Frequently Asked Questions

Automatic projector calibration uses one or more cameras to measure a projected display and compute the geometric and photometric corrections that make multiple projectors behave as one seamless image. Cameras capture structured patterns from each projector, software solves how projector pixels land on the screen, and the correction is applied during playback. It replaces manual alignment. On a large or curved display, manual warping and edge blending takes a highly skilled technician days or weeks, and the result degrades as soon as anything in the room moves. A camera-based calibration is algorithmic, consistent, and repeatable, which means the same system can be returned to specification in minutes rather than rebuilt by hand. Scalable Display Technologies invented camera-based projector calibration. The core technology was developed at MIT by Dr. Rajeev Surati and his thesis advisor Dr. Tom Knight, Jr. in the 90s, and the company has focused on it exclusively since 2004.

Warp is geometric correction: reshaping each projector's image so it lands correctly on a curved, domed, or irregular surface and registers pixel-to-pixel with its neighbors. Blend is photometric correction: shading the overlapping regions between projectors so the seams disappear. Any multi-projector display needs both. Warp and blend is the baseline, not the whole job. A complete calibration also handles color matching between projectors, brightness uniformity across the screen, black level correction in overlap regions, image masking at screen edges, and 3D frustum calculation for perspective rendering. Scalable Display Manager produces all of these in a single pass.

Calibration runs in four stages: data collection, screen alignment, warp generation, and warp execution. Data collection: each projector displays a sequence of patterns — a black image for background subtraction, a full white image spanning the projector's output, a 5-spot image that establishes orientation, and a full spot grid. The cameras capture each pattern. For structural 3D mapping, a more traditional structured light pattern is used. Screen alignment: control points are known physical locations on the screen surface, typically along the top and bottom edges. Each camera must observe a minimum of six. They tell the software where the real screen sits inside each camera image. Control points can be placed in the camera image, the projector image, or physically with an LED or laser point. Warp generation: data from all cameras is combined in a process called camera blending, producing a map from projector space to texture space. That map becomes the basis for creating the Scalable Mesh File. Warp execution: the correction is applied at playback through the Scalable SDK, the graphics driver, a warping box, or the projector's own firmware. Execution is independent of how many cameras were used or how complex the screen is.

A Scalable Mesh File is the output of a calibration. It contains the geometric warp, edge blend, color correction, and intensity correction for a single projector. It also contains eye-point information if you are using a 3D playback engine. Playback systems read the file at runtime and apply the correction. Scalable also exports MPCDI for compatibility with third-party playback systems. The file is the reason Scalable stays hardware-independent. Because the correction lives in a portable file rather than in a piece of hardware, the same calibration can drive an SDK-integrated image generator, an NVIDIA graphics driver, a warping box, or a projector with onboard warp hardware. Change the playback system and the calibration comes with you.

Projectors move. Thermal cycling, building settling, vibration, optics adjustments, and light source replacement all shift the image by fractions of a degree, which is enough to break registration across a blended array. Screens move too. Any display that was aligned by hand will need to be realigned by hand. This is the cost most system designs underestimate. Manual realignment means scheduling a highly skilled technician, taking the system out of service, and accepting that the result depends on who showed up. With camera-based recalibration, the system is returned to specification in minutes by whoever is already on site, and the result is identical every time. The same mechanism covers technology refresh. When a projector is replaced five, ten, or twenty years into a system's life, recalibration absorbs the change rather than triggering a redesign.

Orthographic calibration maps a static 2D content space onto the screen surface, like stretching wallpaper across it. It is independent of viewing position and looks correct from many locations. Perspective calibration renders a 3D scene correctly from a single specified eyepoint, so a circle in the virtual world appears as a circle to the viewer regardless of screen geometry. Choose orthographic for media playback, attractions, projection mapping, and visualization where the audience moves around. Choose perspective for simulation and training, where a pilot or operator sits at a fixed design eyepoint and geometric truth matters. Perspective calibrations depend on that eyepoint, and Scalable Display Manager supports multiple eyepoints and dynamic eyepoints where the application requires them.

The answer depends on your pixel pipeline, not your industry. If content originates on a single Windows PC, ScalableDesktop is the fit. If a cluster of PCs, an image generator, or a media server drives the display, Scalable Display Manager is the fit. If you are calibrating a CAVE or panel-based environment, Scalable Panel Assembly applies. ScalableDesktop warps the Windows desktop itself, so every application — PowerPoint, AutoCAD, video conferencing, digital signage — runs corrected without modification. All content must originate on or pass through the machine running ScalableDesktop, and the pipeline is not HDCP compliant. Scalable Display Manager (SDM) is the flagship. It supports single-PC and distributed rendering architectures, perspective warps for simulation, and multiple methods of delivering the correction. It is the product used in simulation and training, themed entertainment, planetariums, automotive simulators, and immersive research environments. Scalable Panel Assembly (SPA) targets CAVE and simulator environments built from discrete flat panels to create one continuous screen. Purpose built and unique in its ability to handle edge-blended and edge-aligned surfaces. Scalable Atlas handles projection mapping and 3D warp and blend workflows. Atlas uses a special calibration pattern to handle sharp transitions, gaps, obstructions, and other real-world projection mapping challenges.

ScalableDesktop warps the Windows desktop on a single PC and supports standard Windows applications with no integration work. Scalable Display Manager supports single-PC and multi-PC distributed rendering, exports to media servers and projector hardware, and produces perspective warps for simulation. SDM is the product for real-time and mission-critical systems. ScalableDesktop is common in conference rooms, lobby displays, digital signage, and local exhibits. SDM is common in simulation, visualization, AR and VR, theme parks and attractions, and any environment where the display is driven by more than one machine or where perspective accuracy is required.

Built-in blending gives you the controls. It does not give you the measurement or ease-of-use. The difference Scalable makes is the camera: it precisely measures where every projector pixel should land and computes the correction from that measurement, rather than asking someone to nudge sliders until it looks good enough manually. Two consequences follow. First, setup time. What takes a highly skilled projectionist days to perfect by hand is done automatically, and the quality does not depend on which technician was available. Second, and more significant over a system's life, maintenance. When a projector drifts or is replaced, the display is returned to specification at the touch of a button instead of scheduling a costly site visit. Scalable layers on top of hardware you have already specified. If your display is technically achievable with the tools in your projectors or media server, Scalable still earns its place by automating setup and eliminating on-site maintenance for the lifetime of the system.

Yes. The SDK reads and renders Scalable Mesh Files. It does not create them. A license for Scalable Display Manager or Scalable Panel Assembly is required to run the calibration and generate the files. Think of SDM as the warp and blend generator and the SDK as the runtime. The split is what gives you hardware flexibility: SDM handles measurement and computation on the calibration PC, and the SDK lets any integrated application consume the result on Windows or Linux, on any graphics card that meets the application's own requirements.

Scalable delivers simulation-grade accuracy. Edge blending is registered at the sub-pixel level and geometric warping is accurate to 2.82 arc minutes, or 0.047 degrees, as measured with a theodolite. Color matching holds to ∆x 0.01 and ∆y 0.015 using the existing calibration cameras, and brightness uniformity holds within 10% of the desired variation in foot-lamberts. These are system-level measurements, bounded in practice by projector resolution and screen behavior rather than by the software. This is the accuracy standard that has made Scalable the default in defense simulation and training, where a geometric error is a training error.

Recalibration runs in roughly 30 to 60 seconds per projector. A ten-projector system realigns in five to ten minutes. A two-projector system realigns in under a minute. Recalibration is a single button press and requires no specialist knowledge. Initial configuration takes longer, because it is where screen geometry, camera placement, control points, and system-specific variables get established. That work is done once. Every recalibration afterward reuses it, which is why the ongoing cost of a Scalable system is measured in minutes rather than site visits. Calibrations can also be scheduled to run automatically with no user intervention.

Scalable supports flat, cylindrical, dome, hemispherical, toroidal, and articulated screens as built-in geometry types, plus custom geometry imported as an OBJ file. There is no practical limit on screen size or projector count for Scalable Display Manager. The one real constraint is smoothness. The screen should be free of sharp corners, localized discontinuities, and interruptions in the projected imagery. Seams, cutouts, and obstructions can prevent the software from correctly identifying calibration patterns in the camera image. Where a screen does have these characteristics, a binary calibration pattern is available as an alternative capture method. If you are unsure whether a geometry is supported, send the model. Scalable reviews screen geometry as part of system design and will tell you before you buy.

Scalable Display Manager supports unlimited resolution and an unlimited number of projectors, across distributed rendering clusters and media server architectures. ScalableDesktop is capped by Windows itself at 16,000 pixels in any single direction and by the graphics hardware in the machine. Complexity, not count, is the real design variable. A 30-projector dome with clean geometry and uniform hardware calibrates more predictably than an 8-projector system with mixed projector models and obstructed cameras.

Yes. Scalable supports tiled (blended) displays, stacked displays, and both at once. Stacking combines multiple lower-lumen projectors to reach a brightness target that would otherwise require a single expensive high-lumen unit, and Scalable registers the stacked images to each other automatically. Stacked and blended together is common where the display is tiled across a large surface and then layered with additional projectors for redundancy, brightness, or passive stereoscopic 3D. Redundancy matters in installations that cannot go dark: if one projector fails, the stack keeps the image up.

Yes. Scalable supports both active and passive stereoscopic 3D. Passive stereo is typically implemented as a stacked configuration with polarization, which Scalable calibrates as it would any other stacked array.

Yes. Scalable performs automatic color and intensity calibration using the same cameras that handle geometry, adjusting per-projector RGB levels across a grid to bring the display toward uniform color and brightness. A custom color target can also be trained where the screen shows view-dependent behavior. Color matching holds to ∆x 0.01 and ∆y 0.015. In practice, matching is most effective when projectors are the same manufacturer and model, and adjustments are distributed evenly across projectors rather than concentrated on one. Correcting toward the darkest projector in an array will cost you overall system brightness, which is a design consideration worth resolving before purchase rather than after.

Yes. Scalable Display Manager includes a black level panel that applies a software uplift to identified regions, improving black uniformity across the screen. Software black level correction trades away some dynamic range, so a physical baffle or optical blend plate still outperforms it where absolute black matters. For that reason software black level correction is typically not deployed in simulators, where dynamic range in dark scenes is part of the training fidelity. It can be well suited to museum, attraction, and ProAV environments where the display sits in controlled ambient light and the expectation of contrast and quality can be managed.

Yes. Scalable Display Manager includes a collimation panel to simplify the setup of these highly complex visual systems. Enter the pilot and copilot eyepoint positions and the mirror parameters, and the software computes corrections needed to compensate for mirror distortion, with a tweaking grid available for any residual error. Eyepoints and mirror parameters must use the same coordinate system and units as the screen geometry. Line selection mode is the most effective tweaking method on collimated displays.

Complexity introduced at the design stage is the main driver of difficulty, and almost all of it is avoidable if all parties align early. The most common source is trying to map a complex shape without a 3D model or design diagram. This is compounded if the content provider is not aligned with the systems integrator on pixel canvas size and UV texture mapping requirements. Scalable can import custom OBJs and UV to ensure the content looks as-designed, but any gaps in knowledge through the delivery chain can result in challenges on-site. There are also more fundamental things: obstructed cameras, blocked projectors, poor camera modeling. None of these make a system impossible, but they all add time on-site. Every one of them makes it slower to commission and harder to support over its life. The cheapest hour you will ever spend on a multi-projector display is the one where you send Scalable the design before the hardware is ordered.

Fewer is always better, but there is no single right answer — use however many you need. Technically one camera is sufficient if it can see every projected pixel without obstruction. Cameras are added only when complex geometry or limited throw distance leaves part of the screen unobserved. Every additional camera adds calibration time and design and setup complexity. Cameras can be placed almost anywhere. There is no requirement that they sit at screen center or be co-located with the projectors, which is what makes Scalable workable in rooms where the obvious camera position is occupied by the thing you are actually building. Validate the camera layout with Scalable before finalizing it. Camera modeling services are available to determine the right camera count, placement, and lens selection for your geometry.

Scalable supports Canon DSLR Rebels, Basler Ace GigE, ImagingSource GigE, and Logitech webcams. For simulation and training, only the Canon and GigE cameras are recommended. Multiple lens options are available for each, including fisheye lenses for domes and planetariums and longer throw lenses for outdoor projection mapping. Canon DSLR offers high resolution, low noise, good contrast and sharpness, and reliable setup. Field of view is moderate at roughly 90 degrees with a 10-22mm lens. USB cabling needs planning on long runs. The body is large and looks like a consumer camera, which matters in installations where the camera has to disappear. Basler and GigE offer a small machine-vision form factor and ethernet cabling, selected for high resolution and performant optics — these are the most frequently used cameras. Ethernet connectivity allows for long cable runs and PoE means no local power is required. The supported camera list is maintained on the Scalable website and continues to grow. If you need a camera that is not on it, ask.

Design blend zones at 15% to 20% of the projector's image span. Avoid going below 15% as the human eye is very good at picking up differences between projectors with limited overlap. A wider blend gives the software more distance over which to transition intensity and color differences between projectors, which is what makes a seam invisible. Too little overlap and the blend becomes visible or shows artifacts. Go to about 25% for rear-projection screens or displays that will be viewed from very wide angles, where blends can become view-dependent due to screen material or angle of incidence. Other factors affect blend quality: screen material, the angle at which projectors hit the screen, and the specific overlap pattern. Highly oblique projection creates large disparities in pixel density and therefore intensity across the screen. Overlaps that pinch to a narrow region are difficult to blend well and are worth catching on the layout drawing rather than on site.

No. Any make and model of projector will work. Scalable is a pure software product and does not require warp or blend features in the projector itself. Some design guidance still applies. Use projectors from the same manufacturer and the same model where possible, place them at equal distance from the screen, and focus them properly for the best final image quality. Turn off dynamic black, adaptive contrast, auto iris, overscan, and keystone correction, and set input range to full 0-255. Dynamic contrast features continually change brightness based on content, which breaks blends and interferes with color calibration.

For 3D camera placement Scalable accepts STP, DWG, and DXF files. A PDF of a 2D plan is enough to estimate camera quantity and approximate location, but a 3D model produces a far more reliable design. See the Camera Selection and Placement quick guide for details.

Scalable Display Manager supports four methods of warp delivery: the Scalable SDK integrated into an image generator or media server, the NVIDIA graphics driver (NVIDIA Quadro only), an integrated warping box, or the projector's own firmware (Barco and Norxe). The method is chosen to fit your architecture, not the other way around. This is what hardware independence means in practice. Scalable does not sell you a box that has to sit in your signal path. The correction is a file, and you apply it wherever it makes sense for your system.

Yes. Scalable offers the SDK free to customers and partners. It is available for OpenGL, DirectX, and Vulkan, running on Windows and Linux, and is a proven zero-latency implementation. There are five core function calls, and most integrations are completed in about a day of work. Because all correction happens in a software layer, the SDK imposes no graphics card requirement beyond what your own application needs. Software partners who integrate it become Auto-Calibration Ready, which means any customer running Scalable Display Manager can calibrate a display driven by your application without further work from either party. Calibrations are still generated by Scalable Display Manager. The SDK reads and renders the resulting mesh files at runtime.

Yes. Scalable Display Manager exports warp files configured for Unreal Engine, either through the Scalable EasyBlend SDK integration as an nDisplay configuration, or as an MPCDI file built to Unreal's compatibility requirements. Unreal output requires an eyepoint set in the Perspective panel. Scalable was heavily involved in the initial development of nDisplay warp and blend support.

Yes. Scalable Display Manager exports MPCDI with control over version, profile, depth, pixel pitch, and gamma, for use with third-party playback systems that consume the standard. Scalable is a co-author on the MPCDI white paper along with Disney and others. Pixel pitch is the setting worth understanding. Lower numbers produce higher-density warp files. Flat or slightly curved screens are usually fine at a pitch of 8, curved screens often work better at 4, and strongly curved screens or harsh projector angles may need 2. Very dense MPCDI files can affect playback performance, so test your content for stuttering at low pitch values.

Scalable maintains SDK integrations with leading image generators, media servers, and rendering applications across simulation, visualization, and themed entertainment. The current list is published on the Scalable SDK page. This list grows every year.

Scalable Display Manager can connect to supported projectors over the network and control their settings directly. Currently supported: Barco Pulse, used on F70, F80, F90, F40, F400, UDX, and UDM models among others; and Norxe Unify, used on all current and future Norxe projectors excluding the P1, P1+, and P2 — double digit models support Unify, for example P20, P60, P55. Firmware confirmed in testing: Barco 2.4.5 and Norxe 1.5. Other versions generally work provided the communication API did not change between releases. Projector support requires custom driver work per manufacturer, so if you need a model that is not listed, ask and it can be evaluated.

It depends on how the correction is applied. If you are using the Scalable SDK, a warping box, or projector firmware to apply the warp, the graphics card is entirely your choice and should be specified for your application's rendering demands. If you are applying the warp through the NVIDIA driver, a supported NVIDIA professional card is required. With Scalable Display Manager's host and client architecture, the host PC running SDM has no specific graphics card requirement. Only clients applying the warp through the NVIDIA driver are constrained.

Scalable Display Manager runs on Windows. Applications using the Scalable SDK to apply the correction can run on Windows or Linux, since the SDK operates as a software layer inside your rendering pipeline.

Scalable invented camera-based projector calibration and has done nothing else since 2004. The technology originated at MIT, and Scalable remains the only independent, pure-software calibration specialist in the market. Every competitor bundles alignment into hardware or into a broader integration business. Scalable sells calibration, and only calibration. That focus is the argument. It means Scalable is genuinely hardware-independent and non-competitive with the integrators and manufacturers who deploy it. It means the correction suite goes well past warp and blend to color matching, black level uplift, real-time perspective correction, image masking, and frustum calculation. It means cameras can be placed almost anywhere rather than in the one position a hardware vendor designed for. And it means an integration ecosystem broad enough that there is a delivery path for essentially any pixel pipeline.

Scalable Display Manager is deployed in defense simulation and training, themed entertainment, museums, planetariums and dome theaters, automotive simulators, control rooms, corporate ProAV, and academic immersive research environments. Scalable clients include some of the biggest organizations in the world, deploying systems globally at the highest standard of excellence.

Scalable sells through a global network of resellers and integration partners, with company headquarters and R&D in the United States. Contact Scalable directly to be connected with the right partner for your region or to discuss becoming one.

Scalable software is licensed to the physical calibration PC that runs it. A license is requested from Scalable through the software's license request feature and applied through the license wizard. Systems using Display Clients receive a fixed number of Display Client licenses, determined during the sales process based on system architecture. The license itself is based on the number of projectors and any advanced features needed. One thing worth knowing before you deploy: once a Display Client license is assigned to a machine, it cannot be reassigned without contacting Scalable support. Verify your assigned client list is correct before allocating licenses. Back up your configuration often. The software creates an automatic backup at the end of each calibration and retains up to ten, and those backups are what make a cold start recoverable rather than a rebuild.

No. There is no requirement to use a Scalable PC as a playback platform, and customers are free to build their own machines. The case for the Scalable PC is single-point-of-contact support. It is a turn-key machine tested in house and fully supported, which means hardware and software compatibility is not your problem to diagnose. Home-built machines can be difficult to support and maintain over the life of a multi-display system, particularly when a system that worked at commissioning starts behaving differently after a driver update three years later.

Scalable provides direct technical support from the engineering team that builds the software, not a tiered call center. Remote sessions, system design review, camera modeling, and calibration troubleshooting are all available. This is available in either Standard or Premier service tiers. Standard includes email, ticketing system, and dataset exchange. Premier includes email, ticketing system, dataset exchange, phone, TeamViewer, and remote setup. To get the most from a remote session: make sure the calibration PC has a solid internet connection and remote access software installed, have someone physically present at the display to assist, and reserve the space so lighting can be controlled and people are not moving through the calibration area.

Yes. Scalable offers regular partner training courses both online and in person. The curriculum covers SDM workflow and architecture, system design and how to avoid common pitfalls, advanced features and when to apply them, diagnostic image interpretation, and hands-on troubleshooting of complex simulation displays. This training can take place at Scalable HQ, customer site, or remotely, depending on the level of curriculum. If you are interested in becoming better trained, contact your Scalable representative.