Glossy and metallic parts
Painted bumpers, oily steel parts: many perspectives balance out reflections that fool a single view.
PLATFORM / LUMISCAN X
LumiScan X
Depth from 13 viewpoints. In a single shot.
The light-field camera for surfaces where 3D measurement gets difficult: glossy, metallic, film-wrapped. Its 3D data flows through the LumiScan Vision Suite straight into recognition, measurement and robot guidance.
- cameras exposing simultaneously
- 13
- shot for the depth of every pixel
- 1
- colour images
- 8/12 bit
- pattern projector for textureless surfaces
- SL
When light field pays off
Films and packaging
Vacuum-sealed PCBs and shrink-wrapped reels – depth stays measurable even when the film reflects.
Moving objects
All 13 cameras expose at the same time. Parts on a conveyor or on the robot are captured without motion artefacts.
Parts that overlap
Multiple baselines see around edges; the depth estimate stays robust against occlusion and noise.
How depth is created
Every view sees the same point at a slightly different position. From this shift – the disparity – LumiScan computes the distance. Try it:
- Disparity
- – px per view
- Computed depth
- – mm
Drag the yellow point or use the slider. Example values: 7 views, spacing b = 80 mm, focal length f = 200 px.
Start with the workpiece
The surface matters. So does the way it sits.
Parts overlap in a full bin. As the bin empties, different contours come into view and the space available to the gripper changes.
- Shiny or oily metal
- Check where reflections form and whether the part edges stand out from the bin. The light-field camera captures the same area from several viewpoints.
- Castings and complex components
- Turn the part during a trial. Which features does it hide from the camera, and from which directions can the gripper still reach it?
- Matte parts and regular shapes
- Check that the camera resolves enough detail across the working volume. When the surface has little texture, a projected pattern can help.
Key specifications
- Cameras
- 13, exposing simultaneously
- Reconstruction
- Epipolar constraint with multiple baselines
- Image data
- Colour, 8 or 12 bit
- Calibration
- Intrinsic and extrinsic parameters of all 13 cameras in one .hdvcal file
- Pattern projector
- LumiScan SL (optional) for textureless surfaces
- System requirements
- Windows 11 or Ubuntu 24.04, NVIDIA GPU with CUDA 11.8 (cuDNN 8.8.1)
Binding values are in the data sheet and the documentation.
Frequently asked questions
Why does a light-field camera work on shiny parts?
LumiScan X sees every point from 13 viewpoints at once. Reflections that fool a single view are balanced out by the others, and the depth of every pixel is computed from the shift between the views – in a single shot.
Can LumiScan X capture moving parts?
Yes. All 13 cameras expose at the same time, so parts on a conveyor or on the robot are captured without motion artefacts.
What about surfaces with little texture?
The optional LumiScan SL pattern projector gives textureless surfaces an artificial texture, so depth stays measurable.
What are the system requirements?
Windows 11 or Ubuntu 24.04 with an NVIDIA GPU and CUDA 11.8 (cuDNN 8.8.1). Binding values are in the data sheet and the documentation.
LUMISCAN X / DOWNLOADS
The details. To take with you.
Technical data and product information for the LumiScan X light-field camera, in one English data sheet.
Free · We’ll email you the download link.