feat: Add BEN2 background removal functionality and integrate into SnapForm

- Introduced BEN2 model for background removal in the panel and properties.
- Updated runtime to handle background removal commands and model downloads.
- Implemented background removal logic in worker, including image processing and saving results.
- Enhanced tests to cover new background removal features and ensure proper integration.
- Updated version to 0.9.28 to reflect new features and improvements.

README.md

@@ -3,7 +3,8 @@
 `SnapForm` 是一套从图片快速构建空间结构的 Blender Extension。目前以
 [Depth Anything 3](https://github.com/ByteDance-Seed/Depth-Anything-3)
 作为首个 AI 能力，在独立 Runtime 中运行 DA3，可将官方导出结果保存在独立 job 目录，
-也可把 Image Empty 转换成带几何与材质置换的立体面片。
+也可把 Image Empty 转换成带几何与材质置换的立体面片。Debug 面板还可运行
+[BEN2](https://github.com/PramaLLC/BEN2) 背景剔除功能测试。
 
 ## 当前功能
 
@@ -11,11 +12,12 @@
 - 在独立 Python Runtime 中运行 Torch 与 DA3，避免污染 Blender Python 进程
 - Runtime、模型和任务结果默认保存在 `~/.snapform`，可由多个 Blender 版本共享
 - 支持 DA3 Small、Base、Large 1.1、Giant 1.1 和 Nested Giant Large 1.1
+- 支持 BEN2 Base 单图背景剔除测试，输出透明 PNG 与推理耗时
 - 支持单张 JPG、JPEG、PNG、WEBP 图片或常见视频文件输入
 - 视频均匀抽帧，可设置参与重建的最大帧数
 - `Setup` 集中管理 Runtime，并逐项列出和下载所有可用模型
 - Runtime 已就绪时不额外占用面板空间；模型列表每个模型只占一行
-- `Debug` 是独立且默认折叠的 DA3 测试面板
+- `Debug` 是独立且默认折叠的 AI 功能测试面板
 - 使用一个导出类型下拉菜单和一个 `Export` 按钮
 - 只显示当前格式的专属参数；GLB 与 PLY 设置不会同时出现
 - Debug 的模型选项按格式过滤；PLY 只提供两个带 Gaussian head 的 Giant 模型
@@ -23,10 +25,11 @@
   `(Missing)`，已下载模型保持原名
 - 分别调用 DA3 的 `glb`、`gs_ply`、`depth_vis` 和 `mini_npz` exporter
 - Image Empty 右键菜单可直接运行 DA3，根据 `mini_npz` 深度生成面片
+- Image Empty 右键菜单可直接运行 BEN2，保留原图并换用新建的透明背景图片
 - 使用浮点深度贴图、Simple Subdivision、Displace Modifier 和原图 Bump
   构建立体面片
 - 转换对话框可选择推理模式、处理分辨率和面片网格细分
-- Debug 导出完成后只打开对应 job 文件夹，不向当前 Blender 文件添加任何数据
+- Debug 的 DA3 导出或 BEN2 背景剔除完成后只打开对应 job 文件夹，不向当前 Blender 文件添加任何数据
 - 完整保留每次任务生成的所有官方文件
 - 安装、模型下载和推理均在后台子进程运行，状态栏实时显示阶段、百分比进度条与取消按钮
 - 后台任务运行时，Install、Download 和 Export 按钮保持原位置并变灰，任务完成后恢复
@@ -36,11 +39,27 @@
 1. 使用 `uv run python pack.py` 构建 Extension ZIP。
 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。
-4. 点击 `Install Runtime`。首次安装会下载 Python、PyTorch 和 DA3 依赖。
+4. 点击 `Install Runtime`。首次安装会下载 Python、PyTorch、DA3 和 BEN2 依赖。
 5. 在 3D View 按 `N`，打开 `SnapForm` 页签。
 6. 在 `Setup` 中找到需要的模型并点击该行的 `Download`。
 7. 展开默认收起的 `Debug`，选择 `Export Format` 和本次执行使用的模型。
 8. 选择一张图片或一个视频文件。视频可设置均匀抽取的帧数，然后点击 `Export`。
+9. 测试背景剔除时，在 `Setup` 下载 `BEN2 Base`，然后在 `Debug` 的
+   `BEN2 Background Removal` 区域选择单张图片并点击 `Remove Background`。
+   完成后会打开 job 文件夹，其中的 `foreground.png` 带透明 Alpha；状态栏显示
+   BEN2 推理、模型加载与任务总耗时。当前测试固定关闭较慢的 foreground refinement。
+
+## Image Empty 去背景
+
+1. 在 `Setup` 下载 `BEN2 Base`。
+2. 在 3D View 选中 Image Empty，右键选择 `Remove Background`。
+3. 完成后以 Fake User 保留原 Image datablock，新建并 Pack 一张透明 PNG 到当前 `.blend`，然后只把
+   当前 Empty 的 `data` 换成新图；Empty 的 Transform、显示尺寸、Offset 及其他设置保持不变。
+4. 新 Image 使用原图去掉格式后再追加 `.rb` 的名称。例如 `img.png` 生成 `img.rb`，
+   `img.png.001` 生成 `img.001.rb`。已有同名 datablock 时由 Blender 自动追加编号。
+
+原图已 Pack 时会先写出临时 PNG 交给 BEN2，任务结束后删除临时输入；原 Image 仍保持 Pack
+状态和原始内容。BEN2 的 `foreground.png` 也会保留在对应 job 目录中。
 
 ## Image Empty 转深度面片
 
@@ -52,11 +71,17 @@
      转换导出 `mini_npz`，使用前需要在
      `Setup` 下载所选模型，并固定使用 `process_res_method="upper_bound_resize"`；设备和参考视图策略
      继续使用场景设置。
-   - `Original`：直接把原图作为深度贴图，不显示估算参数，也不启动 Runtime 或要求源图片存在于磁盘。
-   `Original` 是默认深度来源。选择 `AI Generated` 时，其 Model、Use Ray Pose 和 Resolution Scale
-   参数会集中显示在独立边框中。两种模式始终显示 Displace via、Subdivision 与 Delete Original。
+   - `Original`：直接把原图作为深度贴图，不显示估算参数。Remove Background 关闭时不启动
+     Runtime，也不要求源图片存在于磁盘。
+   `Original` 是默认深度来源。选择 `AI Generated` 时，其 Model、Use Ray Pose 与 Resolution Scale
+   参数会集中显示在独立边框中。两种模式始终显示 Remove Background、Displace via、Unshaded、
+   Subdivision 与 Delete Original。
+   Remove Background 默认关闭，与 Depth from 无关；开启后需要下载 `BEN2 Base`，并从原图生成、
+   Pack `.rb` 透明图片。它只替换生成面片材质的颜色图片，不修改源 Image Empty 的 Image datablock。
+   AI Generated 仍从原图生成 `.depth`；例如 `img.png` 会得到 `img.rb` 与 `img.depth`，不会生成
+   `img.rb.depth`。Original 仍用原图作为位移来源，不改变原有深度逻辑。
    Displace via 可选 `Modifier`（默认）或 `Material`；
-   Subdivision 默认为 `5`，并默认删除原对象。
+   Unshaded 默认关闭；Subdivision 默认为 `5`，并默认删除原对象。
 3. AI Generated 完成后会自动创建面片。面片几何先由 `mini_npz` 的相机内参与深度确定，再通过对象
    Transform 匹配原 Image Empty 的可见尺寸、位置、旋转和 Offset。Original 模式直接使用源 Empty
    的可见尺寸、Offset 与 Transform。开启 `Delete Original` 时，成功生成后
@@ -105,6 +130,9 @@ Principled BSDF 的 `Base Color`，并通过
 `0.001`，会乘以面片 Transform 的统一缩放倍率。原图 Color 还会经过 `Float Curve` 连接到
 Principled BSDF 的 `Roughness`，曲线两端输出分别为 `0.5` 和 `0.1`；材质 IOR 保持
 Principled BSDF 默认值 `1.5`。
+开启 `Unshaded` 后，Surface 改为原图 Color 直接连接 Emission，再连接 Material Output；不创建
+Principled BSDF、Bump 或 Float Curve。若 Displace via 选择 `Material`，深度贴图与 Displacement
+节点仍会保留并连接 Material Output 的 Displacement。
 
 这里的“原始距离”是 DA3 模型输出的尺度。普通 Small、Base、Large 和 Giant 模型的深度不应
 被解释为现实世界的米；需要现实尺度时应选择带 metric depth 的
@@ -116,6 +144,8 @@ Principled BSDF 默认值 `1.5`。
 `Convert to Depth Plane` 会按当前 Depth from 再次转换。
 面片、深度 Image、Texture 与 Material 会在同一个 Blender Undo 步骤中创建；转换完成后执行
 `Ctrl+Z` 会完整撤销本次转换，不会越过转换去回退更早的 Image Empty 操作。
+开启 Remove Background 时，`.rb` Image 的创建、Pack 和深度面片生成也属于
+同一个 Undo 步骤。
 
 Image Empty 使用 Packed Image 时，SnapForm 会通过 Blender 写出临时 PNG 交给 DA3，
 不会 Unpack 或修改源 Image；任务结束后自动删除临时文件。普通图片仍直接读取源文件。DA3 的原始
@@ -129,7 +159,7 @@ Image Empty 使用 Packed Image 时，SnapForm 会通过 Blender 写出临时 PN
 `DA3 Nested Giant Large 1.1`。Small、Base 和 Large 没有 Gaussian head，无法由
 DA3 生成 `gs_ply`。
 
-每次任务会在 `~/.snapform/jobs/<任务目录>/` 中保留 DA3 官方产物：
+每次任务会在 `~/.snapform/jobs/<任务目录>/` 中保留输出产物：
 
 ```text
 <GLB 任务目录>/
@@ -143,6 +173,9 @@ DA3 生成 `gs_ply`。
 
 <Mini NPZ 任务目录>/
 └── exports/mini_npz/results.npz
+
+<BEN2 背景剔除任务目录>/
+└── foreground.png
 ```
 
 建议从 `DA3 Base` 开始。它使用 Apache 2.0 模型权重；`DA3 Large 1.1` 权重采用
@@ -185,6 +218,7 @@ Extension 本身不包含 Torch。`Install Runtime` 使用 `uv` 创建独立 Pyt
 - TorchVision 0.24.1
 - xFormers 0.0.33.post2（与 PyTorch 2.9.1 对应）
 - 固定到已验证 commit 的 Depth Anything 3
+- 固定到已验证 commit 的 BEN2
 - 固定到 DA3 官方推荐 commit 的 `gsplat`；仅安装 PLY 导出所需的纯 Python 部分，
   不编译本机 CUDA rasterizer
 - `addict` 2.4.0（补充 DA3 上游包元数据暂未声明的运行依赖）
@@ -236,5 +270,6 @@ uv run python pack.py
 ## 上游与许可证
 
 - Depth Anything 3 代码采用 Apache 2.0。
+- BEN2 代码与模型权重采用 MIT License。
 - 本 Extension 代码采用 GPL-3.0-or-later。
 - 模型权重许可证由所选 Hugging Face 模型仓库决定，Extension 下载前会在界面显示。

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "snapform"
-version = "0.9.27"
+version = "0.9.28"
 description = "Fast image-guided structure building tools for Blender"
 readme = "README.md"
 requires-python = ">=3.11"

src/snapform/__init__.py

@@ -1,8 +1,11 @@
 import bpy
 
 from .image_depth import (
+    ApplyRemovedBackground,
     BuildDepthPlane,
+    BuildOriginalDepthPlane,
     ConvertImageToDepthPlane,
+    RemoveImageBackground,
     draw_image_context_menu,
 )
 from .operators import (
@@ -22,7 +25,10 @@ CLASSES = (
     SnapFormSettings,
     RunJob,
     CancelJob,
+    ApplyRemovedBackground,
+    RemoveImageBackground,
     BuildDepthPlane,
+    BuildOriginalDepthPlane,
     ConvertImageToDepthPlane,
     RuntimeModelsPanel,
     DebugPanel,

src/snapform/blender_manifest.toml

@@ -1,7 +1,7 @@
 schema_version = "1.0.0"
 
 id = "snapform"
-version = "0.9.27"
+version = "0.9.28"
 name = "SnapForm"
 tagline = "Build spatial structures from images, fast"
 maintainer = "OmooLab"

src/snapform/bootstrap_runtime.py

@@ -11,7 +11,7 @@ TORCHVISION_VERSION = "0.24.1"
 XFORMERS_VERSION = "0.0.33.post2"
 ADDICT_VERSION = "2.4.0"
 GSPLAT_COMMIT = "0b4dddf04cb687367602c01196913cde6a743d70"
-RUNTIME_SCHEMA = 2
+RUNTIME_SCHEMA = 3
 TORCH_INDEXES = {
     "cu128": "https://download.pytorch.org/whl/cu128",
     "cpu": "https://download.pytorch.org/whl/cpu",
@@ -113,6 +113,22 @@ def install(args):
         ]
     )
 
+    write_status(args.status, 0.68, "Installing BEN2")
+    ben2_requirement = (
+        "ben2 @ git+https://github.com/PramaLLC/BEN2.git"
+        f"@{args.ben2_commit}"
+    )
+    run(
+        [
+            args.uv,
+            "pip",
+            "install",
+            "--python",
+            python,
+            ben2_requirement,
+        ]
+    )
+
     write_status(args.status, 0.78, "Installing gsplat build tools")
     run(
         [
@@ -154,6 +170,7 @@ def install(args):
             (
                 "import torch; "
                 "from depth_anything_3.api import DepthAnything3; "
+                "from ben2 import BEN_Base; "
                 "import gsplat; "
                 "print(torch.__version__, torch.cuda.is_available())"
             ),
@@ -170,6 +187,7 @@ def install(args):
                 "xformers": XFORMERS_VERSION,
                 "backend": args.backend,
                 "da3_commit": args.da3_commit,
+                "ben2_commit": args.ben2_commit,
                 "gsplat_commit": GSPLAT_COMMIT,
                 "gsplat_cuda": False,
             },
@@ -186,6 +204,7 @@ def parse_args():
     parser.add_argument("--runtime-dir", required=True)
     parser.add_argument("--backend", choices=tuple(TORCH_INDEXES), required=True)
     parser.add_argument("--da3-commit", required=True)
+    parser.add_argument("--ben2-commit", required=True)
     parser.add_argument("--status", required=True)
     return parser.parse_args()
 

src/snapform/image_depth.py

@@ -80,6 +80,10 @@ def depth_data_name(source_object):
     return f"{image_base_name(source_object.data)}.depth"
 
 
+def removed_background_data_name(source_object):
+    return f"{image_base_name(source_object.data)}.rb"
+
+
 def scaled_process_resolution(image, scale):
     width, height = image.size
     if width <= 0 or height <= 0:
@@ -108,6 +112,49 @@ def save_packed_image(image, scene):
     return output_path
 
 
+def prepare_image_input(image, scene):
+    source_path = image_path(image)
+    if image_is_packed(image):
+        return save_packed_image(image, scene), True
+    if not source_path.is_file():
+        raise RuntimeError(
+            "The image empty must reference an existing image file"
+        )
+    return source_path, False
+
+
+def create_removed_background_image(
+    source_object,
+    output_path,
+    assign_to_source=True,
+):
+    output_path = Path(output_path)
+    if not output_path.is_file():
+        raise RuntimeError(
+            f"Background removal result does not exist: {output_path}"
+        )
+    try:
+        result_image = bpy.data.images.load(
+            str(output_path),
+            check_existing=False,
+        )
+    except RuntimeError as error:
+        raise RuntimeError("Unable to load the background result") from error
+    source_image = source_object.data
+    original_fake_user = source_image.use_fake_user
+    try:
+        result_image.name = removed_background_data_name(source_object)
+        result_image.pack()
+        if assign_to_source:
+            source_image.use_fake_user = True
+            source_object.data = result_image
+    except (OSError, RuntimeError) as error:
+        source_image.use_fake_user = original_fake_user
+        bpy.data.images.remove(result_image)
+        raise RuntimeError("Unable to pack the background result") from error
+    return result_image
+
+
 def plane_bounds(source_object):
     image_width, image_height = source_object.data.size
     if image_width <= 0 or image_height <= 0:
@@ -422,6 +469,7 @@ def create_material(
     displacement_image=None,
     displacement_midlevel=0.0,
     displacement_scale=1.0,
+    unshaded=False,
 ):
     material = bpy.data.materials.new(name)
     material.use_nodes = True
@@ -430,15 +478,47 @@ def create_material(
     nodes = material.node_tree.nodes
     nodes.clear()
 
-    output = nodes.new("ShaderNodeOutputMaterial")
-    shader = nodes.new("ShaderNodeBsdfPrincipled")
     color_texture = nodes.new("ShaderNodeTexImage")
-    bump = nodes.new("ShaderNodeBump")
-    roughness_curve = nodes.new("ShaderNodeFloatCurve")
+    output = nodes.new("ShaderNodeOutputMaterial")
 
     color_texture.image = color_image
     color_texture.interpolation = "Linear"
     color_texture.extension = "EXTEND"
+    color_texture.location = (-640.0, 180.0)
+    output.location = (460.0, 180.0)
+
+    links = material.node_tree.links
+    if unshaded:
+        emission = nodes.new("ShaderNodeEmission")
+        transparent = nodes.new("ShaderNodeBsdfTransparent")
+        mix_shader = nodes.new("ShaderNodeMixShader")
+        emission.location = (-100.0, 260.0)
+        transparent.location = (-100.0, 20.0)
+        mix_shader.location = (200.0, 180.0)
+        links.new(
+            color_texture.outputs["Color"],
+            emission.inputs["Color"],
+        )
+        links.new(
+            color_texture.outputs["Alpha"],
+            mix_shader.inputs["Fac"],
+        )
+        links.new(
+            transparent.outputs["BSDF"],
+            mix_shader.inputs[1],
+        )
+        links.new(
+            emission.outputs["Emission"],
+            mix_shader.inputs[2],
+        )
+        links.new(
+            mix_shader.outputs["Shader"],
+            output.inputs["Surface"],
+        )
+    else:
+        shader = nodes.new("ShaderNodeBsdfPrincipled")
+        bump = nodes.new("ShaderNodeBump")
+        roughness_curve = nodes.new("ShaderNodeFloatCurve")
         bump.inputs["Distance"].default_value = BUMP_BASE_DISTANCE * depth_scale
         roughness_curve.mapping.initialize()
         curve_points = roughness_curve.mapping.curves[0].points
@@ -446,14 +526,14 @@ def create_material(
         curve_points[-1].location = (1.0, 0.1)
         roughness_curve.mapping.update()
 
-    color_texture.location = (-640.0, 180.0)
         roughness_curve.location = (-300.0, 320.0)
         bump.location = (-300.0, -80.0)
         shader.location = (80.0, 180.0)
-    output.location = (460.0, 180.0)
 
-    links = material.node_tree.links
-    links.new(color_texture.outputs["Color"], shader.inputs["Base Color"])
+        links.new(
+            color_texture.outputs["Color"],
+            shader.inputs["Base Color"],
+        )
         links.new(color_texture.outputs["Alpha"], shader.inputs["Alpha"])
         links.new(color_texture.outputs["Color"], bump.inputs["Height"])
         links.new(bump.outputs["Normal"], shader.inputs["Normal"])
@@ -525,6 +605,8 @@ def build_image_depth_plane(
     subdivision_level,
     delete_source,
     displacement_via="MODIFIER",
+    unshaded=False,
+    color_image=None,
 ):
     source_name = source_object.name
     mesh = create_plane_mesh(
@@ -554,13 +636,14 @@ def build_image_depth_plane(
     depth_scale = abs(float(plane_object.scale.z))
     material = create_material(
         source_name,
-        source_object.data,
+        color_image if color_image is not None else source_object.data,
         depth_scale,
         displacement_image=(
             source_object.data if displacement_via == "MATERIAL" else None
         ),
         displacement_midlevel=ORIGINAL_DISPLACEMENT_MIDLEVEL,
         displacement_scale=ORIGINAL_DISPLACEMENT_STRENGTH,
+        unshaded=unshaded,
     )
     mesh.materials.append(material)
     return finish_depth_plane(
@@ -578,6 +661,9 @@ def build_depth_plane(
     subdivision_level,
     delete_source,
     displacement_via="MODIFIER",
+    unshaded=False,
+    depth_name="",
+    color_image=None,
 ):
     source_name = source_object.name
     depth = extract_depth_array(npz_path)
@@ -589,7 +675,7 @@ def build_depth_plane(
         intrinsics,
         DISPLACEMENT_STRENGTH,
     )
-    depth_name = depth_data_name(source_object)
+    depth_name = depth_name or depth_data_name(source_object)
     depth_image = create_depth_image(
         depth_name,
         depth_values,
@@ -627,13 +713,14 @@ def build_depth_plane(
 
     material = create_material(
         source_name,
-        source_object.data,
+        color_image if color_image is not None else source_object.data,
         depth_scale,
         displacement_image=(
             depth_image if displacement_via == "MATERIAL" else None
         ),
         displacement_midlevel=DISPLACEMENT_MIDLEVEL,
         displacement_scale=DISPLACEMENT_STRENGTH,
+        unshaded=unshaded,
     )
     mesh.materials.append(material)
     plane_object["snapform_reference_depth"] = reference_depth
@@ -660,6 +747,12 @@ class BuildDepthPlane(bpy.types.Operator):
     snapform_npz_path: bpy.props.StringProperty(
         options={"HIDDEN", "SKIP_SAVE"},
     )
+    snapform_depth_data_name: bpy.props.StringProperty(
+        options={"HIDDEN", "SKIP_SAVE"},
+    )
+    snapform_background_path: bpy.props.StringProperty(
+        options={"HIDDEN", "SKIP_SAVE"},
+    )
     snapform_subdivision_level: bpy.props.IntProperty(
         default=5,
         min=0,
@@ -678,6 +771,10 @@ class BuildDepthPlane(bpy.types.Operator):
         default="MODIFIER",
         options={"HIDDEN", "SKIP_SAVE"},
     )
+    snapform_unshaded: bpy.props.BoolProperty(
+        default=False,
+        options={"HIDDEN", "SKIP_SAVE"},
+    )
 
     def execute(self, context):
         source_object = bpy.data.objects.get(self.snapform_source_object_name)
@@ -697,13 +794,23 @@ class BuildDepthPlane(bpy.types.Operator):
             return {"CANCELLED"}
 
         try:
+            color_image = None
+            if self.snapform_background_path:
+                color_image = create_removed_background_image(
+                    source_object,
+                    self.snapform_background_path,
+                    assign_to_source=False,
+                )
             build_depth_plane(
                 context,
                 source_object,
                 npz_path,
                 self.snapform_subdivision_level,
                 self.snapform_delete_source,
-                self.snapform_displacement_via,
+                displacement_via=self.snapform_displacement_via,
+                unshaded=self.snapform_unshaded,
+                depth_name=self.snapform_depth_data_name,
+                color_image=color_image,
             )
         except (OSError, RuntimeError, ValueError) as error:
             self.report({"ERROR"}, str(error))
@@ -711,6 +818,138 @@ class BuildDepthPlane(bpy.types.Operator):
         return {"FINISHED"}
 
 
+class BuildOriginalDepthPlane(bpy.types.Operator):
+    bl_idname = "snapform.build_original_depth_plane"
+    bl_label = "Build Original Depth Plane"
+    bl_options = {"INTERNAL"}
+
+    snapform_source_object_name: bpy.props.StringProperty(
+        options={"HIDDEN", "SKIP_SAVE"},
+    )
+    snapform_background_path: bpy.props.StringProperty(
+        options={"HIDDEN", "SKIP_SAVE"},
+    )
+    snapform_subdivision_level: bpy.props.IntProperty(
+        default=5,
+        min=0,
+        max=10,
+        options={"HIDDEN", "SKIP_SAVE"},
+    )
+    snapform_delete_source: bpy.props.BoolProperty(
+        default=True,
+        options={"HIDDEN", "SKIP_SAVE"},
+    )
+    snapform_displacement_via: bpy.props.EnumProperty(
+        items=(
+            ("MODIFIER", "Modifier", "Use a Displace modifier"),
+            ("MATERIAL", "Material", "Use material displacement"),
+        ),
+        default="MODIFIER",
+        options={"HIDDEN", "SKIP_SAVE"},
+    )
+    snapform_unshaded: bpy.props.BoolProperty(
+        default=False,
+        options={"HIDDEN", "SKIP_SAVE"},
+    )
+
+    def execute(self, context):
+        source_object = bpy.data.objects.get(self.snapform_source_object_name)
+        if not is_image_empty(source_object):
+            self.report(
+                {"ERROR"},
+                "The source image empty is no longer available",
+            )
+            return {"CANCELLED"}
+        try:
+            color_image = create_removed_background_image(
+                source_object,
+                self.snapform_background_path,
+                assign_to_source=False,
+            )
+            build_image_depth_plane(
+                context,
+                source_object,
+                self.snapform_subdivision_level,
+                self.snapform_delete_source,
+                displacement_via=self.snapform_displacement_via,
+                unshaded=self.snapform_unshaded,
+                color_image=color_image,
+            )
+        except (OSError, RuntimeError, ValueError) as error:
+            self.report({"ERROR"}, str(error))
+            return {"CANCELLED"}
+        return {"FINISHED"}
+
+
+class ApplyRemovedBackground(bpy.types.Operator):
+    bl_idname = "snapform.apply_removed_background"
+    bl_label = "Apply Removed Background"
+    bl_options = {"INTERNAL"}
+
+    snapform_source_object_name: bpy.props.StringProperty(
+        options={"HIDDEN", "SKIP_SAVE"},
+    )
+    snapform_output_path: bpy.props.StringProperty(
+        options={"HIDDEN", "SKIP_SAVE"},
+    )
+
+    def execute(self, _context):
+        source_object = bpy.data.objects.get(self.snapform_source_object_name)
+        if not is_image_empty(source_object):
+            self.report(
+                {"ERROR"},
+                "The source image empty is no longer available",
+            )
+            return {"CANCELLED"}
+        try:
+            create_removed_background_image(
+                source_object,
+                self.snapform_output_path,
+            )
+        except RuntimeError as error:
+            self.report({"ERROR"}, str(error))
+            return {"CANCELLED"}
+        return {"FINISHED"}
+
+
+class RemoveImageBackground(bpy.types.Operator):
+    bl_idname = "snapform.remove_image_background"
+    bl_label = "Remove Background"
+    bl_description = "Remove the background from this Image Empty with BEN2"
+    bl_options = {"REGISTER"}
+
+    @classmethod
+    def poll(cls, context):
+        from .operators import active_job
+
+        return is_image_empty(context.object) and active_job() is None
+
+    def execute(self, context):
+        from .operators import IMAGE_BACKGROUND_JOB
+
+        source_object = context.object
+        try:
+            source_path, delete_input = prepare_image_input(
+                source_object.data,
+                context.scene,
+            )
+        except RuntimeError as error:
+            self.report({"ERROR"}, str(error))
+            return {"CANCELLED"}
+
+        result = bpy.ops.snapform.run_job(
+            job_type=IMAGE_BACKGROUND_JOB,
+            snapform_input_path=str(source_path),
+            snapform_source_object_name=source_object.name,
+            snapform_delete_input=delete_input,
+        )
+        if "CANCELLED" not in result:
+            return {"FINISHED"}
+        if delete_input:
+            source_path.unlink(missing_ok=True)
+        return {"CANCELLED"}
+
+
 class ConvertImageToDepthPlane(bpy.types.Operator):
     bl_idname = "snapform.convert_image_to_depth_plane"
     bl_label = "Convert to Depth Plane"
@@ -739,6 +978,16 @@ class ConvertImageToDepthPlane(bpy.types.Operator):
         ),
         default="MODIFIER",
     )
+    snapform_unshaded: bpy.props.BoolProperty(
+        name="Unshaded",
+        description="Display the source color through an Emission shader",
+        default=False,
+    )
+    snapform_remove_background: bpy.props.BoolProperty(
+        name="Remove Background",
+        description="Remove the color image background with BEN2",
+        default=False,
+    )
     snapform_model: bpy.props.EnumProperty(
         name="Model",
         description="Model used to estimate image depth",
@@ -800,15 +1049,20 @@ class ConvertImageToDepthPlane(bpy.types.Operator):
                 "snapform_process_res_factor",
                 text=f"Resolution Scale ({process_res} px)",
             )
+        layout.prop(self, "snapform_remove_background")
         layout.prop(self, "snapform_displacement_via")
+        layout.prop(self, "snapform_unshaded")
         layout.prop(self, "snapform_subdivision_level")
         layout.prop(self, "snapform_delete_original")
 
     def execute(self, context):
-        from .operators import IMAGE_DEPTH_JOB
+        from .operators import IMAGE_DEPTH_JOB, ORIGINAL_BACKGROUND_JOB
 
         source_object = context.object
-        if self.snapform_depth_source == "ORIGINAL":
+        if (
+            self.snapform_depth_source == "ORIGINAL"
+            and not self.snapform_remove_background
+        ):
             try:
                 build_image_depth_plane(
                     context,
@@ -816,6 +1070,7 @@ class ConvertImageToDepthPlane(bpy.types.Operator):
                     self.snapform_subdivision_level,
                     self.snapform_delete_original,
                     self.snapform_displacement_via,
+                    self.snapform_unshaded,
                 )
             except (OSError, RuntimeError, ValueError) as error:
                 self.report({"ERROR"}, str(error))
@@ -831,6 +1086,31 @@ class ConvertImageToDepthPlane(bpy.types.Operator):
                 return {"CANCELLED"}
             return {"FINISHED"}
 
+        if self.snapform_depth_source == "ORIGINAL":
+            try:
+                source_path, delete_input = prepare_image_input(
+                    source_object.data,
+                    context.scene,
+                )
+            except RuntimeError as error:
+                self.report({"ERROR"}, str(error))
+                return {"CANCELLED"}
+            result = bpy.ops.snapform.run_job(
+                job_type=ORIGINAL_BACKGROUND_JOB,
+                snapform_input_path=str(source_path),
+                snapform_source_object_name=source_object.name,
+                snapform_subdivision_level=self.snapform_subdivision_level,
+                snapform_delete_source=self.snapform_delete_original,
+                snapform_delete_input=delete_input,
+                snapform_displacement_via=self.snapform_displacement_via,
+                snapform_unshaded=self.snapform_unshaded,
+            )
+            if "CANCELLED" not in result:
+                return {"FINISHED"}
+            if delete_input:
+                source_path.unlink(missing_ok=True)
+            return {"CANCELLED"}
+
         image = source_object.data
         try:
             process_res = scaled_process_resolution(
@@ -840,26 +1120,20 @@ class ConvertImageToDepthPlane(bpy.types.Operator):
         except ValueError as error:
             self.report({"ERROR"}, str(error))
             return {"CANCELLED"}
-        source_path = image_path(image)
-        delete_input = False
-        if image_is_packed(image):
         try:
-                source_path = save_packed_image(image, context.scene)
+            source_path, delete_input = prepare_image_input(
+                image,
+                context.scene,
+            )
         except RuntimeError as error:
             self.report({"ERROR"}, str(error))
             return {"CANCELLED"}
-            delete_input = True
-        if not source_path.is_file():
-            self.report(
-                {"ERROR"},
-                "The image empty must reference an existing image file",
-            )
-            return {"CANCELLED"}
 
         result = bpy.ops.snapform.run_job(
             job_type=IMAGE_DEPTH_JOB,
             snapform_input_path=str(source_path),
             snapform_source_object_name=source_object.name,
+            snapform_depth_data_name=depth_data_name(source_object),
             snapform_model_key=self.snapform_model,
             snapform_use_ray_pose=self.snapform_use_ray_pose,
             snapform_process_res=process_res,
@@ -867,6 +1141,8 @@ class ConvertImageToDepthPlane(bpy.types.Operator):
             snapform_delete_source=self.snapform_delete_original,
             snapform_delete_input=delete_input,
             snapform_displacement_via=self.snapform_displacement_via,
+            snapform_unshaded=self.snapform_unshaded,
+            snapform_remove_background=self.snapform_remove_background,
         )
         if "CANCELLED" in result:
             if delete_input:
@@ -889,3 +1165,10 @@ def draw_image_context_menu(self, context):
         text="Convert to Depth Plane",
         icon="MOD_DISPLACE",
     )
+    row = self.layout.row()
+    row.operator_context = "INVOKE_DEFAULT"
+    row.operator(
+        "snapform.remove_image_background",
+        text="Remove Background",
+        icon="IMAGE_ALPHA",
+    )

src/snapform/models.py

@@ -9,6 +9,7 @@ class ModelSpec:
     license_name: str
     description: str
     supports_gs: bool
+    download_patterns: tuple[str, ...] = ()
 
 
 MODELS = {
@@ -53,6 +54,17 @@ MODELS = {
         supports_gs=True,
     ),
 }
+
+BEN2_MODEL = ModelSpec(
+    key="BEN2_BASE",
+    label="BEN2 Base",
+    repository="PramaLLC/BEN2",
+    license_name="MIT",
+    description="Fast high-quality background removal model",
+    supports_gs=False,
+    download_patterns=("config.json", "model.safetensors"),
+)
+DOWNLOADABLE_MODELS = {**MODELS, BEN2_MODEL.key: BEN2_MODEL}
 MODEL_NUMBERS = {
     model_key: number for number, model_key in enumerate(MODELS)
 }
@@ -123,3 +135,10 @@ def get_model(model_key):
         return MODELS[model_key]
     except KeyError as error:
         raise ValueError(f"Unknown DA3 model: {model_key}") from error
+
+
+def get_downloadable_model(model_key):
+    try:
+        return DOWNLOADABLE_MODELS[model_key]
+    except KeyError as error:
+        raise ValueError(f"Unknown model: {model_key}") from error

src/snapform/operators.py

@@ -5,11 +5,19 @@ from pathlib import Path
 import bpy
 
 from . import runtime
-from .models import get_model, selected_model_key
+from .models import (
+    BEN2_MODEL,
+    get_downloadable_model,
+    get_model,
+    selected_model_key,
+)
 
 
 RECONSTRUCTION_JOB = "RECONSTRUCT"
 IMAGE_DEPTH_JOB = "IMAGE_DEPTH"
+BACKGROUND_REMOVAL_JOB = "REMOVE_BACKGROUND"
+IMAGE_BACKGROUND_JOB = "IMAGE_BACKGROUND"
+ORIGINAL_BACKGROUND_JOB = "ORIGINAL_BACKGROUND"
 IMAGE_DEPTH_DEFAULT_PROCESS_RES = 1024
 IMAGE_DEPTH_PROCESS_RES_METHOD = "upper_bound_resize"
 
@@ -33,10 +41,13 @@ class ActiveJob:
     result_directory: Path
     export_format: str = ""
     source_object_name: str = ""
+    depth_data_name: str = ""
     process_res: int = IMAGE_DEPTH_DEFAULT_PROCESS_RES
     subdivision_level: int = 5
     delete_source: bool = True
     displacement_via: str = "MODIFIER"
+    unshaded: bool = False
+    remove_background: bool = False
     temporary_input_path: Path | None = None
     cancelled: bool = False
 
@@ -83,6 +94,9 @@ class RunJob(bpy.types.Operator):
             ("DOWNLOAD_MODEL", "Download Model", ""),
             (RECONSTRUCTION_JOB, "Export DA3 Result", ""),
             (IMAGE_DEPTH_JOB, "Convert Image to Depth Plane", ""),
+            (BACKGROUND_REMOVAL_JOB, "Test BEN2 Background Removal", ""),
+            (IMAGE_BACKGROUND_JOB, "Remove Image Empty Background", ""),
+            (ORIGINAL_BACKGROUND_JOB, "Build Original Background Plane", ""),
         ),
         options={"HIDDEN", "SKIP_SAVE"},
     )
@@ -95,6 +109,9 @@ class RunJob(bpy.types.Operator):
     snapform_source_object_name: bpy.props.StringProperty(
         options={"HIDDEN", "SKIP_SAVE"},
     )
+    snapform_depth_data_name: bpy.props.StringProperty(
+        options={"HIDDEN", "SKIP_SAVE"},
+    )
     snapform_model_key: bpy.props.StringProperty(
         default="BASE",
         options={"HIDDEN", "SKIP_SAVE"},
@@ -131,6 +148,14 @@ class RunJob(bpy.types.Operator):
         default="MODIFIER",
         options={"HIDDEN", "SKIP_SAVE"},
     )
+    snapform_unshaded: bpy.props.BoolProperty(
+        default=False,
+        options={"HIDDEN", "SKIP_SAVE"},
+    )
+    snapform_remove_background: bpy.props.BoolProperty(
+        default=False,
+        options={"HIDDEN", "SKIP_SAVE"},
+    )
 
     def execute(self, context):
         global _active_job
@@ -173,13 +198,20 @@ class RunJob(bpy.types.Operator):
                 else ""
             ),
             source_object_name=self.snapform_source_object_name,
+            depth_data_name=self.snapform_depth_data_name,
             process_res=self.snapform_process_res,
             subdivision_level=self.snapform_subdivision_level,
             delete_source=self.snapform_delete_source,
             displacement_via=self.snapform_displacement_via,
+            unshaded=self.snapform_unshaded,
+            remove_background=self.snapform_remove_background,
             temporary_input_path=(
                 Path(self.snapform_input_path)
-                if self.job_type == IMAGE_DEPTH_JOB
+                if self.job_type in {
+                    IMAGE_DEPTH_JOB,
+                    IMAGE_BACKGROUND_JOB,
+                    ORIGINAL_BACKGROUND_JOB,
+                }
                 and self.snapform_delete_input
                 else None
             ),
@@ -196,6 +228,9 @@ class RunJob(bpy.types.Operator):
 
     def _prepare_command(self, _context, settings):
         paths = runtime.new_job_paths(self.job_type)
+        remove_background = bool(
+            getattr(self, "snapform_remove_background", False)
+        )
         if self.job_type == "INSTALL_RUNTIME":
             self._require_online_access()
             return runtime.install_runtime_command(paths["status"]), paths
@@ -205,20 +240,49 @@ class RunJob(bpy.types.Operator):
 
         if self.job_type == "DOWNLOAD_MODEL":
             self._require_online_access()
-            model = get_model(self.model_key)
+            model = get_downloadable_model(self.model_key)
             return (
                 runtime.download_model_command(model.key, paths["status"]),
                 paths,
             )
 
-        configured_input = (
-            self.snapform_input_path
-            if self.job_type == IMAGE_DEPTH_JOB
-            else settings.input_path
-        )
+        if self.job_type in {
+            IMAGE_DEPTH_JOB,
+            IMAGE_BACKGROUND_JOB,
+            ORIGINAL_BACKGROUND_JOB,
+        }:
+            configured_input = self.snapform_input_path
+        elif self.job_type == BACKGROUND_REMOVAL_JOB:
+            configured_input = settings.background_input_path
+        else:
+            configured_input = settings.input_path
         input_path = Path(bpy.path.abspath(configured_input))
         if not input_path.is_file():
             raise ValueError("Select a valid image or video file")
+        if self.job_type in {
+            BACKGROUND_REMOVAL_JOB,
+            IMAGE_BACKGROUND_JOB,
+            ORIGINAL_BACKGROUND_JOB,
+        }:
+            if input_path.suffix.lower() not in {
+                ".jpg",
+                ".jpeg",
+                ".png",
+                ".webp",
+            }:
+                raise ValueError("BEN2 requires a JPG, JPEG, PNG, or WEBP image")
+            if not runtime.model_ready(BEN2_MODEL.key):
+                raise RuntimeError(f"Download {BEN2_MODEL.label} first")
+            return (
+                runtime.remove_background_command(
+                    settings,
+                    input_path,
+                    paths["status"],
+                    paths["result"],
+                    BEN2_MODEL.key,
+                ),
+                paths,
+            )
         if self.job_type not in {RECONSTRUCTION_JOB, IMAGE_DEPTH_JOB}:
             raise RuntimeError(f"Unsupported task type: {self.job_type}")
         export_format = (
@@ -236,6 +300,12 @@ class RunJob(bpy.types.Operator):
         model = get_model(model_key)
         if not runtime.model_ready(model_key):
             raise RuntimeError(f"Download {model.label} first")
+        if (
+            self.job_type == IMAGE_DEPTH_JOB
+            and remove_background
+            and not runtime.model_ready(BEN2_MODEL.key)
+        ):
+            raise RuntimeError(f"Download {BEN2_MODEL.label} first")
         if export_format == "gs_ply" and not model.supports_gs:
             raise RuntimeError(
                 "PLY export requires a DA3 Giant or Nested Giant model"
@@ -255,6 +325,10 @@ class RunJob(bpy.types.Operator):
                     else None
                 ),
                 use_ray_pose=use_ray_pose,
+                remove_background=(
+                    self.job_type == IMAGE_DEPTH_JOB
+                    and remove_background
+                ),
             ),
             paths,
         )
@@ -274,8 +348,15 @@ class RunJob(bpy.types.Operator):
             return f"Starting {settings.export_format} export"
         if self.job_type == IMAGE_DEPTH_JOB:
             return "Starting depth estimation"
+        if self.job_type == BACKGROUND_REMOVAL_JOB:
+            return "Starting BEN2 background removal"
+        if self.job_type == IMAGE_BACKGROUND_JOB:
+            return "Starting Image Empty background removal"
+        if self.job_type == ORIGINAL_BACKGROUND_JOB:
+            return "Starting background removal"
         if self.job_type == "DOWNLOAD_MODEL":
-            return f"Starting {get_model(self.model_key).label} download"
+            model = get_downloadable_model(self.model_key)
+            return f"Starting {model.label} download"
         return labels[self.job_type]
 
     def modal(self, context, event):
@@ -330,15 +411,18 @@ class RunJob(bpy.types.Operator):
             self._redraw_ui(context, force=True)
             return {"CANCELLED"}
 
-        if job.kind == RECONSTRUCTION_JOB:
+        if job.kind in {RECONSTRUCTION_JOB, BACKGROUND_REMOVAL_JOB}:
             try:
                 open_job_directory(job.result_directory)
             except Exception as open_error:
                 self.report({"WARNING"}, str(open_error))
-            self.report(
-                {"INFO"},
-                f"Exported {job.export_format} to {job.result_directory}",
+            if job.kind == RECONSTRUCTION_JOB:
+                message = (
+                    f"Exported {job.export_format} to {job.result_directory}"
                 )
+            else:
+                message = f"Removed background to {job.result_directory}"
+            self.report({"INFO"}, message)
         if job.kind == IMAGE_DEPTH_JOB:
             try:
                 settings.status = "Building depth plane"
@@ -353,6 +437,34 @@ class RunJob(bpy.types.Operator):
                 self._redraw_ui(context, force=True)
                 return {"CANCELLED"}
             status["message"] = "Depth plane is ready"
+        if job.kind == ORIGINAL_BACKGROUND_JOB:
+            try:
+                settings.status = "Building depth plane"
+                settings.progress = 0.96
+                self._redraw_ui(context, force=True)
+                self._build_original_depth_plane(job)
+            except (OSError, RuntimeError, ValueError) as import_error:
+                settings.status = "Depth import failed"
+                settings.progress = 0.0
+                self.report({"ERROR"}, str(import_error))
+                close_active_job()
+                self._redraw_ui(context, force=True)
+                return {"CANCELLED"}
+            status["message"] = "Depth plane is ready"
+        if job.kind == IMAGE_BACKGROUND_JOB:
+            try:
+                settings.status = "Updating Image Empty"
+                settings.progress = 0.96
+                self._redraw_ui(context, force=True)
+                self._apply_removed_background(job)
+            except (OSError, RuntimeError, ValueError) as import_error:
+                settings.status = "Image update failed"
+                settings.progress = 0.0
+                self.report({"ERROR"}, str(import_error))
+                close_active_job()
+                self._redraw_ui(context, force=True)
+                return {"CANCELLED"}
+            status["message"] = "Image background is removed"
 
         settings.status = status.get("message", "Task complete")
         settings.progress = 1.0
@@ -370,9 +482,16 @@ class RunJob(bpy.types.Operator):
         result = bpy.ops.snapform.build_depth_plane(
             snapform_source_object_name=job.source_object_name,
             snapform_npz_path=str(npz_path),
+            snapform_depth_data_name=job.depth_data_name,
+            snapform_background_path=(
+                str(job.result_directory / "foreground.png")
+                if job.remove_background
+                else ""
+            ),
             snapform_subdivision_level=job.subdivision_level,
             snapform_delete_source=job.delete_source,
             snapform_displacement_via=job.displacement_via,
+            snapform_unshaded=job.unshaded,
         )
         if "FINISHED" not in result:
             raise RuntimeError("Unable to build the depth plane")
@@ -380,6 +499,37 @@ class RunJob(bpy.types.Operator):
         if "FINISHED" not in undo_result:
             raise RuntimeError("Unable to create the depth plane Undo step")
 
+    def _build_original_depth_plane(self, job):
+        result = bpy.ops.snapform.build_original_depth_plane(
+            snapform_source_object_name=job.source_object_name,
+            snapform_background_path=str(
+                job.result_directory / "foreground.png"
+            ),
+            snapform_subdivision_level=job.subdivision_level,
+            snapform_delete_source=job.delete_source,
+            snapform_displacement_via=job.displacement_via,
+            snapform_unshaded=job.unshaded,
+        )
+        if "FINISHED" not in result:
+            raise RuntimeError("Unable to build the depth plane")
+        undo_result = bpy.ops.ed.undo_push(message="Convert to Depth Plane")
+        if "FINISHED" not in undo_result:
+            raise RuntimeError("Unable to create the depth plane Undo step")
+
+    def _apply_removed_background(self, job, push_undo=True):
+        output_path = job.result_directory / "foreground.png"
+        result = bpy.ops.snapform.apply_removed_background(
+            snapform_source_object_name=job.source_object_name,
+            snapform_output_path=str(output_path),
+        )
+        if "FINISHED" not in result:
+            raise RuntimeError("Unable to update the Image Empty")
+        if not push_undo:
+            return
+        undo_result = bpy.ops.ed.undo_push(message="Remove Background")
+        if "FINISHED" not in undo_result:
+            raise RuntimeError("Unable to create the background removal Undo step")
+
     def _remove_timer(self, context):
         timer = getattr(self, "_timer", None)
         if timer is None:

src/snapform/panel.py

@@ -1,6 +1,6 @@
 import bpy
 
-from .models import MODELS
+from .models import BEN2_MODEL, MODELS
 from .operators import active_job
 from .runtime import model_ready, runtime_ready
 
@@ -29,6 +29,7 @@ class RuntimeModelsPanel(bpy.types.Panel):
 
         for model in MODELS.values():
             self._draw_model(layout, model, running)
+        self._draw_model(layout, BEN2_MODEL, running)
 
     def _draw_model(self, layout, model, running):
         row = layout.row(align=True)
@@ -106,3 +107,15 @@ class DebugPanel(bpy.types.Panel):
             icon="EXPORT",
         )
         operator.job_type = "RECONSTRUCT"
+
+        background_box = layout.box()
+        background_box.label(text="BEN2 Background Removal")
+        background_box.prop(settings, "background_input_path")
+        background_row = background_box.row()
+        background_row.enabled = active_job() is None
+        operator = background_row.operator(
+            "snapform.run_job",
+            text="Remove Background",
+            icon="IMAGE_ALPHA",
+        )
+        operator.job_type = "REMOVE_BACKGROUND"

src/snapform/properties.py

@@ -9,6 +9,11 @@ class SnapFormSettings(bpy.types.PropertyGroup):
         description="A single image or video file used for reconstruction",
         subtype="FILE_PATH",
     )
+    background_input_path: bpy.props.StringProperty(
+        name="Image",
+        description="A single image used to test BEN2 background removal",
+        subtype="FILE_PATH",
+    )
     video_frames: bpy.props.IntProperty(
         name="Video Frames",
         description="Maximum number of frames sampled evenly from a video",

src/snapform/runtime.py

@@ -8,11 +8,12 @@ from pathlib import Path
 
 import bpy
 
-from .models import get_model, selected_model_key
+from .models import BEN2_MODEL, get_downloadable_model, selected_model_key
 
 
 DA3_COMMIT = "3d835ec1a5802d64a8b8b15f817a1ab54809bfe4"
-RUNTIME_SCHEMA = 2
+BEN2_COMMIT = "2c99a5da477b5523585bfa5c893888a6e818a8f6"
+RUNTIME_SCHEMA = 3
 DEFAULT_STORAGE_ROOT = Path.home() / ".snapform"
 
 
@@ -127,14 +128,16 @@ def install_runtime_command(status_path):
         backend.lower(),
         "--da3-commit",
         DA3_COMMIT,
+        "--ben2-commit",
+        BEN2_COMMIT,
         "--status",
         str(status_path),
     ]
 
 
 def download_model_command(model_key, status_path):
-    model = get_model(model_key)
-    return [
+    model = get_downloadable_model(model_key)
+    command = [
         str(runtime_python()),
         str(Path(__file__).with_name("worker.py")),
         "download",
@@ -145,6 +148,33 @@ def download_model_command(model_key, status_path):
         "--status",
         str(status_path),
     ]
+    for pattern in model.download_patterns:
+        command.extend(["--allow-pattern", pattern])
+    return command
+
+
+def remove_background_command(
+    settings,
+    input_path,
+    status_path,
+    result_directory,
+    model_key,
+):
+    return [
+        str(runtime_python()),
+        str(Path(__file__).with_name("worker.py")),
+        "remove-background",
+        "--input",
+        str(input_path),
+        "--model-dir",
+        str(model_directory(model_key)),
+        "--output-dir",
+        str(result_directory),
+        "--status",
+        str(status_path),
+        "--device",
+        settings.device.lower(),
+    ]
 
 
 def reconstruct_command(
@@ -157,6 +187,7 @@ def reconstruct_command(
     process_res=None,
     process_res_method=None,
     use_ray_pose=None,
+    remove_background=False,
 ):
     snapform_selected_key = model_key or selected_model_key(settings, export_format)
     snapform_selected_process_res = (
@@ -170,7 +201,7 @@ def reconstruct_command(
     snapform_selected_ray_pose = (
         settings.use_ray_pose if use_ray_pose is None else use_ray_pose
     )
-    return [
+    command = [
         str(runtime_python()),
         str(Path(__file__).with_name("worker.py")),
         "reconstruct",
@@ -209,6 +240,14 @@ def reconstruct_command(
         "--ref-view-strategy",
         settings.ref_view_strategy,
     ]
+    if remove_background:
+        command.extend(
+            [
+                "--remove-background-model-dir",
+                str(model_directory(BEN2_MODEL.key)),
+            ]
+        )
+    return command
 
 
 def creation_flags():

src/snapform/worker.py

@@ -188,6 +188,7 @@ def download_model(args):
         repo_id=args.repository,
         local_dir=model_dir,
         tqdm_class=progress_class,
+        allow_patterns=args.allow_pattern or None,
     )
     write_status(args.status, 1.0, "Model is ready")
 
@@ -202,6 +203,76 @@ def select_device(requested, torch):
     return torch.device("cuda" if torch.cuda.is_available() else "cpu")
 
 
+def infer_foreground(input_path, model_dir, device, torch):
+    from ben2 import BEN_Base
+    from PIL import Image
+
+    input_path = Path(input_path)
+    if input_path.suffix.lower() not in IMAGE_SUFFIXES:
+        raise RuntimeError("BEN2 requires a JPG, JPEG, PNG, or WEBP image")
+    load_started = time.perf_counter()
+    model = BEN_Base.from_pretrained(model_dir)
+    model = model.to(device=device)
+    model.eval()
+    load_ms = (time.perf_counter() - load_started) * 1000.0
+
+    with Image.open(input_path) as opened_image:
+        image = opened_image.convert("RGB")
+    if device.type == "cuda":
+        torch.cuda.synchronize()
+    inference_started = time.perf_counter()
+    with torch.inference_mode():
+        foreground = model.inference(
+            image,
+            refine_foreground=False,
+        )
+    if device.type == "cuda":
+        torch.cuda.synchronize()
+    inference_ms = (time.perf_counter() - inference_started) * 1000.0
+    del model
+    if device.type == "cuda":
+        torch.cuda.empty_cache()
+    return foreground, load_ms, inference_ms
+
+
+def remove_background(args):
+    total_started = time.perf_counter()
+    write_status(args.status, 0.02, "Starting BEN2 Runtime")
+
+    import torch
+
+    output_directory = Path(args.output_dir)
+    output_directory.mkdir(parents=True, exist_ok=False)
+    device = select_device(args.device, torch)
+
+    write_status(args.status, 0.12, f"Loading BEN2 on {device.type}")
+    write_status(args.status, 0.35, "Removing background")
+    foreground, load_ms, inference_ms = infer_foreground(
+        args.input,
+        args.model_dir,
+        device,
+        torch,
+    )
+
+    write_status(args.status, 0.92, "Writing transparent PNG")
+    output_path = output_directory / "foreground.png"
+    foreground.convert("RGBA").save(
+        output_path,
+        format="PNG",
+        compress_level=1,
+    )
+    if not output_path.is_file():
+        raise RuntimeError("BEN2 did not write a foreground image")
+
+    total_ms = (time.perf_counter() - total_started) * 1000.0
+    message = (
+        f"Background removed in {inference_ms:.0f} ms "
+        f"(model load {load_ms:.0f} ms, total {total_ms:.0f} ms)"
+    )
+    print(message)
+    write_status(args.status, 1.0, message)
+
+
 def exported_paths(output_directory, export_format):
     output_directory = Path(output_directory)
     if export_format == "glb":
@@ -246,6 +317,30 @@ def reconstruct(args):
 
     try:
         device = select_device(args.device, torch)
+        remove_background_model_dir = getattr(
+            args,
+            "remove_background_model_dir",
+            "",
+        )
+        if remove_background_model_dir:
+            write_status(args.status, 0.12, "Removing image background")
+            foreground, load_ms, inference_ms = infer_foreground(
+                images[0],
+                remove_background_model_dir,
+                device,
+                torch,
+            )
+            foreground.convert("RGBA").save(
+                output_directory / "foreground.png",
+                format="PNG",
+                compress_level=1,
+            )
+            if not (output_directory / "foreground.png").is_file():
+                raise RuntimeError("BEN2 did not write a foreground image")
+            print(
+                f"BEN2 inference {inference_ms:.0f} ms; "
+                f"model load {load_ms:.0f} ms"
+            )
         write_status(args.status, 0.15, f"Loading model on {device.type}")
         model = DepthAnything3.from_pretrained(args.model_dir)
         model = model.to(device=device)
@@ -311,6 +406,18 @@ def parser():
     download.add_argument("--repository", required=True)
     download.add_argument("--model-dir", required=True)
     download.add_argument("--status", required=True)
+    download.add_argument("--allow-pattern", action="append", default=[])
+
+    background = commands.add_parser("remove-background")
+    background.add_argument("--input", required=True)
+    background.add_argument("--model-dir", required=True)
+    background.add_argument("--output-dir", required=True)
+    background.add_argument("--status", required=True)
+    background.add_argument(
+        "--device",
+        choices=("auto", "cuda", "cpu"),
+        required=True,
+    )
 
     reconstruction = commands.add_parser("reconstruct")
     reconstruction.add_argument("--input", required=True)
@@ -322,6 +429,7 @@ def parser():
         required=True,
     )
     reconstruction.add_argument("--status", required=True)
+    reconstruction.add_argument("--remove-background-model-dir", default="")
     reconstruction.add_argument("--process-res", type=int, required=True)
     reconstruction.add_argument(
         "--process-res-method",
@@ -370,8 +478,10 @@ def main():
     try:
         if args.command == "download":
             download_model(args)
-        else:
+        elif args.command == "reconstruct":
             reconstruct(args)
+        else:
+            remove_background(args)
     except Exception as error:
         write_status(args.status, 1.0, "Task failed", str(error))
         raise

tests/test_addon.py

@@ -118,12 +118,26 @@ class AddonTest(unittest.TestCase):
         )
         self.assertFalse(hasattr(self.addon, "ImportDepthPlane"))
 
+    def test_registers_image_background_operators(self):
+        self.assertIn(self.addon.RemoveImageBackground, self.addon.CLASSES)
+        self.assertIn(self.addon.ApplyRemovedBackground, self.addon.CLASSES)
+        self.assertIn(self.addon.BuildOriginalDepthPlane, self.addon.CLASSES)
+        self.assertEqual(
+            self.addon.ApplyRemovedBackground.bl_options,
+            {"INTERNAL"},
+        )
+
     def test_completed_depth_job_uses_undoable_build_operator(self):
         source = inspect.getsource(
             self.addon.RunJob._build_depth_plane
         )
 
         self.assertIn("bpy.ops.snapform.build_depth_plane(", source)
+        self.assertIn("snapform_unshaded=job.unshaded", source)
+        self.assertIn(
+            "snapform_depth_data_name=job.depth_data_name",
+            source,
+        )
         self.assertIn('if "FINISHED" not in result:', source)
         self.assertIn(
             'bpy.ops.ed.undo_push(message="Convert to Depth Plane")',
@@ -131,7 +145,7 @@ class AddonTest(unittest.TestCase):
         )
         self.assertIn('if "FINISHED" not in undo_result:', source)
 
-    def test_image_context_menu_only_exposes_conversion(self):
+    def test_image_context_menu_exposes_depth_and_background_actions(self):
         draw_source = inspect.getsource(
             self.addon.draw_image_context_menu
         )
@@ -140,10 +154,185 @@ class AddonTest(unittest.TestCase):
             '"snapform.convert_image_to_depth_plane"',
             draw_source,
         )
+        self.assertIn(
+            '"snapform.remove_image_background"',
+            draw_source,
+        )
+        self.assertIn('text="Remove Background"', draw_source)
+        self.assertEqual(draw_source.count("row = self.layout.row()"), 2)
         self.assertIn('operator_context = "INVOKE_DEFAULT"', draw_source)
         self.assertNotIn("import_depth_plane", draw_source)
         self.assertNotIn("Rebuild", draw_source)
 
+    def test_background_result_creates_new_packed_image(self):
+        original_image = SimpleNamespace(
+            name="img.png",
+            use_fake_user=False,
+        )
+        transform = object()
+        source_object = SimpleNamespace(
+            data=original_image,
+            matrix_world=transform,
+            empty_display_size=2.5,
+            empty_image_offset=(-0.25, -0.75),
+        )
+        pack_calls = []
+        result_image = SimpleNamespace(
+            name="foreground.png",
+            pack=lambda: pack_calls.append(True),
+        )
+        load_calls = []
+        remove_calls = []
+        self.fake_bpy.data = SimpleNamespace(
+            images=SimpleNamespace(
+                load=lambda *args, **options: load_calls.append(
+                    (args, options)
+                )
+                or result_image,
+                remove=remove_calls.append,
+            ),
+        )
+
+        with TemporaryDirectory() as directory:
+            output_path = Path(directory) / "foreground.png"
+            output_path.touch()
+            created = self.addon.image_depth.create_removed_background_image(
+                source_object,
+                output_path,
+            )
+
+        self.assertIs(created, result_image)
+        self.assertIs(source_object.data, result_image)
+        self.assertIsNot(source_object.data, original_image)
+        self.assertTrue(original_image.use_fake_user)
+        self.assertIs(source_object.matrix_world, transform)
+        self.assertEqual(source_object.empty_display_size, 2.5)
+        self.assertEqual(
+            source_object.empty_image_offset,
+            (-0.25, -0.75),
+        )
+        self.assertEqual(result_image.name, "img.rb")
+        self.assertEqual(pack_calls, [True])
+        self.assertEqual(remove_calls, [])
+        self.assertEqual(
+            load_calls[0][1],
+            {"check_existing": False},
+        )
+
+    def test_background_result_name_handles_numbered_format_suffix(self):
+        source_object = SimpleNamespace(
+            data=SimpleNamespace(name="img.png.001"),
+        )
+
+        self.assertEqual(
+            self.addon.image_depth.removed_background_data_name(
+                source_object
+            ),
+            "img.001.rb",
+        )
+
+    def test_depth_background_image_does_not_replace_source_empty(self):
+        original_image = SimpleNamespace(
+            name="img.png",
+            use_fake_user=False,
+        )
+        source_object = SimpleNamespace(data=original_image)
+        pack_calls = []
+        result_image = SimpleNamespace(
+            name="foreground.png",
+            pack=lambda: pack_calls.append(True),
+        )
+        self.fake_bpy.data = SimpleNamespace(
+            images=SimpleNamespace(
+                load=lambda *_args, **_options: result_image,
+                remove=lambda _image: None,
+            ),
+        )
+
+        with TemporaryDirectory() as directory:
+            output_path = Path(directory) / "foreground.png"
+            output_path.touch()
+            created = self.addon.image_depth.create_removed_background_image(
+                source_object,
+                output_path,
+                assign_to_source=False,
+            )
+
+        self.assertIs(created, result_image)
+        self.assertIs(source_object.data, original_image)
+        self.assertFalse(original_image.use_fake_user)
+        self.assertEqual(result_image.name, "img.rb")
+        self.assertEqual(pack_calls, [True])
+
+    def test_image_empty_background_action_starts_production_job(self):
+        source_object = SimpleNamespace(
+            name="Reference",
+            data=SimpleNamespace(name="img.png"),
+        )
+        context = SimpleNamespace(object=source_object, scene=object())
+        calls = []
+        self.fake_bpy.ops = SimpleNamespace(
+            snapform=SimpleNamespace(
+                run_job=lambda **options: calls.append(options)
+                or {"RUNNING_MODAL"},
+            ),
+        )
+        operator = self.addon.RemoveImageBackground()
+
+        with TemporaryDirectory() as directory:
+            input_path = Path(directory) / "input.png"
+            input_path.touch()
+            with patch.object(
+                self.addon.image_depth,
+                "prepare_image_input",
+                return_value=(input_path, True),
+            ):
+                result = operator.execute(context)
+
+        self.assertEqual(result, {"FINISHED"})
+        self.assertEqual(calls[0]["job_type"], "IMAGE_BACKGROUND")
+        self.assertEqual(calls[0]["snapform_input_path"], str(input_path))
+        self.assertEqual(
+            calls[0]["snapform_source_object_name"],
+            "Reference",
+        )
+        self.assertTrue(calls[0]["snapform_delete_input"])
+
+    def test_original_depth_background_starts_independent_job(self):
+        operator = self.addon.ConvertImageToDepthPlane()
+        operator.snapform_depth_source = "ORIGINAL"
+        operator.snapform_remove_background = True
+        operator.snapform_displacement_via = "MODIFIER"
+        operator.snapform_unshaded = False
+        operator.snapform_subdivision_level = 5
+        operator.snapform_delete_original = False
+        source_object = SimpleNamespace(
+            name="Reference",
+            data=SimpleNamespace(name="img.png"),
+        )
+        context = SimpleNamespace(object=source_object, scene=object())
+        calls = []
+        self.fake_bpy.ops = SimpleNamespace(
+            snapform=SimpleNamespace(
+                run_job=lambda **options: calls.append(options)
+                or {"RUNNING_MODAL"},
+            ),
+        )
+
+        with TemporaryDirectory() as directory:
+            input_path = Path(directory) / "input.png"
+            input_path.touch()
+            with patch.object(
+                self.addon.image_depth,
+                "prepare_image_input",
+                return_value=(input_path, False),
+            ):
+                result = operator.execute(context)
+
+        self.assertEqual(result, {"FINISHED"})
+        self.assertEqual(calls[0]["job_type"], "ORIGINAL_BACKGROUND")
+        self.assertFalse(calls[0]["snapform_delete_source"])
+
     def test_conversion_dialog_exposes_requested_settings(self):
         operator = self.addon.ConvertImageToDepthPlane
 
@@ -159,6 +348,12 @@ class AddonTest(unittest.TestCase):
         self.assertTrue(
             operator.__annotations__["snapform_delete_original"]["default"]
         )
+        self.assertFalse(
+            operator.__annotations__["snapform_unshaded"]["default"]
+        )
+        self.assertFalse(
+            operator.__annotations__["snapform_remove_background"]["default"]
+        )
         self.assertEqual(
             operator.__annotations__["snapform_subdivision_level"]["default"],
             5,
@@ -168,6 +363,8 @@ class AddonTest(unittest.TestCase):
             {
                 "snapform_depth_source",
                 "snapform_displacement_via",
+                "snapform_unshaded",
+                "snapform_remove_background",
                 "snapform_model",
                 "snapform_use_ray_pose",
                 "snapform_process_res_factor",
@@ -192,6 +389,10 @@ class AddonTest(unittest.TestCase):
             [item[0] for item in displacement_via["items"]],
             ["MODIFIER", "MATERIAL"],
         )
+        self.assertEqual(
+            operator.__annotations__["snapform_unshaded"]["name"],
+            "Unshaded",
+        )
         draw_source = inspect.getsource(operator.draw)
         self.assertIn(
             'if self.snapform_depth_source == "ESTIMATE":',
@@ -218,8 +419,16 @@ class AddonTest(unittest.TestCase):
             'estimate_box.prop(self, "snapform_use_ray_pose")',
             draw_source,
         )
+        self.assertIn(
+            'layout.prop(self, "snapform_remove_background")',
+            draw_source,
+        )
         self.assertNotIn("layout.separator", draw_source)
         self.assertNotIn("layout.label", draw_source)
+        self.assertIn(
+            'layout.prop(self, "snapform_unshaded")',
+            draw_source,
+        )
         model_items = operator.__annotations__["snapform_model"]["items"]
         model_items = model_items(None, None)
         self.assertEqual(
@@ -266,6 +475,12 @@ class AddonTest(unittest.TestCase):
         self.assertNotIn("DA3 image depth", starting_source)
         self.assertNotIn("DA3 depth plane", finish_source)
 
+    def test_depth_background_is_loaded_without_changing_source_empty(self):
+        source = inspect.getsource(self.addon.RunJob._build_depth_plane)
+
+        self.assertIn("snapform_background_path=", source)
+        self.assertNotIn("_apply_removed_background", source)
+
     def test_model_options_mark_only_models_that_are_not_downloaded(self):
         items_callback = self.addon.SnapFormSettings.__annotations__["model"][
             "items"
@@ -288,9 +503,12 @@ class AddonTest(unittest.TestCase):
         operator.snapform_use_ray_pose = True
         operator.snapform_process_res_factor = 0.5
         operator.snapform_displacement_via = "MATERIAL"
+        operator.snapform_unshaded = True
+        operator.snapform_remove_background = True
         operator.snapform_subdivision_level = 5
         operator.snapform_delete_original = True
         image = SimpleNamespace(
+            name="img.png",
             filepath="",
             packed_file=object(),
             size=(1920, 1080),
@@ -326,6 +544,9 @@ class AddonTest(unittest.TestCase):
             calls[0]["snapform_displacement_via"],
             "MATERIAL",
         )
+        self.assertTrue(calls[0]["snapform_unshaded"])
+        self.assertTrue(calls[0]["snapform_remove_background"])
+        self.assertEqual(calls[0]["snapform_depth_data_name"], "img.depth")
         self.assertTrue(calls[0]["snapform_delete_input"])
         self.assertEqual(
             self.addon.image_depth._last_depth_plane_model,
@@ -454,6 +675,81 @@ class AddonTest(unittest.TestCase):
                     snapform_ray_pose,
                 )
 
+    def test_image_depth_background_option_requires_ben2(self):
+        settings = SimpleNamespace(
+            input_path="debug-input.png",
+            export_format="mini_npz",
+            model="BASE",
+            gs_model="GIANT_1_1",
+        )
+        operator = self.addon.RunJob()
+        operator.job_type = self.addon.operators.IMAGE_DEPTH_JOB
+        operator.snapform_model_key = "BASE"
+        operator.snapform_use_ray_pose = True
+        operator.snapform_process_res = 1024
+        operator.snapform_remove_background = True
+
+        with TemporaryDirectory() as directory:
+            input_path = Path(directory) / "input.png"
+            input_path.touch()
+            operator.snapform_input_path = str(input_path)
+            with patch.object(
+                self.addon.runtime,
+                "new_job_paths",
+                return_value={
+                    "log": Path(directory) / "job.log",
+                    "status": Path(directory) / "job.json",
+                    "result": Path(directory) / "result",
+                },
+            ), patch.object(
+                self.addon.runtime,
+                "runtime_ready",
+                return_value=True,
+            ), patch.object(
+                self.addon.runtime,
+                "model_ready",
+                side_effect=lambda key: key == "BASE",
+            ):
+                with self.assertRaisesRegex(RuntimeError, "BEN2 Base"):
+                    operator._prepare_command(None, settings)
+
+    def test_image_depth_background_option_reaches_worker_command(self):
+        runtime = importlib.import_module("snapform.runtime")
+        settings = SimpleNamespace(
+            input_path="input.png",
+            model="BASE",
+            gs_model="GIANT_1_1",
+            process_res=504,
+            process_res_method="upper_bound_resize",
+            conf_thresh_percentile=40.0,
+            num_max_points=1_000_000,
+            show_cameras=True,
+            gs_views_interval=1,
+            video_frames=16,
+            device="AUTO",
+            use_ray_pose=False,
+            ref_view_strategy="saddle_balanced",
+        )
+        with patch.object(
+            runtime,
+            "runtime_python",
+            return_value=Path("python"),
+        ), patch.object(
+            runtime,
+            "model_directory",
+            side_effect=lambda key: Path(key.lower()),
+        ):
+            command = runtime.reconstruct_command(
+                settings,
+                "mini_npz",
+                Path("status.json"),
+                Path("result"),
+                remove_background=True,
+            )
+
+        option_index = command.index("--remove-background-model-dir")
+        self.assertEqual(command[option_index + 1], str(Path("ben2_base")))
+
     def test_preferences_use_package_identifier(self):
         self.assertEqual(
             self.addon.SnapFormPreferences.bl_idname,
@@ -625,6 +921,13 @@ class AddonTest(unittest.TestCase):
         self.assertNotIn('text="Missing"', draw_source)
         self.assertNotIn("License:", draw_source)
         self.assertIn('operator.job_type = "RECONSTRUCT"', draw_source)
+        self.assertIn('text="BEN2 Background Removal"', draw_source)
+        self.assertIn(
+            'background_box.prop(settings, "background_input_path")',
+            draw_source,
+        )
+        self.assertIn('text="Remove Background"', draw_source)
+        self.assertIn('operator.job_type = "REMOVE_BACKGROUND"', draw_source)
 
     def test_runtime_models_panel_assigns_each_download_target(self):
         draw_source = inspect.getsource(
@@ -635,6 +938,10 @@ class AddonTest(unittest.TestCase):
         )
 
         self.assertIn("for model in MODELS.values()", draw_source)
+        self.assertIn(
+            "self._draw_model(layout, BEN2_MODEL, running)",
+            draw_source,
+        )
         self.assertIn("operator.model_key = model.key", model_source)
         self.assertIn(
             "download_row.enabled = runtime_ready() and not running",
@@ -648,6 +955,191 @@ class AddonTest(unittest.TestCase):
             "Setup",
         )
 
+    def test_ben2_debug_job_uses_its_image_and_model(self):
+        settings = SimpleNamespace(
+            background_input_path="input.png",
+            device="CUDA",
+        )
+        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,
+            ) as model_ready, patch.object(
+                self.addon.runtime,
+                "remove_background_command",
+                return_value=["worker"],
+            ) as remove_background:
+                command, prepared_paths = operator._prepare_command(
+                    None,
+                    settings,
+                )
+
+        self.assertEqual(command, ["worker"])
+        self.assertEqual(prepared_paths, paths)
+        model_ready.assert_called_once_with("BEN2_BASE")
+        self.assertEqual(remove_background.call_args.args[1], input_path)
+        self.assertEqual(remove_background.call_args.args[4], "BEN2_BASE")
+
+    def test_image_background_job_uses_operator_input(self):
+        settings = SimpleNamespace(device="AUTO")
+        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,
+                "remove_background_command",
+                return_value=["worker"],
+            ) as remove_background:
+                command, prepared_paths = operator._prepare_command(
+                    None,
+                    settings,
+                )
+
+        self.assertEqual(command, ["worker"])
+        self.assertEqual(prepared_paths, paths)
+        self.assertEqual(remove_background.call_args.args[1], input_path)
+
+    def test_completed_image_background_job_applies_result_with_undo(self):
+        calls = []
+        self.fake_bpy.ops = SimpleNamespace(
+            snapform=SimpleNamespace(
+                apply_removed_background=lambda **options: calls.append(
+                    ("apply", options)
+                )
+                or {"FINISHED"},
+            ),
+            ed=SimpleNamespace(
+                undo_push=lambda **options: calls.append(("undo", options))
+                or {"FINISHED"},
+            ),
+        )
+        operator = self.addon.RunJob()
+        job = SimpleNamespace(
+            result_directory=Path("result"),
+            source_object_name="Reference",
+        )
+
+        operator._apply_removed_background(job)
+
+        self.assertEqual(
+            calls,
+            [
+                (
+                    "apply",
+                    {
+                        "snapform_source_object_name": "Reference",
+                        "snapform_output_path": str(
+                            Path("result") / "foreground.png"
+                        ),
+                    },
+                ),
+                ("undo", {"message": "Remove Background"}),
+            ],
+        )
+
+    def test_completed_original_background_job_builds_with_one_undo(self):
+        calls = []
+        self.fake_bpy.ops = SimpleNamespace(
+            snapform=SimpleNamespace(
+                build_original_depth_plane=lambda **options: calls.append(
+                    ("build", options)
+                )
+                or {"FINISHED"},
+            ),
+            ed=SimpleNamespace(
+                undo_push=lambda **options: calls.append(("undo", options))
+                or {"FINISHED"},
+            ),
+        )
+        operator = self.addon.RunJob()
+        job = SimpleNamespace(
+            result_directory=Path("result"),
+            source_object_name="Reference",
+            subdivision_level=5,
+            delete_source=False,
+            displacement_via="MODIFIER",
+            unshaded=False,
+        )
+
+        operator._build_original_depth_plane(job)
+
+        self.assertEqual(calls[0][0], "build")
+        self.assertEqual(
+            calls[0][1]["snapform_background_path"],
+            str(Path("result") / "foreground.png"),
+        )
+        self.assertFalse(calls[0][1]["snapform_delete_source"])
+        self.assertEqual(
+            calls[1],
+            ("undo", {"message": "Convert to Depth Plane"}),
+        )
+
+    def test_ben2_download_command_filters_large_alternative_weights(self):
+        runtime = importlib.import_module("snapform.runtime")
+        with patch.object(
+            runtime,
+            "runtime_python",
+            return_value=Path("python"),
+        ), patch.object(
+            runtime,
+            "model_directory",
+            return_value=Path("model"),
+        ):
+            command = runtime.download_model_command(
+                "BEN2_BASE",
+                Path("status.json"),
+            )
+
+        patterns = [
+            command[index + 1]
+            for index, value in enumerate(command)
+            if value == "--allow-pattern"
+        ]
+        self.assertEqual(patterns, ["config.json", "model.safetensors"])
+
     def test_preferences_install_button_is_disabled_while_busy(self):
         draw_source = inspect.getsource(self.addon.SnapFormPreferences.draw)
 

tests/test_bootstrap_runtime.py

@@ -31,6 +31,7 @@ class BootstrapRuntimeTest(unittest.TestCase):
                 runtime_dir=str(runtime_dir),
                 backend="cu128",
                 da3_commit="test-commit",
+                ben2_commit="test-ben2-commit",
                 status=str(root / "status.json"),
             )
             with patch.object(bootstrap, "run") as mocked_run:
@@ -58,6 +59,22 @@ class BootstrapRuntimeTest(unittest.TestCase):
         )
         self.assertEqual(gsplat_call.args[1], {"BUILD_NO_CUDA": "1"})
 
+        ben2_command = next(
+            command
+            for command in commands
+            if any("ben2 @" in str(value) for value in command)
+        )
+        self.assertTrue(
+            any("test-ben2-commit" in str(value) for value in ben2_command)
+        )
+        check_command = commands[-1]
+        self.assertTrue(
+            any(
+                "from ben2 import BEN_Base" in str(value)
+                for value in check_command
+            )
+        )
+
 
 if __name__ == "__main__":
     unittest.main()

tests/test_image_depth.py

@@ -298,9 +298,52 @@ class ImageDepthTest(unittest.TestCase):
         original_source = inspect.getsource(
             self.image_depth.build_image_depth_plane
         )
-        self.assertIn("depth_name = depth_data_name(source_object)", estimate_source)
+        self.assertIn(
+            "depth_name = depth_name or depth_data_name(source_object)",
+            estimate_source,
+        )
         self.assertIn("depth_data_name(source_object)", original_source)
 
+    def test_removed_background_name_uses_source_image_base_name(self):
+        source_object = SimpleNamespace(
+            data=SimpleNamespace(name="img.png"),
+        )
+
+        self.assertEqual(
+            self.image_depth.removed_background_data_name(source_object),
+            "img.rb",
+        )
+        source_object.data.name = "img.png.001"
+        self.assertEqual(
+            self.image_depth.removed_background_data_name(source_object),
+            "img.001.rb",
+        )
+
+    def test_depth_name_override_keeps_original_base_name(self):
+        source = inspect.getsource(self.image_depth.build_depth_plane)
+
+        self.assertIn(
+            "depth_name = depth_name or depth_data_name(source_object)",
+            source,
+        )
+        self.assertIn(
+            "color_image if color_image is not None else source_object.data",
+            source,
+        )
+
+    def test_original_depth_background_only_changes_material_color(self):
+        source = inspect.getsource(self.image_depth.build_image_depth_plane)
+
+        self.assertIn("texture.image = source_object.data", source)
+        self.assertIn(
+            "source_object.data if displacement_via == \"MATERIAL\" else None",
+            source,
+        )
+        self.assertIn(
+            "color_image if color_image is not None else source_object.data",
+            source,
+        )
+
     def test_material_uses_source_object_name(self):
         da3_source = inspect.getsource(self.image_depth.build_depth_plane)
         image_source = inspect.getsource(
@@ -319,7 +362,10 @@ class ImageDepthTest(unittest.TestCase):
             self.image_depth.build_depth_plane
         )
 
-        self.assertIn("depth_name = depth_data_name(source_object)", source)
+        self.assertIn(
+            "depth_name = depth_name or depth_data_name(source_object)",
+            source,
+        )
         self.assertIn("create_depth_image(\n        depth_name,", source)
         self.assertIn("bpy.data.textures.new(\n            depth_name,", source)
         self.assertIn('texture.extension = "EXTEND"', source)
@@ -385,6 +431,42 @@ class ImageDepthTest(unittest.TestCase):
         )
         self.assertIn("set_displacement_only(material)", source)
 
+    def test_unshaded_material_uses_alpha_mixed_emission(self):
+        source = inspect.getsource(self.image_depth.create_material)
+
+        unshaded_start = source.index("if unshaded:")
+        shaded_start = source.index("else:", unshaded_start)
+        displacement_start = source.index(
+            "if displacement_image is not None:",
+            shaded_start,
+        )
+        unshaded_source = source[unshaded_start:shaded_start]
+        shaded_source = source[shaded_start:displacement_start]
+
+        self.assertIn('nodes.new("ShaderNodeEmission")', unshaded_source)
+        self.assertIn(
+            'nodes.new("ShaderNodeBsdfTransparent")',
+            unshaded_source,
+        )
+        self.assertIn('nodes.new("ShaderNodeMixShader")', unshaded_source)
+        self.assertIn(
+            'color_texture.outputs["Color"]',
+            unshaded_source,
+        )
+        self.assertIn('emission.inputs["Color"]', unshaded_source)
+        self.assertIn('color_texture.outputs["Alpha"]', unshaded_source)
+        self.assertIn('mix_shader.inputs["Fac"]', unshaded_source)
+        self.assertIn('transparent.outputs["BSDF"]', unshaded_source)
+        self.assertIn('mix_shader.inputs[1]', unshaded_source)
+        self.assertIn('emission.outputs["Emission"]', unshaded_source)
+        self.assertIn('mix_shader.inputs[2]', unshaded_source)
+        self.assertIn('mix_shader.outputs["Shader"]', unshaded_source)
+        self.assertNotIn("ShaderNodeBsdfPrincipled", unshaded_source)
+        self.assertNotIn("ShaderNodeBump", unshaded_source)
+        self.assertNotIn("ShaderNodeFloatCurve", unshaded_source)
+        self.assertIn("ShaderNodeBsdfPrincipled", shaded_source)
+        self.assertGreater(displacement_start, unshaded_start)
+
     def test_displacement_values_follow_depth_source(self):
         estimate_source = inspect.getsource(
             self.image_depth.build_depth_plane
@@ -434,6 +516,7 @@ class ImageDepthTest(unittest.TestCase):
                 'if displacement_via == "MATERIAL" else None',
                 source,
             )
+            self.assertIn("unshaded=unshaded", source)
 
     def test_surface_displacement_is_set_to_displacement_only(self):
         class Material:

tests/test_models.py

@@ -43,3 +43,14 @@ class ModelTest(unittest.TestCase):
     def test_unknown_model_is_rejected(self):
         with self.assertRaisesRegex(ValueError, "Unknown DA3 model"):
             self.models.get_model("UNKNOWN")
+
+    def test_ben2_download_only_includes_required_hugging_face_files(self):
+        model = self.models.get_downloadable_model("BEN2_BASE")
+
+        self.assertEqual(model.repository, "PramaLLC/BEN2")
+        self.assertEqual(model.license_name, "MIT")
+        self.assertEqual(
+            model.download_patterns,
+            ("config.json", "model.safetensors"),
+        )
+        self.assertNotIn("BEN2_BASE", self.models.MODELS)

tests/test_pack.py

@@ -29,7 +29,7 @@ class PackTest(unittest.TestCase):
             finally:
                 pack.DIST_DIR = original_dist
 
-            self.assertEqual(output_path.name, "SnapForm.v0.9.27.zip")
+            self.assertEqual(output_path.name, "SnapForm.v0.9.28.zip")
             with ZipFile(output_path) as archive:
                 names = set(archive.namelist())
 

tests/test_worker.py

@@ -47,6 +47,52 @@ class WorkerTest(unittest.TestCase):
             source.index("import torch"),
         )
 
+    def test_optional_background_removal_runs_before_da3_model_load(self):
+        source = inspect.getsource(self.worker.reconstruct)
+
+        self.assertLess(
+            source.index("infer_foreground("),
+            source.index("DepthAnything3.from_pretrained"),
+        )
+        args = self.worker.parser().parse_args(
+            [
+                "reconstruct",
+                "--input",
+                "input.png",
+                "--model-dir",
+                "da3",
+                "--output-dir",
+                "result",
+                "--export-format",
+                "mini_npz",
+                "--status",
+                "status.json",
+                "--remove-background-model-dir",
+                "ben2",
+                "--process-res",
+                "1024",
+                "--process-res-method",
+                "upper_bound_resize",
+                "--conf-thresh-percentile",
+                "40",
+                "--num-max-points",
+                "1000",
+                "--show-cameras",
+                "0",
+                "--gs-views-interval",
+                "1",
+                "--video-frames",
+                "16",
+                "--device",
+                "auto",
+                "--use-ray-pose",
+                "1",
+                "--ref-view-strategy",
+                "first",
+            ]
+        )
+        self.assertEqual(args.remove_background_model_dir, "ben2")
+
     def test_prepares_a_single_image_file(self):
         with tempfile.TemporaryDirectory() as directory:
             image_path = Path(directory) / "input.png"
@@ -69,6 +115,117 @@ 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"],
+        )
+
+    def test_ben2_background_removal_writes_fast_transparent_png(self):
+        inference_options = []
+        save_options = []
+
+        class FakeImage:
+            def __enter__(self):
+                return self
+
+            def __exit__(self, *_args):
+                return None
+
+            def convert(self, mode):
+                self.mode = mode
+                return self
+
+            def save(self, path, **options):
+                save_options.append(options)
+                Path(path).touch()
+
+        class FakeModel:
+            @classmethod
+            def from_pretrained(cls, model_dir):
+                self.assertEqual(model_dir, "model")
+                return cls()
+
+            def to(model_self, device):
+                self.assertEqual(device.type, "cpu")
+                return model_self
+
+            def eval(self):
+                return None
+
+            def inference(self, image, **options):
+                inference_options.append((image, options))
+                return FakeImage()
+
+        fake_torch = ModuleType("torch")
+        fake_torch.cuda = SimpleNamespace(
+            is_available=lambda: False,
+            synchronize=lambda: None,
+        )
+        fake_torch.device = lambda device_type: SimpleNamespace(type=device_type)
+        fake_torch.inference_mode = nullcontext
+        fake_ben2 = ModuleType("ben2")
+        fake_ben2.BEN_Base = FakeModel
+        fake_pil = ModuleType("PIL")
+        fake_pil.Image = SimpleNamespace(open=lambda _path: FakeImage())
+
+        with tempfile.TemporaryDirectory() as directory, patch.dict(
+            sys.modules,
+            {
+                "torch": fake_torch,
+                "ben2": fake_ben2,
+                "PIL": fake_pil,
+            },
+        ):
+            root = Path(directory)
+            input_path = root / "input.png"
+            input_path.touch()
+            args = SimpleNamespace(
+                input=str(input_path),
+                model_dir="model",
+                output_dir=str(root / "result"),
+                status=str(root / "status.json"),
+                device="cpu",
+            )
+            self.worker.remove_background(args)
+            status = json.loads(Path(args.status).read_text(encoding="utf-8"))
+
+            self.assertTrue((root / "result" / "foreground.png").is_file())
+
+        self.assertEqual(
+            inference_options[0][1],
+            {"refine_foreground": False},
+        )
+        self.assertEqual(
+            save_options[0],
+            {"format": "PNG", "compress_level": 1},
+        )
+        self.assertIn("Background removed in", status["message"])
+
     def test_uses_official_da3_export_parameters_for_all_four_formats(self):
         inference_options = []
         export_calls = []

uv.lock

@@ -4,5 +4,5 @@ requires-python = ">=3.11"
 
 [[package]]
 name = "snapform"
-version = "0.9.27"
+version = "0.9.28"
 source = { virtual = "." }