800-980-4649
PortalCam.
Walk through a space. Walk out with a photorealistic, navigable 3D world. This is everything our team has learned putting XGRIDS' handheld 3D Gaussian Splatting platform to work in the field — the parts the manufacturer's documentation doesn't quite cover.
SPATIAL CAPTURE UNIT XGRIDS // PORTALCAM
The PortalCam is a handheld device that captures full 3D environments by walking through them. Not panoramas. Not point clouds. Photorealistic, navigable 3D Gaussian Splatting models with the depth, lighting, and material fidelity of cinema-grade renders — produced from a sub-900-gram device that fits in your hand.
XGRIDS built the hardware, the capture app (LCC Scan), and the desktop processing software (LCC Studio). The pieces are excellent. The documentation explaining how to get the most out of them is… less excellent. This page closes that gap.
Every recommendation below comes from field experience, community knowledge, and time-on-tool. If you're evaluating the PortalCam, deciding between Basic and Premium licensing, or just trying to get a clean scan of a job site you're already on — start here.
From walk — to deliverable.
The PortalCam workflow is intentionally compressed. Four stages, mostly automated, designed so a single operator can capture, process, and deliver a photorealistic 3D model without leaving the project.
Capture
Walk through the space at a natural pace. The LCC Scan app on your phone shows a live quality indicator — green / yellow / red — so you know mid-scan if you need to adjust.
Process
Transfer the scan to LCC Studio on a workstation, or push it to the LCC cloud. The software handles alignment, exposure balancing, and the splat reconstruction.
Review & Edit
Trim, clean, colour-balance, annotate. Add fly-through paths. Drop measurements. Set the entry viewpoint your client will see when they open the share link.
Share & Export
Publish a web link, drop into VR/AR, or export to Unity, Unreal, Cesium, .PLY, USDZ, or 3D Tiles for downstream pipelines.
Four modes. Pick the right one.
The PortalCam has four Scene Modes that adjust exposure, ISO behaviour, IR-cut filtering, and LiDAR range usage to match the environment. The wrong mode won't fail the scan — it'll just hand you noisy textures or blown highlights you can't recover.
Indoor Mode
The default for typical interior environments. Camera settings are tuned for mixed artificial and indirect daylight; the SLAM/vision pipeline expects enclosed spaces and can handle low-texture surfaces like plain walls and uniform flooring.
Works well in furnished rooms, classrooms, lobbies, and any setting where lighting is decent but not extreme. Performs reliably in shaded interior areas — it accounts for these conditions in its SLAM workflow.
If a room is completely unlit (basement, abandoned space), drop to Night Mode. If you're transitioning between bright daylit room and dark basement, record them as separate scans.
Nature Mode
For building exteriors, streets, construction sites, gardens, and open landscapes under daylight. Uses faster shutter speeds, lower ISO, and the LiDAR's full ~60 m range to capture distant geometry.
Designed to handle high dynamic range — bright sky against shaded ground — via adjusted exposure and HDR behaviour. Overcast is ideal; full sun works with awareness of glare and flare. Avoid pointing directly at the sun.
In wide-open areas without nearby features, slow your walking pace. Distant objects get fewer LiDAR returns. Walking a slight zigzag through plazas or along long roads gives SLAM more parallax to lock onto.
Night Mode
Increases camera sensitivity and removes IR-cut filtering so the sensors gather infrared in addition to visible light. LiDAR doesn't need ambient light to work, so geometry comes in fine — Night Mode exists to let the cameras contribute usable colour information when there's almost nothing to see.
Useful for unlit basements, abandoned interiors, nighttime exteriors lit only by streetlights or moonlight. Expect slightly desaturated or shifted colour tones because of the IR contribution and high ISO.
Don't aim at bright light sources — high ISO causes severe flare. The PortalCam's own status LED can reflect in mirrors and dark interiors; many operators tape it over during night scans. Consider a Bushman Halo 360 ring light (one or two) for cleaner texture.
Portrait Mode
Tuned for fine geometry and texture fidelity on close-range, mostly-stationary subjects. Gathers more frames per unit area and uses specialised depth processing to preserve features like facial structure or sculptural surface detail.
Less tolerant of subject motion than the other modes. A stable, photograph-like pose held for 30–60 seconds is the working assumption. Best at 0.5–1.5 m from the subject with a smooth single orbit in 30–60 seconds.
Portrait Mode produces denser datasets and longer processing times. Use it when the detail is worth it. For larger objects or outdoor sculpture under unpredictable light, Nature Mode is often the better choice.
How to actually use the thing.
Capture techniques for the environments operators actually work in — informed by community knowledge, field testing, and the practical limitations the spec sheet doesn't mention.
Indoor Scanning
Indoor environments — homes, offices, museums — are the bread-and-butter use case for the PortalCam. They also bring constraints: tight rooms, mixed lighting, repetitive geometry like long blank corridors.
Pick the Right Mode
Indoor Mode is the starting point for any room with some lighting. Reach for Night Mode only when a room is genuinely dark — an unlit basement, an unpowered building. If you have to scan a bright lobby and an unlit storage room in the same session, record them as separate scans. The wrong mode won't fail the scan, but it will hand you noise you can't fix in post.
Prepare the Lighting
Turn on every available indoor light. Open blinds during daytime — though watch for bright sunbeams creating extreme contrast. More uniform lighting always produces cleaner texture. In high-contrast rooms (bright window, dark corner), slow down in the darker sections so the cameras have time to gather data.
Coverage Pattern
A repeatable pattern that works in most rooms:
- Walk the perimeter first at roughly 1–2 m from the walls.
- Pass through the centre in a zigzag or straight sweep.
- Briefly aim into corners and through doorways — these are where data gaps appear.
- In rooms with low-texture walls (empty white rooms), add a gentle side-to-side weave. The lateral motion introduces parallax that helps SLAM stay locked.
The PortalCam records continuously. Don't stop in doorways — abrupt motion changes through tight transitions are where alignment drift starts. Smooth motion, every time.
Mirrors, Glass, and Polished Floors
Indoor environments are full of reflective surfaces. LiDAR + camera fusion handles a lot of this, but careful technique helps. For shiny floors, vary your angle slightly as you move — that breaks up persistent ceiling-light reflections. For mirrors:
Do
- Capture from multiple angles
- Approach mirrored walls from the side
- Wear dull, non-reflective clothing
- Use Indoor Mode — LiDAR handles geometry
Don't
- Stand directly between mirror and scene
- Wear bright or glossy clothing
- Aim LiDAR head-on into glass
- Linger in front of a mirror mid-scan
Multi-Room Properties
For large interior areas or multi-story buildings, segment the project. Scan each floor or each major zone as its own session, ideally 10–20 minutes each. This keeps drift manageable and processing reliable — and it gives you the option to use Map Fusion to stitch the scans together later (see 4.7).
We've seen operators try to scan an entire two-story house in one continuous session because the battery technically allows it. Don't. Even when the reconstruction succeeds, processing times spiral and small alignment errors compound across stair transitions. Break the job into rooms, floors, and connecting corridors — then fuse.
Outdoor Scanning
Outdoor scans — building exteriors, gardens, streets, sites, landscapes — let you capture fine detail and large-scale context together. The variables shift outdoors: strong directional light, sparse nearby surfaces, and moving objects in the scene.
Default to Nature Mode
Nature Mode is the daytime outdoor mode. It expects longer viewing distances, uses faster shutter and lower ISO for bright conditions, and unlocks the LiDAR's full ~60 m range. Move to Night Mode for nighttime outdoor work.
Don't aim the cameras at the sun — HDR handling helps, but flare and washout still happen. If the sun is right in front of you, walk a few alternate angles. Overcast is the easiest lighting condition; full sun is workable with care.
Walk Slower in Open Areas
Wide outdoor spaces — parking lots, fields, long driveways — have a structural problem: nothing near the camera to triangulate against. Distant objects get sparse coverage. Slowing down gives the system more frames per metre and more chances to register far geometry.
- Open plaza: weave or loop instead of walking straight through the centre.
- Long road or trail: a gentle zigzag adds reference geometry.
- Approach distant features briefly to anchor them, then continue.
Loop Closure
The strongest tool you have against drift outdoors is loop closure — ending the scan near where it started. When SLAM recognises it's back at a known location, it can correct accumulated alignment error across the whole path. Walk a full circuit around a building. Overlap the start of the scan at the end.
On long linear routes where a true loop isn't possible, weave around distinctive features — circle a group of trees, walk an S-curve, briefly approach a building. Each gives the system something to re-localise against.
Moving Objects
Cars, people, animals, swaying vegetation — anything moving in your scene shows up as semi-transparent "ghosting" in 3DGS reconstruction, because the format assumes a mostly static world. Mitigation:
- Scan when foot and vehicle traffic is low.
- Let a passing car or person clear the frame before continuing.
- Don't let the same moving subject occupy a large portion of the scan.
- If a truck obstructs a facade mid-scan, do a short follow-up pass after it leaves.
Weather & Battery
The PortalCam is rated for low-temperature operation down to roughly −20°C, but cold reduces battery performance — keep spares in a pocket. The device isn't waterproof. Light mist with a clear cover is workable; heavy rain, snow, or fog — reschedule.
A monopod or pole extends the PortalCam's effective field of view dramatically — raising it over your head captures upper facades and rooflines that ground-level scanning misses entirely. It also dampens footstep micro-motion. Not in the standard kit; worth adding if exterior work is a regular part of what you do.
Scanning Detailed Objects
When the subject is an object — a sculpture, a vehicle, a piece of equipment — the goal shifts to maximising geometric and textural detail. The PortalCam handles mid-to-large objects (roughly 0.5 m and up) well. Very small subjects (figurines, jewellery) fall outside its strengths because of the wide-angle lenses and a ~10 cm LiDAR minimum range.
Portrait Mode for Detail
Portrait Mode is the right starting point for any stationary, close-range object. It densifies the dataset and uses depth-processing settings tuned for fine surface detail. Trade-offs: longer processing times, larger files. Worth it when the detail matters.
Optimal Distance
Stay between 0.5 m and 2 m of the subject. Closer than 50 cm and you may exceed the cameras' ability to focus. Farther than 3 m and you're sacrificing pixels-on-target.
A combined approach works best: start around 1.5 m to capture overall structure, then gradually move closer to 0.5–1 m for finer detail, while maintaining context. Spiral or gradually tightening circles do this naturally.
Multi-Orbit Coverage
A single horizontal orbit will miss undercuts, top surfaces, and elevation-dependent details. Plan two or three passes:
- One orbit at mid-height for overall coverage.
- A second orbit higher or lower to catch what the first missed.
- Optionally vary radius — one closer pass for detail, one wider for context.
Shiny & Glossy Materials
Polished metal and stone produce specular highlights. Scan in diffuse lighting (shade, overcast, indirect) when possible. Avoid direct midday sun. Capture from multiple angles so the highlights shift relative to viewpoint — this gives the reconstruction more information to work with.
Isolate When Possible
The PortalCam captures 360° of surrounding context. If the object itself is the deliverable, minimise background clutter — angle the camera slightly up while orbiting a statue to reduce ground noise, position yourself so backgrounds are uniform (sky, wall, ground), and ensure the area around the object's base is cleanly captured so you can crop in editing.
For objects with deep recesses or internal cavities — a car's interior, the underside of a chassis — plan a second targeted scan for the hidden geometry. One pass cannot see what one pass cannot see.
Portraits & People
Scanning a live human is its own discipline. People sway, breathe, and adjust. The target is roughly two minutes of total capture — long enough for good data, short enough that subject fatigue doesn't take over.
Pose & Stillness
A natural, photograph-like pose that can be held comfortably for 30–60 seconds. Sitting helps. A subtle support (out of frame) helps. Natural blinking is fine. Weight shifts and hand movements produce double contours in the reconstruction.
Orbit Plan
One complete 360° loop in 30–60 seconds is the working pattern. Keep the subject centred, smooth pace, steady distance.
- Begin with the head and face — the highest-stakes data.
- Transition smoothly to the torso and lower body.
- One well-executed loop usually beats two rushed ones.
- If a second loop is needed, do it at a slightly different elevation.
Distance & Framing
Portrait Mode at 0.5–1.5 m. Step back occasionally for full-body context, especially for taller subjects. The fisheye coverage is wide enough that you don't need to point straight at the feet — just keep them in frame.
What Won't Work Perfectly
Individual strands of hair. Sheer or fluttery fabrics. Anything thinner than a few millimetres at the capture distance. Group portraits where multiple people may move relative to each other — scan individuals separately and composite if needed.
Preventing Drift (SLAM Stability)
Drift — gradual positioning error that accumulates during a scan — affects every SLAM-based device to some degree. Good technique substantially reduces it. Nothing eliminates it.
Movement With Parallax
SLAM relies on translational motion, not rotation. Forward motion gives the algorithm depth information; spinning in place gives it almost nothing. When you change direction, walk a small arc rather than pivoting. In uniform spaces like long hallways, a slight zigzag adds the lateral variation SLAM needs to stay stable.
Stationary Start and Stop
Begin the scan with the device held still for a few seconds. End it the same way. This gives the system clean reference frames for initial pose estimation and final alignment.
Don't Repeat the Same Path
Multiple identical loops add nothing — they don't introduce new viewpoints, just let small differences accumulate. If you need to cover an area more than once, vary the trajectory: one loop close to the subject, one loop farther out, one at a different height.
Limit Indoor / Outdoor Transitions
Doorways between bright outdoor scenes and dim interiors are abrupt environmental shifts that strain tracking. Cross them once if you must. Don't bounce back and forth within a single continuous scan — record interior and exterior as separate scans and fuse them.
Loop Closure
The most powerful drift-correction technique. End the scan near where it began. When SLAM recognises a previously-seen location, it corrects accumulated alignment error backward across the entire path. Always loop when the site geometry allows it.
Featureless environments — bare hallways, blank walls, open snow — are the SLAM killers. If you have to scan one, lean in with technique: slight zigzag, slower pace, briefly approach any anchor feature you can find (doors, curbs, lamp posts, pavement cracks). Every distinctive feature you put in front of the system is a chance for it to correct itself.
Sharing & Export
The PortalCam ecosystem is built around moving the output downstream — web viewers, VR/AR, game engines, GIS platforms, traditional 3D pipelines. Each path has different export formats and constraints.
LCC Web Viewer
The simplest path. LCC Studio publishes a shareable URL that opens the scene in any browser, desktop or mobile, with WebXR VR on supported devices. Three navigation modes are available:
- Flythrough Mode — first-person navigation, best for interior exploration.
- Pivot Mode — rotate around a fixed point; good for inspection.
- Avatar Mode — character walkthrough for immersive presentation (no measurement tools).
For large outdoor or campus-scale scans, the default loading viewpoint is often unhelpfully far. Set the starting view before sharing — first impression matters.
Export Formats
- .LCC — XGRIDS' proprietary format, up to ~90% smaller than standard .PLY. Now supported by third-party viewers like SuperSplat.
- .PLY — standard 3DGS point cloud. Works in CloudCompare, Blender, ReCap, Revit (with plugins), and most splat-aware viewers.
- USDZ — for AR on iOS. Single-Scene only, requires firmware 3.0+, cross-platform portability must be disabled.
- 3D Tiles — OGC 1.1, for WebGIS and digital twins. Supports up to ~4M splats per model. Requires RTK-enabled scan for georeferenced coordinates.
- Unity / Unreal SDKs — Premium tier; for interactive applications and VR.
Third-Party Viewers
SuperSplat (PlayCanvas-based) and Voluma host PortalCam outputs and add capabilities XGRIDS doesn't — URL-parameter-driven presentation options, custom camera controls, and the ability to layer additional 3D content into the scene.
The Mesh Question
LCC Studio doesn't natively export OBJ or other mesh formats. If your downstream pipeline needs a traditional mesh (CAD, BIM, mesh-based renderers), the path is .PLY out of LCC, then mesh in CloudCompare, Blender, or similar. It adds a step. XGRIDS' public roadmap references additional point-cloud and mesh tools — this is a known area of community feedback.
Map Fusion
Map Fusion combines multiple PortalCam capture sessions into a single unified 3DGS model. It's a project-scaling feature — built for environments too large to record continuously — not a precision-correction tool. Fusion does not convert handheld capture into survey-grade data, and it does not remove drift within individual scans.
What It Needs
Successful fusion depends on capture discipline, set up before you arrive on site:
- A sketch or annotated plan showing segment boundaries, routes, and control point locations.
- Reasonable scan durations — 10–20 minutes per segment.
- Consistent Scene Mode across overlapping segments.
- Stable lighting throughout the capture window (a couple of hours, no major transitions).
- Deliberate spatial overlap between sessions.
Premium Tier Only
Map Fusion lives in LCC Studio Premium. Aerial-Ground Fusion (combining PortalCam ground capture with drone photogrammetry) is also Premium-only. If multi-scan projects are central to what you do, this drives the licensing decision.
Hardware Demands
Fusion workflows place much higher demands on the processing machine than single-scene reconstruction. An NVIDIA CUDA-capable GPU is required. Minimum 64 GB RAM; 96–128 GB preferred for larger projects or any project where total scan duration approaches 150+ minutes.
Processing time runs 30–40× total capture duration. Three 20-minute segments fusing together can take 30–40 hours on capable hardware. Plan accordingly: dedicated machine, uninterrupted runtime, no competing tasks.
Start small. Your first fusion project should not be the seven-segment multi-building campus you really need to deliver. Build confidence with a two- or three-segment job first. Get a feel for the overlap, the control point workflow, and the processing time on your hardware. Then scale up.
Basic or Premium?
Every PortalCam kit ships with twelve months of LCC Studio — Basic with the Standard Kit, Premium with the Premium Kit. The hardware is identical. The difference is software capability, and it matters more than most people expect.
- 3DGS Processing Cloud or local processing of single-scene captures into Gaussian Splat models.
- Basic Editing Cleanup, trimming, colour adjustment, annotations.
- Viewing & Sharing Web links, cloud viewer, .LCC and .PLY exports, basic mesh options.
- Navigation Tools First-person, orbit, measurements, annotations, fly-through path recording.
- Cloud Access Mobile app integration and cloud project storage.
Right fit: single-scan projects, real estate, smaller interior spaces, individual operators.
- Map Fusion Combine multiple scans into one unified model. Essential for large or multi-day projects.
- Aerial-Ground Fusion Merge PortalCam ground capture with drone photogrammetry. Roof + facade + interior in one model.
- Spatial Recognition Generates BIM-oriented metadata (walls, doors, windows) for downstream architectural workflows.
- HD Enhancement Increases splat density, improves clarity and texture refinement.
- Cluster Mode & Priority Cloud Multi-GPU and cloud-node processing for very large environments or tight deadlines.
- Unity & Unreal SDKs Integrate scans directly into game engines for interactive or VR applications.
- Collaboration Tools Extended cloud storage, team sharing, project management.
Right fit: multi-segment sites, BIM-adjacent work, drone-integrated capture, production pipelines.
The hardware keeps working forever. Processing new scans after the included 12 months requires an active LCC licence — subscription renewal or an enterprise perpetual on-premises licence. Previously processed models stay accessible. Plan ongoing software cost into total cost of ownership, especially if a Premium-only feature like Map Fusion is central to your workflow.
What it's great at. What it isn't.
No piece of equipment is right for every job. We sell the PortalCam because we believe in what it does — and we'll tell you when it's the wrong tool.
Strengths
- Walk-Through Speed ~280 m² scanned in 15 minutes with one operator. No tripod, no stations, no setup.
- Photorealistic Output Cinema-grade visual fidelity. Lighting, materials, and textures look real because they are.
- Difficult Environments Dim interiors, weak-texture rooms, reflective surfaces — 3DGS handles what photogrammetry struggles with.
- Compact Files .LCC format is up to 90% smaller than equivalent .PLY. Web-streamable. Shareable by link.
- Pipeline-Ready Exports to Unity, Unreal, Cesium, USDZ, 3D Tiles. Drops into existing 3D workflows.
- One-Handed Capture Under 900 g. Operable with one hand. Real estate agents to film crews can use it without special training.
Limitations
- Survey-Grade Precision Geometric accuracy varies by several centimetres over larger areas. Not appropriate for fabrication-ready as-builts or engineering-grade measurement.
- Macro / Small Object Capture 10 cm LiDAR minimum range plus wide-angle lenses mean figurines, jewellery, and fine-detail small objects are outside its envelope.
- Static Scenes Only Anything moving during capture — cars, people, swaying vegetation — appears as ghosting artifacts. 3DGS assumes a static world.
- Processing Hardware 64 GB RAM minimum for 30-minute scans; 128 GB recommended for 60 minutes. CUDA-capable GPU required. Map Fusion processing runs 30–40× capture time.
- No Native Mesh Export LCC Studio outputs splats and point clouds, not meshes. CAD/BIM pipelines that need meshes have an extra conversion step.
- Ongoing Software Cost LCC subscription must remain active to process new scans after the included 12 months.
An authorized XGRIDS dealer that actually uses the gear.
Anatum GeoMobile Solutions has been a national dealer of engineering-grade geospatial equipment since 2014. We're an authorized XGRIDS dealer, an authorized dealer for Eos Positioning, Laser Tech, and CHC Navigation, and an Esri Business Partner. We aren't a box-pusher.
Hands-On Field Knowledge
Everything on this page came from real capture sessions and community knowledge. When you call AGS with a workflow question, you're talking to someone who has actually held the device in the environment you're working in.
Licensing & Workflow Guidance
Basic or Premium? Cloud or local processing? How much RAM do you really need? The answer depends on what you're trying to deliver, and we'll walk through the decision with you instead of upselling by default.
Ecosystem Integration
We sell the GNSS, LiDAR, drones, and ground stations that pair with the PortalCam. If your project crosses into RTK positioning, drone capture, or larger LiDAR scanning, we can fit the whole stack together.
Honest Recommendations
We told you above what the PortalCam isn't good at. We'll do the same on a sales call. If your project needs survey-grade precision, you need a different tool — and we likely sell that tool too.
Post-Sale Support
Setup, firmware updates, processing-machine guidance, scan critique. The PortalCam isn't complicated, but getting a great result on day one is faster when someone who's been there walks you through it.
Direct Access
AGS is owner-operated. You're not routed through three layers of CRM to get a straight answer about a product or an order. You're talking to the people running the business.
Ready to see the PortalCam in action?
Email us about pricing, kit options, demos, or workflow questions specific to your project. We'll respond personally — no form funnels, no scripted decks.
[email protected] · agsgis.com · Authorized XGRIDS Dealer
