Technology

RealityServer®

RealityServer® by mental images provides server-side 3D rendering capabilities that allow extremely detailed and complex models to be interactively explored with virtually any client hardware. RealityServer® applications are typically deployed within a web-browser environment. Unlike client side solutions RealityServer® applications do not require any browser plug-ins and typically work even through corporate firewalls as they use standards based web delivery.

Additionally, as original 3D data is never tranmitted to the client-side, data security is assured and RealityServer® can operate either over the Internet or corporate Intranet. To provide interactive server side rendering, extensive use of 64-bit hardware is made. Luminova houses considerable hardware resources for RealityServer® based applications and developments.

Using RealityServer® the original full complexity data sets from CAD/CAM, GIS and other modelling systems can be utilised, without the need to simplify the data, thus maintaining the full integrity of the information. Multiple users can access the data from any location in the world from virtually any device with Internet connectivity, including Laptop computers, PDAs and Mobile/Cellular phones.

If required, additional metadata may be associated with model detail and integrated during interactive viewing. Luminova RealityServer® applications are currently being utilised on some of the largest construction projects in the world and by some of most well known retail and engineering firms. In many cases, RealityServer® is allowing these businesses to utilise large scale data sets they have previously never been able to fully assemble as one environment.

In addition to a standardised viewer environment and streamlined RealityServer® conversion tools for common applications such as Maya, 3ds max, Pro/E and AutoCAD, Luminova has also developed specialised viewers for use on Mobile/Cellular phones supporting Java applications. Luminova can provide a standardised RealityServer® based environment or highly customised bespoke solutions with wide ranging functionality. Please contact Luminova for further details or to arrange a demonstration.

About RealityServer® (from mental images website):

‘RealityServer® is the unique server-based, scalable infrastructure software platform for creating and deploying 3D Web Services and other applications and application services for interactive and possibly collaborative remote creation, manipulation and visualization of 3D digital content of essentially unlimited complexity. Users can, for instance, access 3D models based on original CAD data, participate in collaborative online design reviews or participate in multi-player games from any device with only a browser and Internet connection - including mobile devices and smart phones.’

mental ray®

Luminova has selected mental ray® as its primary rendering technology for over 5 years. The combination of leading edge features, production proven quality and full programmability make it an ideal solution for the complex rendering requirements involved in typical production work at Luminova. Working constantly with mental images, Luminova seeks to improve and push mental ray® to its limits on a daily basis.

While there are today many rendering technologies available, none exhibit the production maturity of mental ray®. With full support for 64-bit processing on a wide variety of platforms, mental ray® is able to handle extremely complex scenes, in terms of geometric, material and lighting complexity.

About mental ray® (from mental images website):

‘mental ray® is the world leading, Academy Award® winning, high performance, photorealistic rendering software. It produces images of unsurpassed realism for digital content creation and design in the areas of entertainment, product design and data visualization, including such applications as visual effects for motion pictures, full-length feature animations, content creation for computer games, Computer Aided Design (CAD), product design and styling, architectural design, lighting design, fluid flow simulation, seismic data studies and medical imaging.’

Production Environment

In order to greatly accelerate the production of 3D output, Luminova utilises a heavily customised pipeline build around mental ray® rendering technology. All mental ray® shaders used at Luminova have been developed in-house to ensure full control over rendering speed and output results. These have been seemlessly integrated into the standard tools so that they can be used just like conventional material shaders and linked with off-the-shelf components which already exist. The user interfaces for these shaders has been carefully designed to minimise problems in the downstream rendering process.

Building on off-the-shelf tools, significant modelling automation and efficiency enhancements have been created at Luminova to allow for rapid creation of modelling detail for both mental ray® output and RealityServer® applications. Additionally, in order to deal with massively complex scenes which cannot be loaded in current products such as Maya and 3ds max, Luminova has created proprietary technology to efficiently manage this complexity within those tools proir to use in mental images technology (which fully support this level of complexity). These in-house tools greatly minimise processing times and repetitive computation which occurs in standard environments.

For management of re-usable 3D detail such as 3D models, textures and light source photometry, Luminova has created the Luminova Library System. This in-house web-based database manages all re-usable 3D assets at Luminova and is fully integrated within the production pipeline allowing scenes to be rapidly created from libraries of over 20,000 3D models, 10,000 textures and 48,000 photometric light sources. When required, customers can be given read-only access to the Library System in order to select desired modelling detail from pre-existing elements.

Luminova have heavily optimised and geared the production pipeline towards 64-bit technology, allowing massively complex scenes to be processed as single models, removing labour intensive simplification or compositing tasks from the process. For dealing with large volumes of scene data, Luminova has developed a complete suite of tools for processing and manipulating mental ray® scene files (.mi) in any means required with completely scriptable interfaces for building future requirements.

64-bit Processing

The advent of 64-bit processing has vastly increased the amount of memory available to rendering applications. This has a direct impact on the size and complexity of scenes which can be handled by a processing system. 32-bit processors are limited to 2Gb1 or 4Gb of memory and as such, must swap scene data in and out of memory in order to render scenes larger than this memory size. Some users even resort to compositing mulitiple images together to manage the complexity. Using 64-bit technology it is quite feasible to utilise even 64Gb of memory2 and so the entire scene geometry can reside within system memory.

Typically, if a scene is larger than the available memory, utilising 64-bit hardware to eliminate swapping will increase rendering speed by a factor of 10x - 15x. Additionally, far less labour is involved since the scene does not need to be split up in to smaller pieces in order to be rendered. mental images technology is fully 64-bit compliant and has been used on 64-bit hardware for many years. Luminova utilises a full 64-bit pipeline, each element in which there are potential complexity limitations are built with 64-bit technology to ensure scenes of virtually any size can be rendered.

1 Microsoft Windows based platforms impose a limit of 2Gb of physical memory, with the exception of platforms utilising the /3G switch.

2 64-bit systems theoretically allow 2^64bits of memory to be addressed, the current limitations are commercial in nature.

Computing Resources

For both Production work and RealityServer based services, Luminova heavily exploits large scale, 64-bit parallel processing systems. Utilising high density blade servers, Luminova is able to rapidly respond to customer demand for processing resources. Currently over 200 CPUs are available for rendering tasks and more are being added at regular intervals. All new hardware is fully 64-bit with between 16Gb to 64Gb of memory.

To manage the large number of processors required, Luminova has created a customised node management system to manage hardware resource allocation and running jobs. System maintenance, replication and updates are all managed through systemised processes to ensure maximum uptime of all processing resources.

For high end RealityServer applications, Luminova maintains several 4-way Dual Core (8-cores) Opteron based systems with capacity for 64Gb of memory. This system can be used to process models of virtually unlimited size in practical time frames and models of up to approximately 250M triangles in real-time. All software utilised in the Luminova core process is scalable and able to exploit both networked and SMP based multiprocessing.

RPC

For providing scenes with a wide variety of realistic people, trees, vehicles and many other object types, Luminova has implemented RPC for Maya (mental ray) which allows ArchVision's RPC content to be rendered directly within mental ray. RPC for Maya is sold as a commercial product through ArchVision and is used extensively in the internal production process at Luminova. For a more detailed explaination of RPC technology, please visit the ArchVision webite.

In brief (from the ArchVision website):

'RPC stands for Rich Photorealistic Content, a term used to describe the software and content associated with ArchVision's award-winning Image-Based Rendering (IBR) technology. (RPC is also the file extension for this image-based content.) RPC is the first major commercial project in the evolution of image-based rendering technology. Since its introduction in 1998, the company's solutions have become the preferred method of incorporating complex objects into 3D computer graphics environments. Because RPCs rely on photo image data, the incredible richness of detail and quality of images far surpasses even the best computer-generated models.'

In addition to RPC for Maya, Luminova's RPC rendering technology for mental ray® also allows 3ds max users to access RPC content when using mental ray. As well as commercially sold RPC products, Luminova has also developed the worlds only RPC for Linux rendering solution, with full support for 64-bit hardware and its extensive memory capacity. This proprietary solution is deployed exclusively at Luminova to allow RPC technology to be utilised in conjunction with large scale 64-bit parallel processing.

Customers of Luminova production services can choose from all available ArchVision RPC content for inclusion in their output as Luminova maintains a complete library of RPC content for use with its system. RPC capabilities are provided as part of Luminova's core service and virtually any number of RPC objects can be inserted into scenes when utilising Luminova's 64-bit rendering solution. Luminova has an ongoing development partnership with Archvision.

Archvision

About ArchVision Inc.:

ArchVision is the leader in the development of image-based content creation, management and delivery solutions. In addition to its RPC (Rich Photorealistic Content) Plug-in technology, ArchVision's products include a growing library of RPC content ranging from people, trees and shrubs to image-based automobiles and other popular objects.

ArchVision currently serves customers located in over 90 countries around the world. RPC is the image-based rendering technology of choice for many of the leading animation, CAD, real-time simulation, and 3D modeling software packages on the market today. Native and plug-in RPC support now exists in 3dsmax, Autodesk VIZ, Autodesk Civil 3D, Bentley MicroStation, Alias' Maya, Newtek's LightWave 3D, Maxon's Cinema 4D, AccuRender from Robert McNeel & Associates, Revit, Piranesi from Informatix, Adobe Photoshop, Graffiti RenderPro, Cubicspace's rtre, Arc+ Render, VR4MAX, and SGI Performer. For more information, call ArchVision at 859-252-3118 or visit us on the web at www.archvision.com.

Daylight Simulation

Heavily customised mental ray® shaders and production integration have been developed to support straight forward modelling of daylighting based on date, time, location and conditions. The Luminova toolset fully supports CIE and IESNA based daylighting standards if required, however these standards have been found to represent a poor match to real world daylighting situations. As such, Luminova additionally supports the Perez All-Weather daylight model which provides a far more accurate representation of daylight distribution.

An important consideration is that CIE/IESNA and Perez models do not provide any spectral information and so ignore the colour of the sky, they provide luminance information only. Typically, the sun and sky are simply assigned a light yellow and light blue colour respectively and the daylight models intensity is combined with this. To create truely automatic and accurate daylight simulations, Luminova has implemented a full spectral sky model which automatically computes both sky luminance and spectral properties (colour) based on time of day, date, location and atmospheric parameters such as turbidity.

While not directly part of daylight simulation, also available is an atmospheric scattering model to simulate the effect of aerial perspective in large exterior scenes, where the scattering imposed by the atmosphere between the viewer and the object can be significant. All daylighting applications at Luminova draw on significant experience from Lighting Design applications for standards compliance and other common daylighting exercises.

In addition to physically based daylight simulation, Luminova has created enhancements to the Photon Mapping algorithm specifically for dealing with daylight modelling, in particular the difficult problem of interior daylight modelling where Photon Mapping can be extremely problematic where windows or openings represent a small percentage of the model being considered. Finally, as daylight simulation produces HDR output, it must typically be combined with one of the several tone-mapping operators available at Luminova.

High Dynamic Range Imaging

High Dynamic Range or HDR imaging relates to methods for encoding extremely high-contrast information in computer images. To understand this concept, it is important to know what is meant by the term 'Contrast Ratio'. This essentially means the ratio of the brightest representable value to the darkest. Typical computer monitors have a contrast ratio of approximately 100:1 where as the real world can have contrast ratios in excess of 5,000,000:1. Standard computer images store 8-bits of information for each colour channel, allowing 256 possible values. High Dynamic Range images store additional information, usually 32-bits for each colour, allowing 2^32 possible values.

Luminova works with both the capture of and use of HDR imagery. Tools created at Luminova for both production rendering and technology development typically allow for the use of high dynamic range images. Luminova's production process can either output standard low dynamic range images or high dynamic range information. When performing physically based simulation, images contain real-world units (e.g. cd/m2) which can only be encoded in HDR images. HDR technology is now becoming mainstream with companies such as Abobe introducing HDR capabilities into their Photoshop CS2 product.

Considerable research effort has been expended on a topic known as Tone-mapping. This is the process of converting high-contrast images into a form suitable for display on standard low contrast display devices, a process referred to as Dynamic Range compression. Additionally, tone-mapping can also deal with representing the human psycophysical response to high dynamic range data, attempting to simulate such effects as glare, contrast sensitivity and visual acuity. Some Tone-mapping techniques are primarily for asthetic quality, while others are more focused on quantitative analysis. Luminova has implemented many tone-mapping methods which have been put to use on various projects. They include:

  • Simple Logarithimc Methods
  • Gradient Domain Compression
  • Bi-lateral Filtering
  • Photographic Tone-mapping
  • Human Visual System Based Tone-mapping


Several of these tone-mapping techniques are only commercially available at Luminova and can be applied to a variety of applications. Tone-mapping, however is primarily used to overcome inadequecies of current display devices. To reduce the need for sophisticated tone-mapping and to increase the impact of HDR imagery for the user, HDR Display Systems must be used. For these requirments, Luminova works directly with BrightSide Technologies Inc.

BrightSide Technologies

About BrightSide Inc.:

BrightSide Technologies Inc., a Vancouver based developer of advanced display technology for monitors and TV's, announced that it has launched the World's First Extreme Dynamic Range display. BrightSide's new display combines a matrix of individually modulated LEDs in the backlight with an LCD High Definition panel to deliver 10 times the brightness and 100 times the contrast of conventional display products. The displays are 16 bit per color and are capable of going to absolute black while giving a contrast ratio of over 200,000 : 1.

BrightSide's displays allow our customers to see images as they appear in the real world. They will initially be used in "vision critical" applications where customers need to see their images in vivid detail. Ultimately, however, we expect that our technology will be incorporated in more mainstream consumer products.

BrightSide's initial targets for its new display include film post production, satellite imaging, geophysical data visualization, gaming and high end home theatre markets.

PaintExplorer

A revolutionary system for visualising colour selections for the consumer and industrial decorative coatings industries, PaintExplorer simplifies the task of visualising colour selections on users images to a point and click operation. There is no need to manually draw around the areas the user wishes to recolour. Simply selecting the desired colour and clicking, or clicking and dragging on the surface will automatically create the selection for the user as they paint.

PAINTEXPLORER
DEMONSTRATION MOVIE


For cases where the user wishes to select more specific areas, a special 'Scissors' tool is available which follows exsiting contours in the image, removing the need for tedious manual selection. This tool can also be used to split a single surface into multiple surface, for example to paint the upper and lower portion of a wall in different colours.

PaintExplorer utilises advanced image segmentation methods to isolate the portions of the image the user wishes to paint. Typically images can be painted by the user within around 30-60 seconds, where as previous solutions could require anywhere from 10-30 minutes depending on the complexity of the objects in the scene.

Features at a Glance

  • Easy to Use Click and Drag Painting
  • Scissors for More Complex Selection
  • Selection Cuts to Divide Areas
  • Interactive Zooming and Panning
  • Full Undo and Redo
  • Lighting Compensation
    (Incandescent, Halogen, Fluorescent and Daylight)
  • Predictive Colour Matching
  • Automated or Fully Programmable Mask Creation (During Painting)
  • Loading of JPEG, TIFF, PNG, BMP and Other Image Types
  • Automated High-quality Resizing of Photos
  • Feature Blurring Tool (for Pattern Removal)
  • Custom Compressed File Format (No Generational Loss)
  • Localisation Support (UNICODE)
  • Straight Forward API (Application Programming Interface)


Typically applications based on PaintExplorer are small enough for use over the Internet either on websites or in email based marketing efforts. Tools based on PaintExplorer can be as little as 500Kb in size, often smaller than the images being edited by the tool. Other painting applications are typically over 20Mb in size and not suitable for such widespread deployment.

In order to allow broad application of PaintExplorer technology, a full implementation has been created in Java and can be deployed to Windows, MacOS, Linux and any platform supporting the J2SE. Both stand-alone applications and web-page based applets can be implemented. Luminova has heavily optimised the PaintExplorer for speed and memory efficiency.

In addition to decorative coatings applications, PaintExplorer technology has also been applied to such applications as satellite image analysis and shadow detection and removal. Further to the PaintExplorer core tools, considerable on going research and development is taking place at Luminova to advance the technology into other areas, including:

  • Image Completion
  • Pattern Detection and Removal (e.g. Bricks)
  • Improved Selection Algorithms
  • Texture Based Selection
  • Perspective Texturing


Luminova has already created industry leading consumer painting applications for Orica/Dulux and Akzo Nobel. The Dulux MyColour system, now in its third generation has been in commercial use for over 5 years. PaintExplorer is a proven technology with a significant track-record in providing measurable marketing impact.

PaintExplorer technology can be provided either as a bespoke customer solution or a programming API for in-house integration. When provided as a customer solution, Luminova will work directly with the customer to implement their own user interface requirements, colour databases and other brand specific application features. If in-house development is preferred, Luminova can provide an straight forward API called PELIB which provides developers access to the core PaintExplorer technology via Java. Please contact Luminova for further details or to request a demonstration.

LIBAT

Luminova Image Based Analysis Technology or LIBAT is a suite of tools for the extraction of 3D geometric information from single satellite images and advanced satellite image analysis. LIBAT based modelling is avaialble to customers of Luminova production services. Customers who are interested in utilising LIBAT in-house should contact Luminova for additional information.

The primary focus of LIBAT is obtaining meaningful 3D information from a single image. Unlike tools which require stereo pairs or sophisticated calibration information such as the Rational Function Model, LIBAT relies on shadowing information within the image to obtain 3D height information. With a very simple interface, users outline buildings in the scene and then correspond shadows with these buildings. Several modelling primitives are available, including buildings, cylinders, pipes, trees and walls. Secondary features such as pitched roofs can also be created easily within the tool.

LIBAT DEMONSTRATION MOVIE

Example LIBAT Site

LIBAT loads industry standard GeoTIFF images and accepts both projected and unprojected imagery. As satellite imagery can be extremely large (16k x 16k pixels is not uncommon), LIBAT has been designed specifically to handle extremely large image files, interactively on commodety hardware. A 1.5G GeoTIFF image can be easily manipulated on a standard laptop computer using approximately 80Mb of memory. Additionally, LIBAT maintains the full dynamic range of the image data, so for example, 11-bit QuickBird imagery retains all detail and dynamic range of the original data. Panchromatic, Natural Colour and Multispectral imagery are all supported.

LIBAT models can be exported to multiple formats, including Wavefront OBJ, Maya, Google Earth and LIBAT Viewer. 3D information can then be further manipulated in standard modelling packages or merged with modelling data from other systems. LIBAT scene data has already been extensively utilised for military, intelligence, architectural and urban planning projects.

LIBAT Viewer is a companion tool to LIBAT which allows LIBAT scenes to be viewed interactively in real-time, without the LIBAT tool. LIBAT Viewer also adds significant analysis capabilities, including real-time sun shadowing (date and time configurable), observer viewing volume visualisation (for location of blind spots) and annotation capabilities (lines, marker signs, solid areas, text). LIBAT Viewer scenes can be created directly from LIBAT with a simple export.

Additional tools under development which fall under the LIBAT umbrella of products are Shadow Removal tools (utilising PaintExplorer technology), Object Recognition and RealityServer integration.

Please see product brochures for more detailed information on LIBAT and LIBAT Viewer.