7.2 Advanced Scanning Features
Initialization, the two-phase capture method, control point marking, multi-battery projects, and Portrait mode. Each follows a specific sequence. Steps performed out of order produce incomplete data, failed fusions, or file corruption.
Preparation and Initialization
Where you initialize determines where the finished model opens. The reconstruction sets its initial viewpoint at the initialization location, so the first thing a client sees when they open the model is whatever the PortalCam was looking at when the scan began. Treat this as a composition decision, not a formality.
Set layered optimization before you scan a repetitive space. It is on by default and it is the right default for most scenes. Where the space has heavy repeating texture, the published guidance is to turn it off, with cubicles given as the example. Open-plan offices, data halls, racked warehouses and uniform corridors are all the same problem. This is set in the app before the scan starts and cannot be applied afterward.
Prepare the Site Before You Prepare the Device
Most PortalCam failures are lighting and movement failures rather than technique failures, and both are cheaper to fix before the scan than after.
- Open every door and passage on the route in advance. The point is to move continuously without interrupting the route once recording starts.
- Turn on all fixed lights. Where ambient brightness is low the camera extends its shutter time, which produces motion blur, image noise, and unreliable feature matching. Do not use flash: it makes the lighting differ between frames.
- Fix the state of doors, windows and curtains for the duration. Do not open or close them mid-capture.
- Remove or secure anything that can move: people, pets, curtains near vents. Turn off fans and secure lightweight items near air-conditioning outlets.
- Deal with glass and mirrors by covering them where that is possible, or by noting them so the route can cross rather than dwell.
- Check the floor for large reflective puddles or wet areas.
- Clean every lens. Fingerprints and dust on the fisheye or front cameras degrade texture quality across the entire model and cannot be corrected in processing. Use the included microfiber cloth. Never touch the optical surfaces directly.
- Set up on the tripod, on stable level ground. The device must be stationary and level through the full startup sequence. An uneven or soft surface during startup causes calibration errors that produce distorted data or an outright scan failure.
- Initialize in the main display area. Choose the space you want the model to open on: the lobby, the entry, the primary room. Capture detailed data around the initialization point so the opening view is clear and blur free from every angle.
- Point the front camera at a clean area. Avoid crowds, vehicles, and any high-movement zone. Moving objects in the initialization frame corrupt the opening view and can destabilize early tracking.
- Allow 25 seconds for startup. Do not move the device during this window.
- After pressing record, wait about 6 seconds for scanning to begin, then wait for the slow green flash before lifting the device. Picking the PortalCam up before the slow green flash appears starts the scan from an unstable state.
Plan the end of the scan at the same time you plan the start. Returning to the initialization area at the end of the segment gives the tracking algorithm a closing loop against geometry it already knows. An effective loop means returning to a previously scanned location with the device facing roughly the same direction, within about 40 degrees of the original line of sight. Returning to the same spot facing the opposite way does not count as a loop.
The Two-Phase Scanning Approach
PortalCam capture is two distinct phases inside one continuous segment: scene scanning for full environment coverage, then high-resolution scanning for local detail. Do not stop and restart between them.
The PortalCam is the slowest of the three devices, and by a wide margin. Its ceiling anywhere is 1.6 ft/s (0.5 m/s), which is the indoor speed for the L2 Pro and K2, not their ceiling. Indoors, in dim light, at doorways and around corners, slow to 1.0 ft/s (0.3 m/s). Applying an L2 Pro pace to a PortalCam job produces degraded data with no warning at the time.
Phase 1: Scene Scanning
The goal is complete spatial coverage of the environment. Aim for 60 to 80 percent overlap between frames throughout: that is what the slow pace buys, and it is the figure the reconstruction actually depends on.
The Three-Pass Route Structure
XGRIDS publishes a three-pass route for indoor capture. Alpine runs it as the structure of the scene phase, with the detail phase in Section 2 following it.
Pass 1: Wall Perimeter Loop
Walk slowly along the walls with the lens pointed across the room toward the far side. This records the overall structure and gives stable side-view overlap between frames.
- Stay 1.6 to 3.3 ft (0.5 to 1 m) from the wall
- Walk at 1.0 to 1.6 ft/s (0.3 to 0.5 m/s)
- Move smoothly, avoiding shake
- In a multi-room space, loop each room and pause longer at doorways
Pass 2: Grid Traversal
Move into the centre of the room and cross it on a grid. This captures objects in the middle of the space and the depth relationships between them.
- Walk 2 to 3 lines in each direction to form the grid
- Keep the lens at eye level, facing your direction of travel
- This is what rescues weak-texture areas such as white walls and open floors, by adding multi-angle constraint
Pass 3: Ceiling and Floor
Walk the same routes again with the lens angled up, then angled down. This is the pass people skip and it is the one that fixes ceiling shapes and floor detail.
- Upward recovers the ceiling, light fixtures and air vents
- Downward recovers floor material, baseboards and the underside of furniture
- You can also alternate forward, up and down every few steps along the same route
- Multi-height trajectories. Scan at low, chest, and overhead positions. A single-height pass captures a fraction of the surface area the reconstruction needs. Change height between passes rather than within one pass, and keep each change small enough that your line of sight stays within about 40 degrees of the previous pass.
- Bidirectional coverage. Walk each major corridor or route in both directions. Forward and reverse passes on the same trajectory give the cameras different angles on the same surfaces and produce noticeably better texture.
- Serpentine routes by default. Use an S-shaped path rather than a straight line everywhere except extremely narrow corridors. Straight single-line passes through open space leave gaps between adjacent trajectories.
- Front camera faces the direction of travel. The front cameras capture the highest resolution imagery. Surfaces seen only by the fisheye cameras carry less texture detail in the finished model.
- Work at about an arm's length from surfaces. The hard floor is 1.6 ft (0.5 m) from walls and from any subject. Closer than that and the cameras cannot match features between frames. Avoid extremely narrow spaces where the clearance cannot be held.
- Back off to about 3.3 ft (1 m) from highly reflective surfaces. Glass, polished metal, and mirrors return unreliable geometry at close range. Add passes from additional angles rather than moving closer.
- Keep the device between negative 60 and positive 60 degrees of tilt. Never point it straight up or straight down.
- Turn gently. Avoid rapid rotation, shaking, and swinging, especially in low light. Use a wide turning radius.
- Keep people out of the scan path. Nobody should follow the operator closely or for an extended stretch.
- Stay back from dynamic elements. Digital and video screens, foliage moving in wind, traffic, and pedestrian flow all change between frames and corrupt the data in that area.
Phase 2: High-Resolution Scanning
After the scene passes, revisit the areas that need detail. Aim the front camera directly at the target and move slowly for close-range, multi-angle coverage. This phase is what makes signage, artwork, product displays, and equipment labels legible in the finished model.
- Keep the front camera aimed at the subject. The fisheye cameras do not resolve enough detail for this work.
- Standing still does not help. Unlike a terrestrial LiDAR scanner, the PortalCam gains nothing from a static hold. Detail comes from movement across the subject, not from dwell time.
- For objects and sculptures, walk a full circle around the subject to capture it from every angle.
- For text, nameplates, and fine labels, bring the front lens to about 1.6 ft (0.5 m) and move in a slow S-pattern across the text.
- For pipes and MEP runs, capture from multiple positions along the run. For ceiling services, use a pitch angle to reach them. Do not point the camera straight up, which exceeds the tilt limit and distorts the data in that section.
- Watch the live camera feed in LCC Scan to confirm image clarity while you capture.
When there is no single focus object, saturate the scene. Multiple routes, multiple heights, multiple angles across the same space produces the most reliable result. Plan the time for it. A 15-minute scene pass often needs another 5 to 10 minutes of detail work on top.
Control Point Marking
Control points are in-scan markers that anchor separate segments to a shared spatial reference. They are what makes Map Fusion, Aerial-Ground Map Fusion, and HD Enhancement possible on the PortalCam. Because the PortalCam does not output absolute georeferenced coordinates, control points are the only way to connect one segment to another. There is no RTK fallback.
The PortalCam procedure differs from the L2 Pro and K2 in one respect that matters: the device goes on the tripod and is set down at the point. It is not marked handheld.
Control point marking requires a current LCC Scan app and current device firmware. On an older build the control point button will not behave as documented. Check the firmware version in the LCC Scan device status panel before any project that depends on control points, and do it before you travel, not after you arrive.
Marking Procedure
- Choose the location. Good ambient light, clear surface texture, and open unobstructed space around it. Avoid corners, narrow passages, and harsh light falling directly on the device cameras. The cameras need clear sight of the surrounding geometry to record usable orientation data.
- Deploy the tripod and set the device on it. The tripod must be stable and not rocking. Avoid soft ground and uneven surfaces where it could shift while you mark.
- Tap the control point button and complete the dialog. Select the type, then enter the name. The name is how the software matches the same physical point across segments, and it is case-sensitive. A point named Map1A in one segment and map1a in another will not be recognized as the same point.
- Tap Add and wait 5 seconds. The device needs those 5 seconds to finish writing the control point to the project file. Powering off or pulling the battery inside that window loses the point. The scan data survives, but the control point never reaches LCC Studio and any fusion that depends on it fails.
- Walk one full circle around the control point. Do not skip this. Circling the point gives the fusion algorithm 360 degree coverage of the surrounding geometry and measurably improves its stability. A point marked without the circle is a weaker anchor.
- Photograph the location with your phone. The photo is how you find the identical position when you start the segment that has to share this point. Placement tolerance is generous, but the photo removes the ambiguity entirely.
Control Point Types
Select the type that matches the job. The type determines how the point is used at reconstruction.
Shared points do not need to be replaced exactly. When the same control point appears in two segments, it needs to be within about 1.6 ft (0.5 m) horizontally, within 20 degrees of the same orientation, and within about 4 in (10 cm) in height. Approximate consistency is enough. Gross inconsistency between sessions degrades the fusion.
Multi-Battery Projects
The PortalCam runs about 60 minutes on a charge and has no pause function. A project that outlasts one battery is a multi-segment project, and the segments are joined afterward in LCC Studio using Map Fusion. Plan the segment boundaries before you start scanning, not when the battery warning sounds.
What Actually Limits Segment Length
Three separate limits apply, and the one that binds is almost never the software ceiling.
Full collection strategy is in 7.4 Map Fusion Collection.
Battery cutoffs decide the segment boundary if you do not. The device sounds an audible alert at 20 percent and again at 10 percent. Below 10 percent during recording it force-stops and saves on its own, ending the segment wherever you happen to be standing. Below 5 percent it shuts down. A battery holding less than 10 percent will not power the device on at all, so confirm your spares are charged before leaving.
Segment Changeover Procedure
- Reach a good handoff location above 20 percent. A corridor intersection or a furnished room gives the next segment strong geometry to initialize against. A blank wall at the end of a hallway does not. Navigate there while you still have margin.
- Capture the overlap before you stop. The next segment has to re-cover the last stretch of this one. Scan through the planned overlap zone, then stop inside it. Aim for about 50 ft (15 m) of shared geometry, and avoid dropping below 33 ft (10 m) or letting one segment sit almost entirely inside another.
- Mark the shared control points. If the segments will be joined by Map Fusion, place the shared points now, following the marking procedure in Section 3. Record the names exactly as you will type them in the next segment.
- Hold the stop button and confirm. The device begins writing the project file immediately.
- Wait out the fast green flash before touching anything. Fast green means saving. Removing the battery or powering off during it corrupts the project file, and the data is not recoverable. Large scans can flash for several minutes. Wait.
- Swap the battery and start the next segment. Set the device down, change the battery, and initialize the new segment following Section 1. Re-cover the overlap zone first, then advance into new territory.
Finish the whole project within about 2 hours. Segments that will be fused need consistent lighting. Capturing across a major lighting change, morning into evening or sunny into overcast, degrades the fusion at the seams. When a longer window is unavoidable, place the overlap zones in well-lit areas with clear texture.
Portrait Mode
PORTRAIT is one of the four scan modes on the PortalCam, alongside NIGHT, INDOOR, and NATURE. It exists for reconstructing a person rather than a space, and it is the one mode where the subject can move and ruin the capture. Select it in the LCC Scan capture interface before scanning starts; it is not a processing option and cannot be applied afterward.
For an AEC, data centre, or arena workload this is a specialist mode. It earns its place when a client wants a person placed inside a captured environment, for a marketing piece, a training asset, or a digital twin that includes staff.
- Select PORTRAIT before you begin. Slide to the mode in the capture interface. The other three modes are tuned for environments and will not reconstruct a person well.
- Set uniform lighting. Even, soft light across the subject. Avoid harsh direct light, which blows out skin detail, and avoid bright backgrounds, which make the subject silhouette rather than resolve.
- Keep the front camera off the subject during initialization. Point it at the surrounding scene while the device starts up. This gives the subject time to settle into position rather than holding a pose through the startup sequence.
- Wait for the ready prompt before the subject holds still. Asking someone to freeze 30 seconds early guarantees they move at the moment it matters.
Capture Procedure
- Circle the subject with the front camera aimed at them throughout. The front cameras carry the detail. Keep them on the subject from every angle of the circle rather than letting the subject drift into fisheye-only coverage.
- Work top down: head and face first, then the body, then the lower extremities. The face is what a viewer looks at, and it is captured while the subject is freshest and stillest.
- Do not dwell on the face. Extended time on one region invites the small movements that produce motion blur. Move steadily across it and come back on a later pass if you need more coverage.
- Target about 2 minutes for the whole capture. That figure is a balance between reconstruction quality and how long a person can genuinely hold still. Past it, subject fatigue costs you more than the extra coverage gains.
- Verify critical regions in LCC Scan before you release the subject. Check the live views while they are still standing there. Re-staging a person is far more expensive than re-scanning a wall.
The subject must stay as still as possible. Changes in posture or expression during capture produce reconstruction errors that no processing setting corrects, because the algorithm receives two different shapes for the same surface. Brief the person before you start: still stance, relaxed but fixed expression, no shifting weight, no talking.
Multiple subjects raise complexity sharply. Two or more people moving independently multiply the chances that one of them shifts during the capture. Scan people individually where the deliverable allows it. If a group shot is required, brief everyone together, keep the total capture short, and expect to repeat it.
©2026 Alpine Reality Capture LLC • XGRIDS Pro Guide™ • Site Disclaimer