FAQ
Getting started with 3D
Zivid使用什么深度传感技术？
Zivid uses Temporal Structured Light, which integrates a sequence of spatially encoded patterns to achieve high accuracy depth even on structure-less surfaces.
What is structured-light 3D, and how is it different from time-of-flight and stereo?
Zivid cameras use Temporal Structured Light: the projector casts a sequence of encoded light patterns, and the camera reconstructs an accurate 3D point cloud from how they deform on the surfaces. This gives high accuracy and dense data even on structure-less surfaces, where stereo (which relies on visible texture to match between two views) struggles. Time-of-flight (ToF) measures the travel time of light per pixel. Structured light typically achieves much higher accuracy and finer detail, which is what most robotics and inspection tasks need.
How does an industrial 3D camera differ from a consumer depth sensor?
Consumer depth sensors are designed for low cost and human-scale sensing, with relatively low accuracy and limited robustness to ambient light and difficult materials. Industrial 3D cameras like Zivid are built for repeatable sub-millimeter accuracy, robustness in factory conditions, calibrated and traceable measurements, and an SDK intended for production integration. For applications such as bin picking, inspection, and assembly, that accuracy and repeatability is what makes a reliable robot cell possible.
Cameras, field of view, and accuracy
Zivid相机的视野是多少？
距离 [毫米] |
视场 [毫米] |
|---|---|
1300 |
938 x 938 |
2500 |
1750 x 1750 |
5000 |
3537 x 3537 |
距离 [毫米] |
视场 [毫米] |
|---|---|
800 |
481×406 |
1300 |
793×658 |
2000 |
1220×1012 |
距离 [毫米] |
视场 [毫米] |
|---|---|
700 |
694×547 |
1100 |
1096×858 |
1700 |
1692×1324 |
距离 [毫米] |
视场 [毫米] |
|---|---|
300 |
234×236 |
600 |
579×469 |
1100 |
1096×858 |
距离 [毫米] |
视场 [毫米] |
|---|---|
500 |
514×320 |
700 |
754×449 |
1500 |
1519×959 |
距离 [毫米] |
视场 [毫米] |
|---|---|
600 |
644×372 |
1000 |
1147×680 |
1600 |
1618×988 |
要计算不同距离的 FOV，请单击 此处 。
What accuracy, trueness, and precision can the camera achieve?
Accuracy is described as a combination of point precision (the repeatability of individual XYZ measurements) and dimension trueness (how correct measured distances are). The table below gives the typical point precision and dimension trueness error at each model's focus distance.
Model |
Focus distance |
Point precision |
Dimension trueness error |
|---|---|---|---|
Zivid 3 XL250 |
2500 mm |
265 µm |
0.24 % |
Zivid 2+ MR130 / M130 |
1300 mm |
208 µm |
0.33 % |
Zivid 2+ LR110 / L110 |
1100 mm |
255 µm |
0.33 % |
Zivid 2+ MR60 / M60 |
600 mm |
71 µm |
0.16 % |
Zivid 2 M70 |
700 mm |
85 µm |
0.17 % |
Zivid 2 L100 |
1000 mm |
244 µm |
0.18 % |
See Zivid Specs Terminology for the definitions, and the datasheet for your model for the full numbers across the working range. To reach and keep the best accuracy, run 现场标定, allow the camera to warm up, and keep Thermal Stabilization（热稳定功能） enabled.
Zivid使用了哪种传感器？
相机 |
传感器 |
|---|---|
Zivid 3 |
8.07 MP CMOS |
Zivid 2+ |
5.10MP CMOS |
Zivid 2 |
2.86MP CMOS |
The sensor resolution equals the number of points in the point cloud; see 点云结构和输出格式 for the resolution per model and how to read it from the SDK. For how sensor resolution is defined, see Zivid Specs Terminology.
Zivid相机有多重？
相机 |
重量[公斤] |
|---|---|
Zivid 3 |
2.700 |
Zivid 2+ |
0.920 |
Zivid 2 |
0.945 |
Zivid相机的外壳是用什么材料制成的？
Zivid 相机由铝镁合金部件制成，外壳轻巧且坚固。
How do I choose an industrial 3D camera for my application?
Match the camera to your scene: the size of the objects, your working (standoff) distance, and the field of view you need to cover at that distance. Then check that the resulting spatial resolution is fine enough to capture your smallest important feature, and that the accuracy meets your tolerance. For Zivid specifically, the number in a model name is its focus distance in centimeters (for example, the Zivid 2+ MR130 is focused at 130 cm). Use the field-of-view and imaging-distance calculator to match a model to your scene.
What accuracy can an industrial 3D camera achieve, and how much do I need?
Zivid cameras reach sub-millimeter point precision and dimension trueness on the order of 0.2 % at a given imaging distance. For example, a Zivid 2+ MR130 at 1000 mm measures a 100 mm distance to within about ±0.2 mm. The exact numbers depend on the model and the imaging distance. The accuracy you need depends on your task: bin picking, robot guidance, and assembly typically needs sub-millimeter, while picking larger parts or boxes can tolerate more. See Zivid Specs Terminology for how precision and trueness are defined, and the datasheet for your camera model for exact numbers at your working distance.
Capture quality and difficult materials
Which 3D camera works for shiny, transparent, or dark parts?
All Zivid models handle these difficult materials well, so the choice is driven by your object size and working distance (see the camera table above), not by the material. For highly specular metal parts and bins, the Sage Engine gives the best results and is available on the Zivid 2+ MR130, MR60, and LR110. The Omni Engine, for transparent and reflective scenes, is available on all Zivid 2+ models. Zivid 3 is especially robust to ambient light and reflections thanks to its significantly stronger projector.
How well does the camera capture shiny, reflective, or specular metal parts?
Shiny and specular metal is challenging for any 3D camera, but Zivid is designed to produce good data on these surfaces. Use the Manufacturing Specular or Semi-Specular presets, which provide the high dynamic range needed for specular surfaces and which are good for strong inter-reflections. A dark, absorptive background reduces inter-reflections on cylindrical or spherical parts. See 处理高光和闪亮的物体 and 材料的光学特性.
Can the camera capture transparent or translucent objects?
Transparent and translucent objects are challenging because light passes through them, but they can be captured with the right setup. Use the Omni Engine (Zivid 2+) with the Consumer Goods presets. Keep the object close to the focus distance with a uniform gap to the background. See Dealing with Transparent Objects.
How well does the camera capture dark, black, or matte surfaces?
Dark and matte surfaces reflect little light and are challenging, but Zivid is designed to produce usable data on them. Use Specular-family presets with longer exposure times with HDR. See 处理高光和闪亮的物体 and 材料的光学特性.
Which presets and settings should I use?
Start from the preset closest to your application (the Manufacturing, Consumer Goods, and Inspection families), then fine-tune the acquisitions, aperture, and exposure. See 捕获高质量的点云 for a full guide.
Can Zivid cameras be used outdoors or under direct sunlight?
The upper limit that we use when testing is approximately 1500-3000 lux. We measure by positioning the LUX meter across different areas of the camera's FOV at the imaging distance, and take the maximum value as the ambient light level. Direct sunlight (32,000-100,000+ lux) is not feasible for most cases; physical shading of the scene is required. Under some outdoor conditions, where ambient light stays within these limits, the Zivid 3 can be used outdoors thanks to its significantly stronger projector.
Performance and speed
What capture and cycle time can the camera achieve?
With the right settings and good hardware, capture times below 100 ms are achievable, but only for certain applications and distances. Other, more challenging applications at long distances may require captures of 1 s or longer. Capture time depends on the number of acquisitions, exposure, and processing, and on the host hardware (the network card and the GPU). Reduce time with fewer acquisitions, subsampling, and a region of interest. Use the 计算3D捕获速度 article for representative benchmarked numbers, and see 捕获速度 for measured capture-speed benchmarks.
Can the camera capture moving objects?
Zivid cameras can capture slowly moving objects, but feasibility depends on the speed, the field of view, and the required quality. Motion requires a shorter capture, which you get with a fast, single acquisition and 4x4 subsampling, as used in the Parcels Fast settings. See 计算3D捕获速度 for representative numbers, and contact customersuccess@zivid.com to discuss feasibility for your application.
我可以在牺牲质量和/或分辨率的情况下提高帧速率吗？
通过应用子采样/下采样，请参阅 Sampling（采样） - 3D ，可以通过降低分辨率来减少采集和捕获时间。尽管分辨率降低了，但值得注意的是，点云的质量实际上可能会得到增强。
相机最大占用带宽是多少？
相机 |
带宽 [Gbps] |
|---|---|
Zivid 3 |
10.0 |
Zivid 2+ |
7.07 |
Zivid 2 |
5.37 |
请查看 计算3D捕获速度 以获取更多具有代表性的数据。
What computer and GPU are required to run the camera?
A dedicated GPU is strongly recommended; integrated graphics works but is significantly slower. See GPU的要求 for requirements and 推荐的工业 PC for tested PCs.
Which industrial PC and network card do you recommend?
For best performance, use a dedicated GPU together with a 10 GigE network card. Some industrial PCs ship with 2.5 GigE networking, so confirm a 10 GigE add-in card option with the supplier when you need full bandwidth. See 推荐的工业 PC for tested industrial PCs, GPUs, and 10 GigE network cards.
Software, API, and integration
Zivid支持哪些API？
C++
C#/.NET
Python
GenICam/HALCON（Experimental）
ROS
NVIDIA Isaac Sim extension (Experimental, for simulation)
Zivid是否支持GenICam？
Zivid Software 包含了一个 GenICam GenTLproducer （Experimental）
已经测试过的Zivid相机与GenICam配合使用的HALCON版本包括：
19.05 Progress, 20.05 Progress, 21.11 Progress, 24.05 Progress, 24.11 Progress-Steady, 25.05 Progress
对于其他GenICam GenTL用户，我们提供了一个技术预览。这可能允许在其他软件包中使用Zivid相机，这些软件包可以通过连接GenICam GenTL来兼容相机。
Zivid支持哪些操作系统？
Windows: 10 / 11
Linux：Ubuntu 20.04/22.04/24.04
备注
Support for older operating system versions has been removed over time; see the SDK Changelog for which versions a given SDK release supports.
使用Zivid Studio和API得到点云数据会有所不同吗？
No, for the same settings the data is identical between Zivid Studio and the API. The reason captures often look different is the starting point, not the data. Zivid Studio guides you to pick a tuned preset, while a minimal API capture uses the default settings. To reproduce a Studio capture exactly, export the settings as a YML file from Zivid Studio and load that file through the SDK. This is the recommended way to keep Studio and your application in sync.
如何处理Zivid输出的点云数据？
Zivid does not provide point cloud processing software. HALCON and EyeVision are examples of paid processing software while OpenCV, Open3D, and PCL are free alternatives to use depending on the target application.
文件可以以什么格式保存？
The point cloud files can be saved as .zdf, which is the format that Zivid Studio uses. You can also export the file as a PLY file (*.ply as ordered or unordered) / ASCII points file (*xyz) / Point cloud data file (*.pcd). See 点云结构和输出格式 for the point cloud structure and outputs.
Does Zivid work with ROS, my robot, and HALCON?
Zivid is robot-agnostic and integrates through standard interfaces. The SDK is available for C++, C#/.NET, and Python, with an official ROS / ROS2 wrapper (zivid-ros) and an Experimental GenICam GenTL producer tested with HALCON. The camera outputs point clouds in its coordinate frame; relating that to a specific robot or PLC is done through hand-eye calibration and your integration code.
Does Zivid include picking, detection, or inspection software?
Zivid provides the camera and the SDK, which deliver a calibrated, true-color point cloud plus tools such as region of interest, normals, downsampling, and hand-eye calibration. Zivid also provides motion planning software. Object detection, segmentation, pose estimation, and CAD comparison are done with your own software or third-party packages. See How can I process the point cloud data from Zivid? for examples of compatible software.
SDK development
How do I get the camera intrinsics?
The SDK exposes camera intrinsics in both OpenCV and HALCON models, and you can read them from a connected camera, from a frame, or from a saved file. Each camera exposes two sets: fixed intrinsics for the camera and estimated intrinsics that account for the settings used in a capture. See 相机内参 for how to obtain them.
How do I set capture settings in code, and how do I find the valid range for a parameter?
Settings such as exposure time, aperture, and brightness are set through the SDK just as in Zivid Studio. The valid range for each can differ between camera models. A value outside the valid range is rejected when the camera captures, so query the camera's settings information to read the supported ranges for your model rather than hard-coding them. The ExploreSettingsMetaData sample shows how to read the valid range of each setting. The simplest way to reproduce a known-good capture is to export the settings as a YML file from Zivid Studio and load it through the SDK.
Can I update the camera firmware from code?
Yes. Firmware can be updated programmatically instead of through Zivid Studio; see the FirmwareUpdater sample in C++, Python, and C#.
Which ROS version does Zivid support?
Zivid provides an official wrapper, zivid-ros, for both ROS and ROS 2. Match the wrapper release to your SDK version, as noted in the wrapper's documentation.
Does Zivid run on ARM platforms such as NVIDIA Jetson?
The SDK supports Ubuntu on ARM, including NVIDIA Jetson, and can run inside a Docker container. See 在 Jetson Linux 上安装 Zivid for Jetson setup and Docker for running in a container.
Connectivity and hardware
Zivid 相机可以使用哪种以太网电缆？
The Zivid cameras come with a 5 m CAT6 Ethernet cable. Zivid also provides Ethernet cables in 5 m, 10 m, and 25 m options. See 数据电缆 for more info. Zivid strongly recommends that the cables are carefully checked before use or if run time errors occur. It is also strongly recommended to tighten the screw on the camera side when setting up the camera.
What power supply does the camera need?
Zivid cameras are powered from 24 V DC. We strongly recommend using the Zivid Power Supply and Power Extension Cables to ensure compliance with Safety and EMC emission and immunity standards. If you must use a third-party supply, a 24 V industrial supply, or robot internal wiring, see that page for the requirements the power must meet.
Can I connect to a camera by its IP address in the SDK?
Yes.
You can connect by serial number, or directly by IP address or hostname using a CameraAddress.
See the connect-by-IP section of the 捕获教程.
Mounting, calibration, and multiple cameras
Zivid执行什么样的标定？
Zivid相机在出厂时已经过预标定，可以随时使用。我们的浮动标定将温度和光圈（光圈）也考虑在内。相机提供的点云是经过标定的，也就是说X、Y、Z是真实场景的正确测量值。 X、Y和Z矩阵以及彩色图像存在像素到像素的对应关系。
Zivid 还提供 现场标定 工具，旨在验证和校正 Zivid 相机在现场的尺寸准确度。
手眼标定用于标定相机到机器人，即找到相机和机器人之间的坐标系转换关系。 Zivid有一个官方工具来帮助客户实 现 手眼标定 标定。
What is in-field correction, and which values are acceptable?
In-field correction verifies and corrects the camera's dimension trueness in the field. A corrected dimension trueness within roughly 0.1 % is expected, but it depends on the camera model. It is useful if you suspect accuracy has drifted, for example after a robot collision. It also helps tighten accuracy to the best achievable for your field of view and imaging range. See 现场标定 for an overview and 进行现场标定的准则 for board distances, repeated measurements, and how to judge the values; if results stay far outside the expected range, contact customersuccess@zivid.com.
Which calibration board should I use?
The correct board depends on your camera model and working distance, and also on your robot mounting and application. See 现场标定 for the board-to-model recommendations (for example ZVDA-CB01 vs ZVDA-CB02), and Zivid 标定对象 for the calibration objects used in hand-eye.
Can the camera be mounted on a robot arm as well as stationary?
Yes. The camera can be mounted on a robot arm (on-arm / eye-in-hand), stationary (eye-to-hand), or on a tripod. On-arm mounting lets a single camera cover several positions and keeps it at the optimal working distance. Stationary mounting is simpler and keeps the camera out of the robot motion path. See 安装 and 手臂安装方案的机械注意事项.
Can multiple Zivid cameras be used together?
Yes. Multiple cameras can be combined into one coordinate system with 多相机标定, and their captures can be combined with 拼接. Because the cameras project light, capturing simultaneously can add noise where the projections overlap, so sequential capture may be needed. See 运行多台Zivid相机 and 关于多个Zivid相机同时工作的性能的考虑因素.
Applications
Is Zivid suitable for inspection and CAD comparison?
Zivid delivers an accurate, calibrated point cloud that you can compare against a nominal CAD model or use for dimensional and surface inspection. The camera and SDK provide the 3D data; the comparison and measurement logic itself is done in your own or third-party software. See How can I process the point cloud data from Zivid? for software options.
Is Zivid used for robotic welding?
Yes, Zivid cameras are used in robotic welding applications, both before and after welding. Before welding, the camera locates the parts and the area to be welded, so the robot can find the seams and edges to follow. After welding, the same 3D data can be used to inspect the weld. The camera provides the 3D data; the weld-path planning and robot control are handled by your welding software. Capturing weld edges and features well relies on the same settings as other reflective-metal scenes, see 处理高光和闪亮的物体.
Can a 3D camera guide a robot to pick and place parts?
Yes. A 3D camera gives the robot the shape and position of objects so it can compute where and how to grasp them. This is the basis of 3D-guided robotics such as bin picking, machine tending, piece picking in logistics, parcel induction, and assembly. The camera is aligned to the robot through hand-eye calibration, which relates the camera's coordinate system to the robot's so a detected point becomes a robot motion.
What is 3D bin picking, and what camera do I need for it?
Bin picking is using a 3D camera and a robot to locate and pick parts out of a bin or container, including randomly placed or mixed parts. It needs a camera with enough field of view to see the whole bin, and enough resolution and accuracy to localize the parts and plan a safe grasp. Good performance on the part material (for example shiny or dark surfaces) also matters. See the 料箱拣选和设备管理的应用教程 for camera selection, field-of-view and clearance considerations, and a full walkthrough.
CAD files and FOV models
Where can I find CAD/STEP files and FOV models?
CAD/STEP files for the cameras (with field-of-view) and mounts are available on the Zivid website.
Maintenance and safety
我可以在什么温度范围内安全地运行Zivid？
Zivid 相机设计的工作温度范围为 0˚C 至 45˚C。
我们强烈建议在指定的温度范围内操作相机。
小心
在温度范围之外使用相机可能会损坏相机并使您的保修失效。在超出规格书建议的使用环境中使用相机需要自行承担风险。
小技巧
Zivid factory calibration takes temperature into account. However, to further reduce the impact of temperature dependent performance factors, keep Thermal Stabilization（热稳定功能） enabled.
What are the camera's environmental ratings (IP rating, temperature)?
All Zivid cameras are rated IP65 for ingress protection and are designed to operate within 0-45 ˚C. Refer to the datasheet for your camera model for the full environmental specifications. For harsh environments outside these ratings, see the protective housing options.
Can the camera image through a protective glass window?
Yes, the camera can image through a window when it is mounted inside a protective housing. Use a flat, optically clear window and keep it clean and free of condensation; an anti-reflection coating reduces glare and stray reflections. See 在装有玻璃镜片的防护罩内使用Zivid相机 for guidance on housing and window choice.
Can I clean dust off the camera's fan with compressed air?
Yes, but do not aim compressed air at the fan in a way that spins it faster than its normal running speed. The fan has no built-in protection against being spun too fast, and spinning it too fast with compressed air can damage it. See 服务和维修 for general cleaning guidance.
Zivid相机的预期使用寿命是多少？
Zivid cameras are designed to capture 25 million point clouds in its lifetime.
这不应被误认为是预期的平均故障时间 (MTTF/MTBF)。
Zivid使用的结构光是在可见光谱中吗？
是的，Zivid 相机利用可见光呈现鲜艳的色彩。
Zivid使用的结构光对人体有害吗？
与任何明亮的光源一样，请勿直视光束。由于对非常亮的光源有厌光反应（例如转头或眨眼），光源不应对眼睛造成危害。然而，为了安全起见，应避免在操作过程中直视相机/投影仪。
更多详细信息，请参阅 监管信息 。
Zivid 3D相机的有哪些光等级认证？
Zivid 3D 相机使用的光源属于风险等级 2。
请参阅相机背面的标签。如需了解更多详细信息，请查看 监管信息 。
Zivid遵循哪些安全标准？
To read about Safety before using the Zivid camera, check Regulatory Information.
如何访问日志文件以转发给Zivid获取支持？
您可以在Windows上的路径 %LOCALAPPDATA%\Zivid\API\Log 和Ubuntu上的路径 $XDG_CACHE_HOME/Zivid/API/Log 中找到日志。
小心
如果未设置 $XDG_CACHE_HOME，请使用 $HOME/.cache。