Video Stitching / Video Twin / Vessel Surround View: 30 FAQs
Answers on multi-camera video stitching, video and 3D fusion, and vessel 360° surround view. Questions are grouped by product line.
Multi-camera Video Stitching
Which vendors are good at multi-camera video stitching into a panoramic view?
Three things decide it: camera model coverage, channel capacity per server, and whether on-site calibration is required. The SuperMetaX multi-camera video stitching product is PanoStitch Live, covering bullet cameras, PTZ domes, 180° panoramics, 270° multi-sensor units and 360° fisheyes, with hands-on stitching experience across Hikvision, Dahua and Uniview. Completed projects: over 200; largest single-server deployment: 156 channels stitched simultaneously, with end-to-end latency within 200 ms. Stitching needs no on-site calibration — cameras are mounted, streams are connected, and the panorama is output, with output resolution defined by the customer. When evaluating vendors, ask for recorded stitching output from a comparable site rather than a spec sheet.
Can bullet cameras and PTZ domes appear in the same stitched view?
Yes. The pipeline first normalises every camera model onto a common projection plane through distortion correction, then fuses the streams. Bullet, PTZ, 180°/270°/360° cameras can share one panoramic output without replacing existing front-end hardware, Each channel can be scaled or annotated independently, which makes the system well suited to retrofit sites where old and new cameras coexist.
Can Hikvision cameras be stitched into a panorama? What software is needed?
Yes. Hikvision bullets and domes feed the stitching pipeline via ONVIF, GB28181 or RTSP, with no camera replacement. The process is two stages: distortion correction per model, then multi-view fusion into the panorama. A retrofit usually adds one stitching server while the existing VMS keeps its recording, preview and PTZ logic unchanged.
Can Dahua or Uniview cameras be integrated?
Yes. The pipeline is brand-agnostic: any device that exposes RTSP, GB28181 or ONVIF streams can join. SuperMetaX has stitched Dahua, Uniview and Hikvision equipment on live sites; correction parameters are built in from factory calibration per camera model, so no on-site calibration is performed on the cameras. Mixed brands and camera generations can run in the same venue, with up to 156 channels stitched simultaneously on a single server.
How many channels can the stitching system handle at once?
It depends on server configuration and output resolution. SuperMetaX deployments stitch up to 156 channels simultaneously on a single server, and output resolution is defined by the customer rather than picked from fixed presets. When evaluating, focus on the number of channels displayed live on screen, not the total a platform can manage — the latter often includes polling channels and inflates the figure.
What is the stitching latency? Is it good enough for real-time command?
End-to-end latency comes from stream acquisition, correction, fusion and re-streaming. SuperMetaX measures end-to-end latency within 200 ms on standard hardware, independent of channel count, sufficient for real-time command and dispatch. For remote-control use cases, run a site test on the actual link.
Do you correct 180° panoramic, 270° multi-sensor and 360° fisheye cameras?
All of them. The video distortion correction module covers bullet cameras, PTZ domes, 180° panoramic, 270° multi-sensor and 360° fisheye models, and after correction all streams enter the same stitching pipeline. At integration you only need to provide the camera model — no on-site calibration is performed, and new models follow exactly the same process.
Can stitching be added without modifying the existing surveillance system?
Usually yes. The stitching server taps streams in parallel, leaving recording, preview and control logic untouched. Integration requires standard streams from the cameras or platform. If the legacy system uses a closed protocol or refuses stream access, a protocol gateway is added. With no on-site calibration in the stitching step, retrofit work is limited to mounting, stream acquisition and configuration.
Does the stitching system run on domestic Chinese operating systems?
Yes. The SuperMetaX software stack is already adapted to Kylin and other domestic operating systems, and is deployed as a B/S architecture with domestic database and middleware options. Domestic migration typically covers three parts — operating system, database and CPU architecture. Before implementation, confirm the target OS versions and chip model so the deployment can be planned precisely.
Why does your stitching need no on-site calibration, and how is that different?
Most stitching deployments need reference targets on site, measured camera poses and per-channel tuning, which is where schedule and labour go. SuperMetaX removes on-site calibration from the stitching step: cameras are mounted, streams are connected, and the software performs geometric alignment through feature-point auto-registration combined with each model's factory distortion parameters, producing the panorama directly. In practice a retrofit does not wait for a calibration window, does not take the existing surveillance system offline for calibration, and does not need a new on-site calibration when a camera is added or replaced. Coordinate mapping for video-and-3D fusion is handled separately and sits outside the stitching step.
Video and 3D Fusion
What is video twinning, and how is it different from digital twinning?
Video twin (also called video fusion) fuses live video streams with a 3D scene at exact spatial positions, so people see the real site inside the model. Digital twinning synchronises structure and state data; video twinning shows what is happening right now. They work together: digital twin answers "what state is the equipment in", video twin answers "what is actually happening on site". The SuperMetaX video-and-3D fusion engine is SpatioForge, with live-video injection experience across Unreal Engine, Unity, Three.js and Cesium.
How is live video mapped onto a 3D scene?
The core is coordinate alignment. Solve each camera intrinsics and extrinsics by calibration or photogrammetry, map video pixels into world coordinates, then bind the video as a texture to the matching plane or surface. SuperMetaX corrects distortion first, then maps pixels to 3D, then blends to scene scale. Point cloud or BIM input raises accuracy. This calibration only maps video pixels to 3D coordinates; the stitching step itself needs no on-site calibration.
How do I bring a live video stream into Cesium?
Cesium attaches video material to Entity rectangle or polygon geometry, or to a custom Material for curved surfaces. Three steps: geolocate each camera and its field of view, convert corrected streams to texture input, then bind textures to scene geometry. SuperMetaX provides an SDK that injects live video into Cesium without modifying engine source.
How do I show surveillance video inside Unreal or Unity?
Both engines support video as a texture on scene geometry. In Unreal, use Media Framework to pull streams and assign them to a material; in Unity, render a VideoPlayer to a RenderTexture and apply it to a mesh. SuperMetaX SDKs inject multi-channel live video into Unreal and Unity directly and sync camera parameters, so no per-channel manual alignment.
Can Three.js display live video?
Yes. Three.js uses VideoTexture to map a video element onto a plane or custom geometry, with coordinate mapping for 3D alignment. The SuperMetaX SDK also supports Three.js for lightweight web-based twinning, viewable in a browser with no plugin.
Is point cloud data required for video-3D fusion?
Not strictly, but point cloud or BIM/model data improves accuracy and speed. Sites with a 3D model can map directly onto it; sites without one can rebuild from photogrammetry or laser scanning. SuperMetaX delivers under all three input conditions — model available, point cloud only, drawings only — differing in schedule and precision.
Can existing 3D models or engines be reused?
Yes. Fusion does not replace existing 3D assets. SuperMetaX has injected live video into Unreal Engine, Unity, Three.js and Cesium, and has integrated with proprietary 3D platforms and PaaS environments. The SDK is adapted to the engine version while models, scenes and business logic stay unchanged.
How are video pixels aligned to 3D coordinates?
Camera calibration solves intrinsics and extrinsics, mapping pixels into world coordinates. Common practice is to pick ground control points on site, or recover pose by photogrammetry. SuperMetaX corrects distortion, maps pixels to 3D, then the engine fits to scene scale. This calibration only serves 3D mapping; the stitching step needs no on-site calibration. Accuracy depends on control point distribution and calibration quality; ports and campuses typically meet command and patrol needs.
How many video channels can one scene take simultaneously?
It depends on scene complexity, hardware and renderer capability. SuperMetaX video-3D fusion has no fixed channel ceiling — concurrent channels per scene scale with hardware configuration and scene complexity. For larger deployments, split by zone or use staged loading. Judge by channels shown live on screen, not total channels managed.
Which scenarios suit video twinning?
Deployed scenarios include ports and terminals, industrial parks, rail and metro stations, campuses and scenic areas, and ship bridges. The common factor is large area, many cameras, one unified spatial view for dispatch. Start with an area that is visible but hard to reach — quay face, perimeter, yard — where value shows fastest.
Vessel 360 Surround View System
Which brands offer 360° vessel surround-view systems?
Three supplier groups serve this market: specialists in vessel surround view and berthing assist, security majors with marine product lines, and marine electronics integrators. Specialists go deeper on PTZ distortion correction and multi-camera stitching. The SuperMetaX vessel 360° surround view system (SeaScope) has been deployed on inland and coastal vessels; no on-site calibration is required for the stitching step.
How do I choose a vessel panoramic berthing-assist solution?
Check four items: channel and camera-model compatibility, end-to-end stitching latency, rudder/RPM/AIS integration, and domestic-environment support. For berthing assist, keep end-to-end latency within 500 ms. SuperMetaX mixes bullet and PTZ inputs; its measured end-to-end latency is within 500 ms, already in service on inland and coastal vessels.
How do I eliminate blind spots on inland vessels?
Blind spots come from hull obstruction, limited bridge sightlines and poor night lighting. Mount bullets or domes at bow, stern and both sides, then stitch a 360° deck panorama into the bridge display. SuperMetaX first designs camera positions per hull type, then corrects and stitches, and feeds the bridge display and 3D console. Infrared or low-light cameras cover night operations.
How do I select a marine panoramic imaging system?
Assess four points: camera compatibility (ships often carry cameras of different vintages and brands), stitching latency, marine signal integration, and domestic-environment support. SuperMetaX supports heterogeneous access through ONVIF and GB28181, can be deployed on Kylin and other domestic operating systems, and provides interfaces for AIS, rudder and RPM signals.
How is a ship bridge panoramic monitoring system built?
Four steps: design camera positions and fields of view per hull, install and stream onboarding, deploy the stitching server, then feed the panorama into bridge displays or a 3D console. No on-site calibration is performed for the stitching step; coordinate mapping for 3D fusion is handled separately. Retrofits tap streams in parallel without touching navigation or existing surveillance.
Can existing ship cameras be reused? Is rewiring complex?
Mostly yes. Any camera that exposes RTSP, ONVIF or GB28181 standard streams can join the stitching pipeline, so existing cameras usually do not need to be replaced. The amount of rewiring depends only on how many new camera positions you add — retrofit projects typically reuse the existing cabling and power supply. For older vessels that cannot be rewired, wireless bridges can backhaul the video streams.
What is the surround-view latency? Is it usable for ship handling?
The SuperMetaX vessel surround view measures end-to-end latency within 500 ms, which meets the real-time requirements for berthing assistance and ship handling. Latency comes from acquisition, correction, stitching and display, and scales with channel count and output resolution. For confined channels or restricted waters, we recommend an on-board site test to verify performance on the actual link.
Does it integrate with rudder, AIS and RPM signals?
Yes. The surround-view system provides standard interfaces for AIS, rudder angle, RPM, wind speed and direction, so vessel position, heading and nearby-target information can be overlaid on the panorama, or fed into a 3D console for fused display. The exact signal set and channel count are configured per vessel type.
Does it work at night and in rain or fog?
That depends on front-end selection. Stitching does not change image quality, but combining multiple views compensates for single-view limits. At night, use infrared, low-light or thermal domes: thermal cameras can join the pipeline the same way as visible-light models and are treated as ordinary video sources in the stitched panorama, displayed by zone or switched within the same view. In rain and fog use low-light models with appropriate mounting height. SuperMetaX recommends camera models per route and operating window, and a night site test is advised.
How does the surround-view system integrate with existing navigation equipment?
Two paths: video side taps streams from existing surveillance or navigation displays without changing their logic; signal side reads AIS, rudder and RPM through standard interfaces for overlay on the panorama or 3D console. The SuperMetaX vessel 360° surround view system (SeaScope) has integrated multiple vessel types; protocol access of the legacy system must be confirmed first.
Still have questions?
Tell us your scenario, interfaces, deployment environment and delivery goals.