlib import Path
 
 import bpy
 
+from .runtime import device_enum_items
+
 
 class SnapFormPreferences(bpy.types.AddonPreferences):
     bl_idname = __package__
@@ -17,31 +17,14 @@ class SnapFormPreferences(bpy.types.AddonPreferences):
         subtype="DIR_PATH",
         default=str(Path.home() / ".snapform"),
     )
-    uv_executable: bpy.props.StringProperty(
-        name="uv Executable",
+    device: bpy.props.EnumProperty(
+        name="Device",
         description=(
-            "Path to uv; leave empty to auto-download a standalone uv "
-            "when it is missing"
+            "Inference device used by production tasks; only devices supported "
+            "on this machine are listed"
         ),
-        subtype="FILE_PATH",
-        default=shutil.which("uv") or "",
-    )
-    torch_backend: bpy.props.EnumProperty(
-        name="Torch Backend",
-        description="PyTorch build installed into the isolated Runtime",
-        items=(
-            ("CU128", "CUDA 12.8", "NVIDIA CUDA 12.8 PyTorch wheels"),
-            ("CPU", "CPU", "CPU-only PyTorch wheels"),
-        ),
-        default="CPU" if sys.platform == "darwin" else "CU128",
-    )
-    pypi_index_url: bpy.props.StringProperty(
-        name="PyPI Index",
-        description=(
-            "Package index for the isolated Runtime; leave empty to auto-detect "
-            "with mirror fallback. Example: https://pypi.tuna.tsinghua.edu.cn/simple"
-        ),
-        default="",
+        items=device_enum_items,
+        default=0,
     )
 
     def draw(self, _context):
@@ -50,9 +33,7 @@ class SnapFormPreferences(bpy.types.AddonPreferences):
 
         layout = self.layout
         layout.prop(self, "storage_root")
-        layout.prop(self, "uv_executable")
-        layout.prop(self, "torch_backend")
-        layout.prop(self, "pypi_index_url")
+        layout.prop(self, "device")
         layout.separator()
         layout.label(text=f"Runtime: {runtime_directory()}")
         if runtime_ready():

src/snapform/properties.py

@@ -1,6 +1,7 @@
 import bpy
 
 from .models import gs_model_enum_items, model_enum_items
+from .runtime import device_enum_items
 
 
 class SnapFormSettings(bpy.types.PropertyGroup):
@@ -18,7 +19,7 @@ class SnapFormSettings(bpy.types.PropertyGroup):
         name="Extract Frames",
         description="Maximum number of consecutive frames extracted from the start of a video or frame sequence",
         default=16,
-        min=2,
+        min=1,
         soft_max=64,
     )
     model: bpy.props.EnumProperty(
@@ -32,15 +33,13 @@ class SnapFormSettings(bpy.types.PropertyGroup):
         items=gs_model_enum_items,
     )
     device: bpy.props.EnumProperty(
-        name="Device",
-        description="Inference device used by the external Runtime",
-        items=(
-            ("AUTO", "Auto", "Use CUDA, then MPS, otherwise CPU"),
-            ("CUDA", "CUDA", "Require an NVIDIA CUDA device"),
-            ("MPS", "MPS", "Use Apple Metal (MPS)"),
-            ("CPU", "CPU", "Run on CPU"),
+        name="Debug Device",
+        description=(
+            "Inference device used by Debug tasks only; production tasks use "
+            "the Device setting in the Add-on Preferences"
         ),
-        default="AUTO",
+        items=device_enum_items,
+        default=0,
     )
     process_res: bpy.props.IntProperty(
         name="Process Res",

src/snapform/runtime.py

@@ -1,5 +1,6 @@
 import json
 import os
+import platform
 import secrets
 import shutil
 import socket
@@ -15,7 +16,12 @@ from urllib.request import ProxyHandler, Request, build_opener
 
 import bpy
 
-from .models import BEN2_MODEL, get_downloadable_model, selected_model_key
+from .models import (
+    BEN2_MODEL,
+    get_downloadable_model,
+    model_directory_name,
+    selected_model_key,
+)
 
 
 DA3_COMMIT = "3d835ec1a5802d64a8b8b15f817a1ab54809bfe4"
@@ -107,7 +113,16 @@ def models_directory():
 
 
 def model_directory(model_key):
-    return models_directory() / model_key.lower()
+    root = models_directory()
+    directory = root / model_directory_name(model_key)
+    legacy_directory = root / model_key.lower()
+    if (
+        not directory.is_dir()
+        and legacy_directory.is_dir()
+        and legacy_directory != directory
+    ):
+        legacy_directory.rename(directory)
+    return directory
 
 
 def model_ready(model_key):
@@ -117,6 +132,61 @@ def model_ready(model_key):
     )
 
 
+def runtime_devices():
+    """Devices detected by the installed Runtime, or None when unknown."""
+    ready_path = runtime_directory() / "snapform_ready.json"
+    if not ready_path.is_file():
+        return None
+    try:
+        payload = json.loads(ready_path.read_text(encoding="utf-8"))
+    except (json.JSONDecodeError, OSError):
+        return None
+    devices = payload.get("devices")
+    if not isinstance(devices, list) or not devices:
+        return None
+    return [str(name).lower() for name in devices]
+
+
+def machine_devices():
+    """Devices the machine can run, fastest first; used before Runtime install."""
+    devices = ["cpu"]
+    if shutil.which("nvidia-smi"):
+        devices.insert(0, "cuda")
+    elif sys.platform == "darwin" and platform.machine().lower() in {
+        "arm64",
+        "aarch64",
+    }:
+        devices.insert(0, "mps")
+    return devices
+
+
+def supported_devices():
+    """Inference devices available on this machine, fastest first."""
+    return runtime_devices() or machine_devices()
+
+
+DEVICE_LABELS = {
+    "cuda": ("CUDA", "NVIDIA CUDA"),
+    "mps": ("MPS", "Apple Metal (MPS)"),
+    "cpu": ("CPU", "CPU"),
+}
+
+
+def device_enum_items(_owner, _context):
+    """Blender EnumProperty items listing only devices supported on this machine."""
+    return [
+        (name.upper(), *DEVICE_LABELS[name], index)
+        for index, name in enumerate(supported_devices())
+    ]
+
+
+def production_device():
+    """Device configured for production tasks, falling back to Auto."""
+    addon_preferences = preferences()
+    configured = getattr(addon_preferences, "device", "")
+    return configured or "AUTO"
+
+
 def jobs_directory():
     path = storage_root() / "jobs"
     path.mkdir(parents=True, exist_ok=True)
@@ -153,13 +223,7 @@ def new_job_paths(job_type):
 
 
 def uv_executable():
-    addon_preferences = preferences()
-    configured = getattr(addon_preferences, "uv_executable", "uv")
-    resolved = shutil.which(configured)
-    if resolved:
-        return resolved
-    if Path(configured).is_file():
-        return str(Path(configured))
+    """Return the bundled uv path inside the shared storage."""
     return str(uv_cache_path())
 
 
@@ -171,9 +235,7 @@ def uv_cache_path():
 
 def install_runtime_command(status_path):
     executable = uv_executable()
-    addon_preferences = preferences()
-    backend = getattr(addon_preferences, "torch_backend", "CU128")
-    pypi_index = getattr(addon_preferences, "pypi_index_url", "") or ""
+    backend = "cu128" if "cuda" in machine_devices() else "cpu"
     script = Path(__file__).with_name("bootstrap_runtime.py")
     command = [
         sys.executable,
@@ -185,7 +247,7 @@ def install_runtime_command(status_path):
         "--models-dir",
         str(models_directory()),
         "--backend",
-        backend.lower(),
+        backend,
         "--da3-commit",
         DA3_COMMIT,
         "--ben2-commit",
@@ -193,8 +255,6 @@ def install_runtime_command(status_path):
         "--status",
         str(status_path),
     ]
-    if pypi_index:
-        command.extend(["--pypi-index", pypi_index])
     return command
 
 
@@ -640,6 +700,7 @@ def remove_background_request(
     model_key,
     video_frames=None,
     start_frame=1,
+    device=None,
 ):
     return HostJobRequest(
         {
@@ -650,7 +711,7 @@ def remove_background_request(
                 "model_dir": str(model_directory(model_key)),
                 "output_dir": str(result_directory),
                 "status": str(status_path),
-                "device": settings.device.lower(),
+                "device": (device or settings.device).lower(),
                 "video_frames": (
                     settings.video_frames
                     if video_frames is None
@@ -701,6 +762,7 @@ def reconstruct_request(
     use_ray_pose=None,
     remove_background=False,
     start_frame=1,
+    device=None,
 ):
     snapform_selected_key = model_key or selected_model_key(settings, export_format)
     snapform_selected_process_res = (
@@ -732,7 +794,7 @@ def reconstruct_request(
         "gs_views_interval": settings.gs_views_interval,
         "video_frames": settings.video_frames,
         "start_frame": start_frame,
-        "device": settings.device.lower(),
+        "device": (device or settings.device).lower(),
         "use_ray_pose": 1 if snapform_selected_ray_pose else 0,
         "ref_view_strategy": settings.ref_view_strategy,
         "remove_background_model_dir": "",

tests/test_addon.py

@@ -1356,6 +1356,104 @@ class AddonTest(unittest.TestCase):
             ["GIANT_1_1", "NESTED_GIANT_LARGE_1_1"],
         )
 
+    def test_device_enums_list_only_supported_devices(self):
+        runtime = importlib.import_module("snapform.runtime")
+        with patch.object(
+            runtime,
+            "supported_devices",
+            return_value=["cuda", "cpu"],
+        ):
+            items = runtime.device_enum_items(None, None)
+
+        self.assertEqual([item[0] for item in items], ["CUDA", "CPU"])
+        self.assertEqual(items[0][2], "NVIDIA CUDA")
+
+    def test_device_enums_default_to_the_fastest_supported_device(self):
+        for class_type in (
+            self.addon.SnapFormSettings,
+            self.addon.SnapFormPreferences,
+        ):
+            with self.subTest(class_type=class_type.__name__):
+                self.assertEqual(
+                    class_type.__annotations__["device"]["default"],
+                    0,
+                )
+
+    def test_preferences_expose_production_device(self):
+        preferences = self.addon.SnapFormPreferences
+
+        self.assertIn("device", preferences.__annotations__)
+        items_callback = preferences.__annotations__["device"]["items"]
+        with patch.object(
+            self.addon.runtime,
+            "supported_devices",
+            return_value=["cpu"],
+        ):
+            items = items_callback(None, None)
+        self.assertEqual([item[0] for item in items], ["CPU"])
+
+    def test_remove_background_poll_requires_ben2_model(self):
+        context = SimpleNamespace(
+            object=SimpleNamespace(
+                type="EMPTY",
+                empty_display_type="IMAGE",
+                data=object(),
+            )
+        )
+        with patch.object(
+            self.addon.runtime,
+            "host_busy",
+            return_value=False,
+        ), patch.object(
+            self.addon.runtime,
+            "model_ready",
+            return_value=False,
+        ):
+            self.assertFalse(self.addon.RemoveImageBackground.poll(context))
+        with patch.object(
+            self.addon.runtime,
+            "host_busy",
+            return_value=False,
+        ), patch.object(
+            self.addon.runtime,
+            "model_ready",
+            return_value=True,
+        ):
+            self.assertTrue(self.addon.RemoveImageBackground.poll(context))
+
+    def test_convert_to_depth_plane_poll_requires_da3_model(self):
+        context = SimpleNamespace(
+            object=SimpleNamespace(
+                type="EMPTY",
+                empty_display_type="IMAGE",
+                data=object(),
+            )
+        )
+        with patch.object(
+            self.addon.runtime,
+            "host_busy",
+            return_value=False,
+        ), patch.object(
+            self.addon.runtime,
+            "model_ready",
+            side_effect=lambda key: key == "BASE",
+        ):
+            self.assertTrue(
+                self.addon.ConvertImageToDepthPlane.poll(context)
+            )
+        with patch.object(
+            self.addon.runtime,
+            "host_busy",
+            return_value=False,
+        ), patch.object(
+            self.addon.runtime,
+            "model_ready",
+            return_value=False,
+        ):
+            self.assertFalse(
+                self.addon.ConvertImageToDepthPlane.poll(context)
+            )
+
     def test_debug_panel_is_closed_by_default(self):
         self.assertIn(
             "DEFAULT_CLOSED",
@@ -1485,6 +1583,95 @@ class AddonTest(unittest.TestCase):
         self.assertEqual(remove_background.call_args.args[1], input_path)
         self.assertEqual(remove_background.call_args.args[4], "BEN2_BASE")
 
+    def test_debug_jobs_use_the_scene_device(self):
+        settings = SimpleNamespace(
+            background_input_path="input.png",
+            device="CPU",
+        )
+        operator = self.addon.RunJob()
+        operator.job_type = self.addon.operators.BACKGROUND_REMOVAL_JOB
+
+        with TemporaryDirectory() as directory:
+            root = Path(directory)
+            input_path = root / "input.png"
+            input_path.touch()
+            settings.background_input_path = str(input_path)
+            paths = {
+                "log": root / "job.log",
+                "status": root / "job.json",
+                "result": root / "result",
+            }
+            with patch.object(
+                self.addon.runtime,
+                "new_job_paths",
+                return_value=paths,
+            ), patch.object(
+                self.addon.runtime,
+                "runtime_ready",
+                return_value=True,
+            ), patch.object(
+                self.addon.runtime,
+                "model_ready",
+                return_value=True,
+            ), patch.object(
+                self.addon.runtime,
+                "remove_background_request",
+                return_value=self.addon.runtime.HostJobRequest(
+                    {"job_id": "test"}
+                ),
+            ) as remove_background:
+                operator._prepare_command(None, settings)
+
+        self.assertEqual(
+            remove_background.call_args.kwargs["device"],
+            "CPU",
+        )
+
+    def test_production_jobs_use_the_preferences_device(self):
+        settings = SimpleNamespace(device="CPU", video_frames=16)
+        operator = self.addon.RunJob()
+        operator.job_type = self.addon.operators.IMAGE_BACKGROUND_JOB
+
+        with TemporaryDirectory() as directory:
+            root = Path(directory)
+            input_path = root / "input.png"
+            input_path.touch()
+            operator.snapform_input_path = str(input_path)
+            paths = {
+                "log": root / "job.log",
+                "status": root / "job.json",
+                "result": root / "result",
+            }
+            with patch.object(
+                self.addon.runtime,
+                "new_job_paths",
+                return_value=paths,
+            ), patch.object(
+                self.addon.runtime,
+                "runtime_ready",
+                return_value=True,
+            ), patch.object(
+                self.addon.runtime,
+                "model_ready",
+                return_value=True,
+            ), patch.object(
+                self.addon.runtime,
+                "production_device",
+                return_value="MPS",
+            ), patch.object(
+                self.addon.runtime,
+                "remove_background_request",
+                return_value=self.addon.runtime.HostJobRequest(
+                    {"job_id": "test"}
+                ),
+            ) as remove_background:
+                operator._prepare_command(None, settings)
+
+        self.assertEqual(
+            remove_background.call_args.kwargs["device"],
+            "MPS",
+        )
+
     def test_image_background_job_uses_operator_input(self):
         settings = SimpleNamespace(device="AUTO")
         operator = self.addon.RunJob()
@@ -1804,8 +1991,8 @@ class AddonTest(unittest.TestCase):
             return_value="C:/uv/uv.exe",
         ), patch.object(
             runtime,
-            "preferences",
-            return_value=SimpleNamespace(torch_backend="CU128"),
+            "machine_devices",
+            return_value=["cuda", "cpu"],
         ), patch.object(
             runtime,
             "runtime_directory",
@@ -1821,11 +2008,15 @@ class AddonTest(unittest.TestCase):
         self.assertEqual(command[0], sys.executable)
         self.assertIn("--uv", command)
         self.assertIn("C:/uv/uv.exe", command)
+        self.assertEqual(
+            command[command.index("--backend") + 1],
+            "cu128",
+        )
         self.assertTrue(
             any("bootstrap_runtime.py" in str(value) for value in command)
         )
 
-    def test_install_runtime_command_passes_configured_pypi_index(self):
+    def test_install_runtime_command_auto_selects_backend(self):
         runtime = importlib.import_module("snapform.runtime")
         with patch.object(
             runtime,
@@ -1833,11 +2024,8 @@ class AddonTest(unittest.TestCase):
             return_value="C:/uv/uv.exe",
         ), patch.object(
             runtime,
-            "preferences",
-            return_value=SimpleNamespace(
-                torch_backend="CU128",
-                pypi_index_url="https://pypi.tuna.tsinghua.edu.cn/simple",
-            ),
+            "machine_devices",
+            return_value=["cpu"],
         ), patch.object(
             runtime,
             "runtime_directory",
@@ -1849,11 +2037,12 @@ class AddonTest(unittest.TestCase):
         ):
             command = runtime.install_runtime_command(Path("status.json"))
 
-        index_position = command.index("--pypi-index")
+        self.assertNotIn("--pypi-index", command)
         self.assertEqual(
-            command[index_position + 1],
-            "https://pypi.tuna.tsinghua.edu.cn/simple",
+            command[command.index("--backend") + 1],
+            "cpu",
         )
+        self.assertIn("C:/uv/uv.exe", command)
 
     def test_completed_export_opens_job_directory(self):
         calls = []

tests/test_ai_host.py

@@ -109,7 +109,7 @@ class HostTest(unittest.TestCase):
 
         self.assertIs(first, second)
         self.assertEqual(loads, ["model"])
-        self.assertEqual(cache.status()["da3"], "da3_model")
+        self.assertEqual(cache.status()["da3"], "model")
 
     def test_host_tracks_clients_and_rejects_shared_shutdown(self):
         fake_models = SimpleNamespace(

tests/test_bootstrap_runtime.py

@@ -62,7 +62,11 @@ class BootstrapRuntimeTest(unittest.TestCase):
             ) as mocked_run, patch.object(
                 bootstrap,
                 "run_install",
-            ) as mocked_run_install:
+            ) as mocked_run_install, patch.object(
+                bootstrap,
+                "run_verify",
+                return_value={},
+            ):
                 bootstrap.install(args)
 
         commands = [call.args[0] for call in mocked_run.call_args_list]
@@ -113,10 +117,11 @@ class BootstrapRuntimeTest(unittest.TestCase):
         self.assertIn(f"uvicorn=={bootstrap.UVICORN_VERSION}", host_command)
         self.assertIn("setuptools", host_command)
         self.assertNotIn("ninja", host_command)
-        check_command = next(
-            command
-            for command in commands
-            if any("import torch;" in str(value) for value in command)
+        check_command = bootstrap.verify_command(
+            bootstrap.environment_python(runtime_dir)
+        )
+        self.assertTrue(
+            any("import torch;" in str(value) for value in check_command)
         )
         self.assertTrue(
             any(
@@ -130,6 +135,12 @@ class BootstrapRuntimeTest(unittest.TestCase):
         self.assertFalse(
             any("import gsplat" in str(value) for value in check_command)
         )
+        self.assertTrue(
+            any(
+                "SNAPFORM_DEVICES=" in str(value)
+                for value in check_command
+            )
+        )
         default_downloads = [
             command
             for command in commands
@@ -190,7 +201,11 @@ class BootstrapRuntimeTest(unittest.TestCase):
             ), patch.object(
                 bootstrap,
                 "run_install",
-            ) as mocked_run_install:
+            ) as mocked_run_install, patch.object(
+                bootstrap,
+                "run_verify",
+                return_value={},
+            ):
                 bootstrap.install(args)
 
         commands = [
@@ -372,7 +387,11 @@ class BootstrapRuntimeTest(unittest.TestCase):
                 bootstrap,
                 "run_stage",
                 side_effect=fake_run_stage,
-            ), patch.object(bootstrap, "run"):
+            ), patch.object(bootstrap, "run"), patch.object(
+                bootstrap,
+                "run_verify",
+                return_value={},
+            ):
                 bootstrap.install(args)
 
             starts = [item[1] for item in stage_ranges]
@@ -604,34 +623,6 @@ class BootstrapRuntimeTest(unittest.TestCase):
             "https://pypi.tuna.tsinghua.edu.cn/simple",
         )
 
-    def test_child_environment_injects_registry_proxy(self):
-        bootstrap = load_bootstrap_runtime()
-        with patch.object(
-            bootstrap.urllib.request,
-            "getproxies",
-            return_value={
-                "http": "http://proxy:8080",
-                "https": "http://proxy:8080",
-            },
-        ):
-            env = bootstrap.child_environment()
-
-        self.assertEqual(env["HTTP_PROXY"], "http://proxy:8080")
-        self.assertEqual(env["HTTPS_PROXY"], "http://proxy:8080")
-
-    def test_child_environment_keeps_explicit_proxy(self):
-        bootstrap = load_bootstrap_runtime()
-        with patch.object(
-            bootstrap.urllib.request,
-            "getproxies",
-            return_value={"https": "http://proxy:8080"},
-        ):
-            env = bootstrap.child_environment(
-                {"HTTPS_PROXY": "http://explicit:3128"}
-            )
-
-        self.assertEqual(env["HTTPS_PROXY"], "http://explicit:3128")
-
     def test_ready_json_records_installed_packages(self):
         bootstrap = load_bootstrap_runtime()
         with tempfile.TemporaryDirectory() as directory:
@@ -657,6 +648,10 @@ class BootstrapRuntimeTest(unittest.TestCase):
                 bootstrap,
                 "run_stage",
                 return_value=["torch==2.9.1+cu128", "numpy==1.26.4"],
+            ), patch.object(
+                bootstrap,
+                "run_verify",
+                return_value={"cuda": True, "mps": False},
             ):
                 bootstrap.install(args)
 
@@ -669,6 +664,7 @@ class BootstrapRuntimeTest(unittest.TestCase):
                 ready["packages"],
                 ["numpy==1.26.4", "torch==2.9.1+cu128"],
             )
+            self.assertEqual(ready["devices"], ["cuda", "cpu"])
 
     def test_broken_runtime_is_cleaned_before_reinstall(self):
         bootstrap = load_bootstrap_runtime()
@@ -693,6 +689,10 @@ class BootstrapRuntimeTest(unittest.TestCase):
             ), patch.object(
                 bootstrap,
                 "run_install",
+            ), patch.object(
+                bootstrap,
+                "run_verify",
+                return_value={},
             ):
                 bootstrap.install(args)
 
@@ -728,6 +728,10 @@ class BootstrapRuntimeTest(unittest.TestCase):
             ), patch.object(
                 bootstrap,
                 "run_install",
+            ), patch.object(
+                bootstrap,
+                "run_verify",
+                return_value={},
             ):
                 bootstrap.install(args)
 
@@ -758,7 +762,11 @@ class BootstrapRuntimeTest(unittest.TestCase):
                 bootstrap,
                 "run",
                 side_effect=fake_run,
-            ), patch.object(bootstrap, "run_install"):
+            ), patch.object(bootstrap, "run_install"), patch.object(
+                bootstrap,
+                "run_verify",
+                return_value={},
+            ):
                 bootstrap.install(args)
 
             self.assertTrue((runtime_dir / "snapform_ready.json").is_file())
@@ -872,6 +880,10 @@ class BootstrapRuntimeTest(unittest.TestCase):
                 bootstrap,
                 "run_install",
                 side_effect=fake_run_install,
+            ), patch.object(
+                bootstrap,
+                "run_verify",
+                return_value={},
             ):
                 bootstrap.install(args)
 
@@ -918,6 +930,10 @@ class BootstrapRuntimeTest(unittest.TestCase):
             ), patch.object(
                 bootstrap,
                 "run_install",
+            ), patch.object(
+                bootstrap,
+                "run_verify",
+                return_value={},
             ):
                 bootstrap.install(args)
 
@@ -995,6 +1011,110 @@ class BootstrapRuntimeTest(unittest.TestCase):
         self.assertIn(f"depth-anything-3 @ {da3_local}", ai_command)
         self.assertIn(f"ben2 @ {ben2_local}", ai_command)
 
+    def test_install_stages_download_models_into_prefixed_directories(self):
+        bootstrap = load_bootstrap_runtime()
+        with tempfile.TemporaryDirectory() as directory:
+            root = Path(directory)
+            runtime_dir = root / "runtime"
+            runtime_dir.mkdir()
+            sources = root / "sources"
+            sources.mkdir()
+            (sources / "depth-anything-3.tar.gz").write_bytes(b"da3")
+            (sources / "ben2.tar.gz").write_bytes(b"ben2")
+            uv_path = root / "uv"
+            uv_path.write_bytes(b"uv")
+            args = SimpleNamespace(
+                uv=str(uv_path),
+                runtime_dir=str(runtime_dir),
+                models_dir=str(root / "models"),
+                backend="cu128",
+                da3_commit="test-commit",
+                ben2_commit="test-ben2-commit",
+                status=str(root / "status.json"),
+            )
+
+            stages = bootstrap.install_stages(
+                args,
+                runtime_dir,
+                root / "python",
+                uv_path,
+            )
+
+        downloads = {
+            label: command
+            for label, command, _environment, kind in stages
+            if kind == "download"
+        }
+        self.assertIn("da3_small", " ".join(map(str, downloads["da3 small"])))
+        self.assertIn("da3_base", " ".join(map(str, downloads["da3 base"])))
+        self.assertIn("ben2_base", " ".join(map(str, downloads["ben2 base"])))
+
+    def test_run_verify_returns_detected_devices(self):
+        bootstrap = load_bootstrap_runtime()
+
+        class FakeProcess:
+            def __init__(self, lines):
+                self.stdout = lines
+
+            def wait(self):
+                return 0
+
+        fake_process = FakeProcess(
+            ['SNAPFORM_DEVICES={"cuda": false, "mps": true}']
+        )
+        with patch.object(
+            bootstrap.subprocess,
+            "Popen",
+            return_value=fake_process,
+        ):
+            report = bootstrap.run_verify(
+                "python",
+                "status.json",
+                clock=lambda: 0.0,
+            )
+
+        self.assertFalse(report["cuda"])
+        self.assertTrue(report["mps"])
+
+    def test_run_verify_writes_heartbeat_progress(self):
+        bootstrap = load_bootstrap_runtime()
+
+        class FakeProcess:
+            def __init__(self, lines):
+                self.stdout = lines
+
+            def wait(self):
+                return 0
+
+        clock_state = {"now": 0.0}
+
+        def fake_clock():
+            clock_state["now"] += 6.0
+            return clock_state["now"]
+
+        def delayed_empty_lines():
+            time.sleep(0.6)
+            if False:
+                yield
+
+        fake_process = FakeProcess(delayed_empty_lines())
+        with patch.object(
+            bootstrap.subprocess,
+            "Popen",
+            return_value=fake_process,
+        ), patch.object(bootstrap, "write_status") as mocked_write:
+            report = bootstrap.run_verify(
+                "python",
+                "status.json",
+                clock=fake_clock,
+            )
+
+        messages = [call.args[2] for call in mocked_write.call_args_list]
+        self.assertTrue(
+            any("verifying runtime" in message for message in messages)
+        )
+        self.assertEqual(report, {})
+
     def test_install_uv_extracts_bundled_archive(self):
         bootstrap = load_bootstrap_runtime()
         with tempfile.TemporaryDirectory() as directory:

tests/test_image_depth.py

@@ -209,6 +209,21 @@ class ImageDepthTest(unittest.TestCase):
             (1, 1),
         )
 
+    def test_extract_frame_range_honors_single_frame_setting(self):
+        source_object = SimpleNamespace(
+            data=SimpleNamespace(source="MOVIE", frame_duration=97),
+            image_user=SimpleNamespace(
+                frame_start=1,
+                frame_offset=70,
+                frame_duration=1,
+            ),
+        )
+
+        self.assertEqual(
+            self.image_depth.extract_frame_range(source_object),
+            (71, 1),
+        )
+
     def test_configure_image_user_preserves_cyclic_setting(self):
         image_user = SimpleNamespace(
             frame_start=0,

tests/test_models.py

@@ -54,3 +54,21 @@ class ModelTest(unittest.TestCase):
             ("config.json", "model.safetensors"),
         )
         self.assertNotIn("BEN2_BASE", self.models.MODELS)
+
+    def test_da3_model_directories_share_the_da3_prefix(self):
+        for model_key, model in self.models.MODELS.items():
+            with self.subTest(model_key=model_key):
+                self.assertTrue(
+                    model.directory_name.startswith("da3_"),
+                    model.directory_name,
+                )
+                self.assertEqual(
+                    self.models.model_directory_name(model_key),
+                    model.directory_name,
+                )
+
+    def test_ben2_model_directory_uses_ben2_prefix(self):
+        self.assertEqual(
+            self.models.model_directory_name("BEN2_BASE"),
+            "ben2_base",
+        )

tests/test_worker.py

@@ -385,79 +385,6 @@ class WorkerTest(unittest.TestCase):
             with self.assertRaisesRegex(RuntimeError, "Unsupported input type"):
                 self.worker.prepare_input(input_path, 16, directory)
 
-    def test_download_passes_optional_file_patterns(self):
-        calls = []
-        fake_hugging_face = ModuleType("huggingface_hub")
-        fake_hugging_face.snapshot_download = lambda **options: calls.append(
-            options
-        )
-        args = SimpleNamespace(
-            repository="PramaLLC/BEN2",
-            model_dir="model",
-            status="status.json",
-            allow_pattern=["config.json", "model.safetensors"],
-        )
-
-        with tempfile.TemporaryDirectory() as directory, patch.dict(
-            sys.modules,
-            {"huggingface_hub": fake_hugging_face},
-        ), patch.object(
-            self.worker,
-            "create_download_progress_class",
-            return_value=object,
-        ):
-            args.model_dir = str(Path(directory) / "model")
-            args.status = str(Path(directory) / "status.json")
-            self.worker.download_model(args)
-
-        self.assertEqual(
-            calls[0]["allow_patterns"],
-            ["config.json", "model.safetensors"],
-        )
-        self.assertEqual(calls[0]["endpoint"], self.worker.HF_ENDPOINT)
-
-    def test_download_model_falls_back_to_modelscope_when_direct_fails(self):
-        calls = []
-        fake_hugging_face = ModuleType("huggingface_hub")
-
-        def fake_snapshot_download(**options):
-            calls.append(options)
-            raise OSError("Connection reset by peer")
-
-        fake_hugging_face.snapshot_download = fake_snapshot_download
-        args = SimpleNamespace(
-            repository="PramaLLC/BEN2",
-            model_dir="model",
-            status="status.json",
-            allow_pattern=[],
-        )
-
-        with tempfile.TemporaryDirectory() as directory, patch.dict(
-            sys.modules,
-            {"huggingface_hub": fake_hugging_face},
-        ), patch.object(
-            self.worker,
-            "create_download_progress_class",
-            return_value=object,
-        ), patch.dict(os.environ, {"HF_ENDPOINT": ""}), patch.object(
-            self.worker,
-            "download_from_modelscope",
-        ) as mocked_modelscope:
-            def fake_download_from_modelscope(*args, **kwargs):
-                model_dir = Path(args[1])
-                model_dir.mkdir(parents=True, exist_ok=True)
-                (model_dir / "config.json").write_text("{}")
-                (model_dir / "model.safetensors").write_bytes(b"x")
-
-            mocked_modelscope.side_effect = fake_download_from_modelscope
-            args.model_dir = str(Path(directory) / "model")
-            args.status = str(Path(directory) / "status.json")
-            self.worker.download_model(args)
-
-        self.assertEqual(len(calls), 1)
-        self.assertEqual(calls[0]["endpoint"], self.worker.HF_ENDPOINT)
-        self.assertEqual(mocked_modelscope.call_count, 1)
-
     def test_download_model_skips_unreachable_endpoint(self):
         calls = []
         fake_hugging_face = ModuleType("huggingface_hub")

uv.lock

@@ -2,7 +2,78 @@ version = 1
 revision = 3
 requires-python = ">=3.11"
 
+[[package]]
+name = "colorama"
+version = "0.4.6"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/d8/53/6f443c9a4a8358a93a6792e2acffb9d9d5cb0a5cfd8802644b7b1c9a02e4/colorama-0.4.6.tar.gz", hash = "sha256:08695f5cb7ed6e0531a20572697297273c47b8cae5a63ffc6d6ed5c201be6e44", size = 27697, upload-time = "2022-10-25T02:36:22.414Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/d1/d6/3965ed04c63042e047cb6a3e6ed1a63a35087b6a609aa3a15ed8ac56c221/colorama-0.4.6-py2.py3-none-any.whl", hash = "sha256:4f1d9991f5acc0ca119f9d443620b77f9d6b33703e51011c16baf57afb285fc6", size = 25335, upload-time = "2022-10-25T02:36:20.889Z" },
+]
+
+[[package]]
+name = "iniconfig"
+version = "2.3.0"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/72/34/14ca021ce8e5dfedc35312d08ba8bf51fdd999c576889fc2c24cb97f4f10/iniconfig-2.3.0.tar.gz", hash = "sha256:c76315c77db068650d49c5b56314774a7804df16fee4402c1f19d6d15d8c4730", size = 20503, upload-time = "2025-10-18T21:55:43.219Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/cb/b1/3846dd7f199d53cb17f49cba7e651e9ce294d8497c8c150530ed11865bb8/iniconfig-2.3.0-py3-none-any.whl", hash = "sha256:f631c04d2c48c52b84d0d0549c99ff3859c98df65b3101406327ecc7d53fbf12", size = 7484, upload-time = "2025-10-18T21:55:41.639Z" },
+]
+
+[[package]]
+name = "packaging"
+version = "26.3"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/7d/fa/3944b40b07da9ce895c0e6303a5ab7d53da063554f534556b134a54d6093/packaging-26.3.tar.gz", hash = "sha256:94edc256424af38762eb31306eed28beb9f0efc50a8837492c9d6fd6004aed79", size = 313412, upload-time = "2026-08-04T18:15:28.737Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/63/34/ba1c580383c9eada3711951fef0795c80b829a078d72188184bcab9dd527/packaging-26.3-py3-none-any.whl", hash = "sha256:d7193f7c8e4e93f444fde0262bf90af30e16fa0ad0ad44cb553c87339b23cd1c", size = 129956, upload-time = "2026-08-04T18:15:27.159Z" },
+]
+
+[[package]]
+name = "pluggy"
+version = "1.6.0"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/f9/e2/3e91f31a7d2b083fe6ef3fa267035b518369d9511ffab804f839851d2779/pluggy-1.6.0.tar.gz", hash = "sha256:7dcc130b76258d33b90f61b658791dede3486c3e6bfb003ee5c9bfb396dd22f3", size = 69412, upload-time = "2025-05-15T12:30:07.975Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/54/20/4d324d65cc6d9205fabedc306948156824eb9f0ee1633355a8f7ec5c66bf/pluggy-1.6.0-py3-none-any.whl", hash = "sha256:e920276dd6813095e9377c0bc5566d94c932c33b27a3e3945d8389c374dd4746", size = 20538, upload-time = "2025-05-15T12:30:06.134Z" },
+]
+
+[[package]]
+name = "pygments"
+version = "2.20.0"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/c3/b2/bc9c9196916376152d655522fdcebac55e66de6603a76a02bca1b6414f6c/pygments-2.20.0.tar.gz", hash = "sha256:6757cd03768053ff99f3039c1a36d6c0aa0b263438fcab17520b30a303a82b5f", size = 4955991, upload-time = "2026-03-29T13:29:33.898Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/f4/7e/a72dd26f3b0f4f2bf1dd8923c85f7ceb43172af56d63c7383eb62b332364/pygments-2.20.0-py3-none-any.whl", hash = "sha256:81a9e26dd42fd28a23a2d169d86d7ac03b46e2f8b59ed4698fb4785f946d0176", size = 1231151, upload-time = "2026-03-29T13:29:30.038Z" },
+]
+
+[[package]]
+name = "pytest"
+version = "9.1.1"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "colorama", marker = "sys_platform == 'win32'" },
+    { name = "iniconfig" },
+    { name = "packaging" },
+    { name = "pluggy" },
+    { name = "pygments" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/e4/47/b9efed96c114afcfa3c9d3fe98a76a1d14c74a9e266d397cf6eb64be5e01/pytest-9.1.1.tar.gz", hash = "sha256:1088fbde8f2b49d95a549a195707afa7a76a3ce9bcadc26b6d71f0ffda5fe313", size = 1636369, upload-time = "2026-06-19T10:58:32.857Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/24/25/1de2678b631f5a49215c6c96fff41ba892b0a34df68d6d80292b1b48aa7f/pytest-9.1.1-py3-none-any.whl", hash = "sha256:37a86b45efb9a47a61a36449063e8e18d0cab3161329fc099eb21783169c4f0c", size = 386536, upload-time = "2026-06-19T10:58:31.347Z" },
+]
+
 [[package]]
 name = "snapform"
 version = "0.10.10"
 source = { virtual = "." }
+
+[package.dev-dependencies]
+dev = [
+    { name = "pytest" },
+]
+
+[package.metadata]
+
+[package.metadata.requires-dev]
+dev = [{ name = "pytest" }]