| { |
| "id": "single-software-execution--cli--blender--task-13--ubuntu", |
| "snapshot": "Blender-4.2.3", |
| "instruction": "Compositor multi-output: beauty + depth. `/home/user/Desktop/scene.blend` contains: - Three cubes named `Cube_Y1`, `Cube_Y3`, `Cube_Y5` at y=1, y=3, y=5 respectively (staggered along X so they don't fully occlude one another). - A `Camera` at (0, -5, 0) rotated 90 degrees around X so it looks along +Y (the cubes appear at increasing depth from the camera). - A `Sun` light at (4, -4, 6). - Render engine BLENDER_EEVEE at 256x256 (samples 16). - The active view layer has `use_pass_z = True` (Z/Depth pass already enabled). - `scene.use_nodes = False` (the compositor node tree has not been built yet). Launch Blender by running `blender` from a terminal. For the steps below, use `/opt/blender/blender` for Blender command-line or background runs when working on `scene.blend`.\n\nTask: build a compositor node graph that writes two images side-by-side via a single File Output node. 1. Enable the compositor: - `scene.use_nodes = True` - `tree = scene.node_tree` - Clear any default nodes: `for n in list(tree.nodes): tree.nodes.remove(n)` 2. Add nodes: - `render_layers = tree.nodes.new('CompositorNodeRLayers')` - `map_range = tree.nodes.new('CompositorNodeMapRange')` - `inputs['From Min'].default_value = 0.0` - `inputs['From Max'].default_value = 20.0` - `inputs['To Min'].default_value = 0.0` - `inputs['To Max'].default_value = 1.0` - `use_clamp = True` - `file_out = tree.nodes.new('CompositorNodeOutputFile')` - Set `file_out.base_path` to the Desktop directory, for example `/home/user/Desktop` or `//` after saving the blend there. The final file paths will be `<base_path>/<slot_path>0001.png`. - Set `file_out.format.file_format = 'PNG'` (8-bit RGBA is fine). - Configure two file slots: - Repurpose the default slot as `beauty_`: `file_out.file_slots[0].path = 'beauty_'` - Add a depth slot: `file_out.file_slots.new('depth_')` 3. Wire the links: - `render_layers.outputs['Image']` -> `file_out.inputs[0]` (the `beauty_` slot) - `render_layers.outputs['Depth']` -> `map_range.inputs['Value']` - `map_range.outputs['Value']` -> `file_out.inputs[1]` (the `depth_` slot) 4. Ensure `scene.frame_start = scene.frame_end = 1` and render: `bpy.ops.render.render(write_still=True)` -- this triggers the File Output node to write `beauty_0001.png` and `depth_0001.png` in `file_out.base_path`. 5. Save the .blend to `/home/user/Desktop/answer.blend`. ## Requirements 1. The .blend opens. 2. The three cubes `Cube_Y1`, `Cube_Y3`, `Cube_Y5` have y=1, 3, 5 respectively. 3. The active view layer has `use_pass_z = True`. 4. `scene.use_nodes == True`. 5. The compositor node tree contains a `CompositorNodeOutputFile`. 6. The compositor node tree contains a `CompositorNodeMapRange`. 7. The File Output node has at least 2 `file_slots`. 8. After a fresh render, `beauty_0001.png` exists in `file_out.base_path`. 9. `depth_0001.png` exists in the same directory. 10. Sampling the depth PNG at the projected screen positions of the three cube centers (via `bpy_extras.object_utils.world_to_camera_view`), the R-channel values are strictly increasing along Cube_Y1 -> Cube_Y3 -> Cube_Y5 with each step delta >= 0.03 (closer cubes have lower depth because Map Range 0..20 -> 0..1 clamped).", |
| "source": "https://docs.blender.org/manual/en/4.1/; https://docs.blender.org/api/4.1/; https://developer.blender.org/docs/release_notes/4.1/; https://www.loc.gov/preservation/digital/formats/fdd/fdd000507.shtml", |
| "config": [ |
| { |
| "type": "upload_file", |
| "parameters": { |
| "files": [ |
| { |
| "local_path": "task-13/init_file/scene.blend", |
| "path": "/home/user/Desktop/scene.blend" |
| } |
| ] |
| } |
| } |
| ], |
| "trajectory": "trajectories/", |
| "related_apps": [ |
| "blender" |
| ], |
| "artifact_paths": [ |
| "/home/user/Desktop/0001.png", |
| "/home/user/Desktop/answer.blend", |
| "/home/user/Desktop/beauty_0001.png", |
| "/home/user/Desktop/depth_0001.png" |
| ], |
| "evaluator": { |
| "postconfig": [ |
| { |
| "type": "upload_file", |
| "parameters": { |
| "files": [ |
| { |
| "local_path": "task-13/eval.py", |
| "path": "/home/user/Desktop/eval.py" |
| } |
| ] |
| } |
| } |
| ], |
| "func": "exact_match", |
| "result": { |
| "type": "vm_command_line", |
| "command": "python3 /home/user/Desktop/eval.py", |
| "shell": "true" |
| }, |
| "expected": { |
| "type": "rule", |
| "rules": { |
| "expected": "True\n" |
| } |
| } |
| }, |
| "proxy": false, |
| "fixed_ip": false, |
| "possibility_of_env_change": "low" |
| } |
|
|
