Thursday, 16 August 2007

Urban modelling via a multi-touch display wall

Urban modelling is arguably an area that has yet to benefit from 3D interaction and quite understandably so since the work involved is often very complex and large-scale and doesn't really lend itself to HCI experimentation. However, Autodesk Labs plans to change all that and has recently unveiled vids of a new research prototype of a novel urban design application.

This prototype runs on a Multi-Touch Wall, a multi-touch input device produced by Perceptive Pixel (link here) and invented by researcher and TED conference luminary Jeff Han. Single touch devices, such as a for example a tablet PC or kiosk at a shopping mall, have been around for years. What distinguishes a multi-touch device from its predecessors is that it recognizes more than one input simultaneously. That means that instead of just touching the screen with the tip of one finger, the device recognizes gestures that the user makes with one or more fingers. Recognizing these gestures can provide computer applications with new opportunities for processing user input...

Therefore, Autodesk Labs believes multi-touch human-computer interfaces may dramatically change how products, infrastructure, and buildings are designed. In the vid above you can see someone grab a section of a city and mock up a building on the site. It looks really exciting (despite the lack of audio commentary on the vid) and could well spark off new avenues in urban modelling research when the project is complete and made commercially available.

Wednesday, 15 August 2007

Spellbinder project: mobile snaps revealing invisible art

A very interesting research project, unveiled at the recent SIGGRAPH 2007 conference, that has only recently come to my attention is the Spellbinder one, developed by Scottish researchers at the Division Of Informatics at the University of Edinburgh. The main objective of the project is to turn to camera phones, an accessory available on just about every cellphone these days, to help bridge the gap between virtual and real worlds.

The essence of the project is that via the use of image-matching algorithms the real world is adorned with digital content. The technology developed has already been used to create a guide of Edinburgh that allows people to find virtual artworks (such as for example graffiti, pictured above) placed around the city using their mobile... Another related spin-off application derived from the Spellbinder project uses the same technology to automatically update a person's blog with their location. According to Dr Mark Wright who came up with the original idea "it's about using a camera phone as a magic wand".

A key component of the Spellbinder is the database of all the places that participants have added data to. Thus, people can query it by taking a snap of a location with their phone then using multimedia text messages to send it to Spellbinder. Moreover, again according to the creators, powerful image-matching algorithms are used to analyse the image that can deal with snaps of the same place being taken under different lighting conditions or orientations. Once the system has worked out the location of an image it consults the database and sends back an image with the extras added to it.

All in all this looks like a very interesting concept that very notably manages to combine virtual & augmented reality issues, mobile blogging, urban navigation & representation, invisible art and even location based services and could indeed have many extensions and forays in other fields too...

Below you will find a link for the original BBC News story on Spellbinder:

New version of 3D Studio Max out in October

Autodesk, Inc. has just announced the new version of this writer's favourite 3D modelling application, 3ds Max 2008. While the naming is somewhat perplexing (what happened to version 10?!) 3ds Max 2008 is said to containe new features that enable artists and designers to further overcome challenges associated with large and complex 3D datasets (such as urban models). The software has been used to create numerous notable projects, including visualizations for the EMI Records building in London, the Mass Effect game and the Grindhouse movie. 3ds Max 2008 was showcased during SIGGRAPH 2007, being held at the San Diego Convention Center between August the 7th and the 9th.

Robert Stava, creative director at 3D Media Group and 3ds Max 2008 beta tester said, "Autodesk 3ds Max 2008 quickly became a standard part of our production pipeline. With many of our architectural design projects tipping the scales at millions of polygons, 3ds Max 2008 software's improved handling of complex data is a very welcome addition." 3D Media Group recently used 3ds Max to create visualizations for the Fulton Street Transit Center in New York City, the Zayed Stadium in Abu Dhabi and the New Songdo City in Korea.

Here's a quick overview of some of the key new features of this new version:

1. Accelerated Performance
The integration of new technology into the software’s Adaptive Degradation System improves interactive performance by automatically simplifying scene display to meet a user-defined target frame rate. You control how 3ds Max adjusts scene display—whether the smallest objects are hidden, or distant objects have less detail, etc.—and 3ds Max calculates how best to achieve it. When combined with the new Direct3D® mesh caching that groups objects by materials, the result is that tens of thousands of objects can be just as interactive as ten objects. In addition, loading, arrays, Autodesk® FBX® and OBJ export, and other areas of the software perform significantly faster.

2. Scene Explorer Scene Management
3ds Max 2008 delivers Scene Explorer, a robust new tool that provides you with a hierarchical view of scene data and fast scene analysis, along with editing tools that facilitate working with even the most complex, object-heavy scenes. Scene Explorer gives you the ability to sort, filter, and search a scene by any object type or property (including metadata), with stackable filtering, sorting, and searching criteria. This new tool also enables you to save and store multiple Explorer instances and to link, unlink, rename, hide, freeze, and delete objects, regardless of what objects are currently selected in the scene. You can also configure columns to display and edit any object property, and because this feature is scriptable and SDK extendable, you can use callbacks to add custom column definitions.

3. Review
This powerful new toolset gives you immediate feedback on various render settings, enabling you to iterate rapidly. This means you can now quickly hone in on your desired look without waiting for a software render—erfect for over-the-shoulder client/boss feedback sessions and other iterative workflows. Based on the latest game engine technology, Review delivers interactive viewport previews of shadows (including self-shadowing and up to 64 lights simultaneously), the 3ds Max sun/sky system, and mental ray® Architectural and Design material settings.

4. MAXScript ProEditor
3ds Max 2008 marks the debut of the new MAXScript ProEditor. This intuitive new interface for working with MAXScript includes multilevel undo functionality; fast, high-quality code colorization; rapid opening of large documents; line number display; regular expressions in search/replace; folding of sections of the script; support for user customization; and many other features.

5. Enhanced DWG Import
3ds Max 2008 delivers faster, more accurate importing of DWG™ files. Significantly improved memory management enables you to import large, complex scenes with multiple objects in considerably less time. Improved support for material assignment and naming, solid object import, and normals management facilitate working with software products such as Revit Architecture 2008. Plus, a new Select Similar feature identifies all objects in an imported DWG scene that contain characteristics similar to those of a selected object. This capability lets you select and edit multiple imported objects simultaneously—dramatically streamlining DWG-based workflows.

6. Artist Friendly Modelling Options
3ds Max 2008 gives you a more streamlined, artist-friendly modeling workflow through a collection of hands-on modeling options that let you focus more on the creative process. These options include selection previewing and the ability to have existing modeling hotkeys and pivots become temporary overrides.

7. Biped Enhancements
This latest release provides you with new levels of flexibility with regard to your Biped rigs. A new Xtras tool lets you create and animate extraneous Biped features anywhere on your rig (for example, wings or additional facial bones) and save them as BIP files. These files are supported in Mixer and Motion Flow, as well as in Layers, where new layering functionality enables BIP files to be saved as offsets from each layer to isolate character motion. As a result, each layer can be saved as its own asset for export into a game.

8. Expanded Platform Support
3ds Max 2008 is the first full release of the software officially compatible with Microsoft® Windows Vista™ 32-bit and 64-bit operating systems and the Microsoft DirectX® 10 platform.

Autodesk anticipates that 3ds Max 2008 will be available in English in October 2007. For a complete list of new features and enhancements in Autodesk 3ds Max 2008, visit

Monday, 13 August 2007

Dell Axim x51v and Windows Mobile 6

With the release of Windows Mobile 6, various PDA manufacturers have taken the plunge to make sure their already released Pocket PC devices can be upgraded to the new operating system. Well, Dell is certainly not one of those manufacturers... While it is understandable to an extent (the Axim line seems to be one that will be discontinued) it nonetheless is very frustrating to a user of the device (such as myself, it's been with me since it was released and has proved to be an extremely robust, reliable and very powerful graphics-wise PDA).

The video above (showing a Dell Axim x51v running Windows Mobile 6!) is the proof that user communities can sometimes overcome little issues such as devices getting obsolete by their manufacturer. It seems that a ROM is out there that allows the new version of the operating system to be loaded on the device. While I'd rather not post any links as to where one can find this ROM, it does reportedly provide Axim users with the capability of upgrade denied by Dell (albeit as a slighty more risque proposition, ROMs like this often tend to be incredibly buggy and are not really recommended). Moreover, it applies further pressure to the manufacturer to step up to the challenges of a still-existing customer base. Have I attemped the upgrade? No way, and I certainly won't as I am not convinced WM5 and WM6 differ greatly but also because I value my device way too much and will always wait for an official update from Dell...

Nintendo developing a new motion sensing portable device?

For today's post, a new (or perhaps even better, hybrid) concept which is certainly interesting, concerning interaction & mobile devices capable of 3D graphics. It seems that Nintendo has caught on as to how important their Wii motion-sensing controller has been to their sales and are trying to apply it to its mobile counterpart, the Nintendo DS. That seems to be the idea behind a recent patent application from Nintendo, which is for a handheld gaming device with Wii remote-esque motion-sensing abilities.

The housing of the system incorporates an XY-axis acceleration sensor to detect an acceleration in an X-axis and Y-axis direction and a Z-axis contact switch to detect an acceleration in a Z-axis direction. In diagrams (see pic above), the system has been represented by a Game Boy-esque figure, but it could well be that's just a temporary placeholder.

While it should be noted that mobile devices with motion sensors are certainly nothing new, this has been more or less limited to cellphones so far and seeing that a device capable of displaying more complex 3D graphics extend to include it as a viable interaction way is very exciting indeed! Time will show what Nintendo will make of this of course as this is a long way from implementation yet. Still, the idea alone sends the imagination reeling when it comes to potential applications...

Monday, 6 August 2007

Tutorial on creating a photorealistic 3D virtual city in under 5 minutes

Quite often I've had to produce random visualisations and renders of 3D urban models for my work with time being a key element. In this post I have decided to describe an easy, cost-efficient and most of all very fast (less than 5 minutes all in all!) way to do that.

First of all, you will need a version of Autodesk's 3D Studio Max to follow this quick tutorial. Also, you will need a copy of the excellent Greeble plug-in, developed by Tom Hudson. This is a modifier plug-in useful for generating random detail for everything ranging from spaceship models, cityscapes to Death Stars (!), which can be downloaded from (to install it simply copy and paste the Greeble .dlm file in the 3D Studio Max Plugins directory).

Step 1: Create a plane in 3D Studio Max with say 25 Length and 25 Width. Make sure Edged Faces is on from the viewport options (very important!).

Step 2: Apply an Edit Mesh modifier to the plane and then proceed to select every third or fourth row of polygons horizontally. Non-uniform scale these rows of polygons to 50%-60% of their size and then delete them. You have created horizontal streets in your random 3D city. Now repeat the process vertically in exactly the same way only this time it doesn't have to be in continuous rows. Remember we're trying to keep the appearance somewhat not-too-"blocky" looking and fairly randomized...

Step 3: Keeping this in mind move around some vertices here and there in the corners of all buildings to randomize the appearance of the cityscape further.

Step 4: Open up the Material Editor and in a new material slot create a Multi/Sub-Object material. Apply this to the cityscape plane. This is done in order to differentiate different parts of the cityscape. I usually use about 4-5 slots in this but let's keep it simple for this tutorial; use only two and name the 1st Base and the 2nd Garden. Assign a different sub-material to each; say a yellow-ish diffuse color for the Base and green for the Garden.

Step 5: Now under the Polygons selection of the Edit Mesh modifier, assign different material IDs to polygons accordingly. The easy way to do this is to select all polygons, assign them as ID1 (Base) and then select the few Garden polys you want and assign ID2 to them (you will see that the Garden ones turn green while all others remain yellow in color). Finally select ID1, all the Base polys that is.

Step 6: Apply the Greeble plug-in! You probably have what is more or less a random 3D city by now. With a few minor tweakings you can get it even better but even the default settings should be good enough! I usually set Taper to 0, make sure I use all 5 Widgets available, with their Density set at 5 etc.

Step 7: To produce a clay-like render add a Skylight to the scene and then render with some Advanced Lighting->Light Tracer (using the added Skylight). This should produce a render like the one below.

As you can see, it is fairly simple and time efficient to achieve results such as this for purely random visualisations. Obviously the shortcomings of this are obvious (for starters geographically it is difficult, nigh-on impossible to adapt this technique to a more realistic GIS-driven target) but this method (and the Greeble plug-in itself!) take some beating in the time/quality of result ratio for purely "beauty" renders. (Also do bear in mind that I've tried to keep this really really simple, there are many ways to build on and improve on the few steps described above, by no means is this definitive, feel free to experiment!).

For a future tutorial I will concentrate on the production of expressively rendered (i.e. in a non-photorealistic manner) views of 3D urban models, check the pic above for a cartoon-shaded view of an urban model...

MaPZone, a very useful texturing editing tool

One utility I have recently found extremely useful for my work, MaPZone, Allegorithmic’s free procedural texture generation tool, will soon be upgraded to version 2.6. The tool enables users to create seamless, high-quality textures based on a compositing graph which will automatically tile and can be locally edited (amongst many other features, check the link below for all of them). The update, set to be released sometime in the fall of this year, will bring about the ability to input bitmaps so that they can be mixed with the procedurally generated layers that the program specialises in, as well as several other usability improvements.

Well worth a look for all 3D modellers also heavily involved with texturing work, more information on the application, tutorials, a community forum, as well as of course the ability to download the freeware version at

Saturday, 4 August 2007

Google PhotoSketch, a method for image-based rapid photomodelling

According to a Google Tech Talk recently given by George Wolberg a technology has been developed designed to take casual photographs and convert them quickly into texture-mapped 3D buildings. The technology, called PhotoSketch is intended to automate as much of the process as possible, while still requiring some human input. The key is that the methods do not require expensive cameras, laser finders, etc. It operates by extrapolating camera angle and position from the derived 3D of the building based on the photos. The key contributions of this work is that it merges the benefits of automatic feature extraction, multiview geometry, an intuitive sketching interface and also dynamic texture mapping to produce lightweight photorealistic 3D models of buildings.

The talk can be viewed at the link below and while it clocks in at slightly less than an hour (!) it is well worth watching since it presents general city modelling issues as they stand today, an overview of the system proposed and also results from experiments in large-scale urban scenes. Finally, the results are designed to integrate with Google SketchUp and are thus particularly suited to creating buildings for display in Google Earth.

Google SketchUp For Dummies book

Another useful book I have come across recently is Google SketchUp For Dummies, written by Aidan Chopra. While Google's SketchUp is far easier and less complicated than something such as a more dedicated modelling application like 3D Studio Max, it is nonetheless in need of some sort of reference manual to take you through the full variety of features it is equipped with.

While the book itself is more than adequate on that, what is even more impressive is that the author has uploaded 62 tutorial vids to compliment the book release and all of these tutorials are entirely free and can be found on YouTube! A great move both in a marketing and also SketchUp-community sense and one that leads me to recommend this book even more! For more info on the book see

The author's vid tutorial site on YouTube can be found at

City GML, providing an open standard for 3D city models

Services such as the very popular Google Earth and Virtual Earth (both covered in this blog), while pushing the 3D urban modelling field to its limits in more ways than one also carry a lot of limitations. One of the most important ones is that these web services largely provide only graphic or geometric models, neglecting the semantic and topological aspects of the buildings and terrain being modeled.

These 3D models can only be used for visualization purposes but not, in most cases, for thematic queries, analytical tasks or spatial data mining. Another problem with these and other city 3D modeling efforts is lack of interoperability. Web-aligned open standard geometry models for computer graphics (e.g., X3D and geospatial technologies (Geography Markup Language 3 (GML3)) are available but still relatively new and not yet widely adopted. Though their common XML foundation provides the elements needed for convergence, this has not been addressed (not yet anyway). Moreover, in those cases where semantic and topological aspects are considered, there are no widely used standard application schemas, though such schemas are under development, drawing on a considerable body of previous data modeling work. Without a common definition of the basic entities, attributes and relations that can be shared over different applications, those who develop and own the models find it difficult to share and maintain them.

Therefore, for this post, I will provide with a few details on an information model that tackles all these issues and more, called City GML. CityGML, began as a research project and has been developed since 2002 by the members of the Special Interest Group 3D (SIG 3D) of the initiative Geodata Infrastructure North-Rhine Westphalia (GDI NRW) in Germany, and is a common information model for representing 3D urban objects. It defines classes and relations for the most relevant topographic objects in cities and regional models with respect to their geometric, topological, semantic and appearance properties. It should be noted that "City" is broadly defined to include not just built structures, but also elevation, vegetation, water bodies, sidewalk furniture and more. Included are generalization hierarchies between thematic classes, aggregations, relationships between objects and spatial properties.These thematic information types go beyond graphic exchange formats and allow users to employ virtual 3D city models for sophisticated analysis tasks in different application domains such as simulation, urban data mining, facilities management, decision support and thematic inquiries.

So, in short, some of City GML's features include:

- Geospatial information model for urban landscapes based on the ISO 191xx family
- GML3 representation of 3D Geometries, based on the ISO 19107 model
- Texture and material representation of object surfaces
- Taxonomies and aggregations
- Digital Terrain Models as a combination of (including nested) triangulated irregular networks (TINs), regular rasters, break and skeleton lines, mass points
- Sites (currently buildings; bridges and tunnels tbd. in the future)
- Vegetation (areas, volumes, and solitary objects with vegetation classification)
- Water bodies (volumes, surfaces)
- Transportation facilities (both graph structures and 3D surface data)
- City furniture
- Generic City objects and attributes
- User definable (recursive) grouping
- Multiscale model with 5 well-defined consecutive Levels of Detail (LOD):
a) LOD 0 : Regional, landscape
b) LOD 1 : City, region
c) LOD 2 : City districts, projects
d) LOD 3 : Architectural models (outside), landmarks
e) LOD 4 : Architectural models (interior)
- Multiple representations in different LODs simultaneously; generalization relations
- Optional topological connections between feature (sub)geometries
- Application Domain Extensions (ADE): specific "hooks" in the CityGML schema allow to define application specific extensions, for example for noise pollution simulation, or to augment CityGML by properties of the new National Building Information Model Standard (NBIMS) in the US

To provide (only some) names of bodies working on and supporting CityGML: municipalities of Berlin, Hamburg, Cologne, Düsseldorf, Recklinghausen, Leverkusen; British Ordnance Survey, State Mapping Agency of North-Rhine Westphalia; companies T-Mobile, Rheinmetall Defence Electronics, Snowflake, CPA Geo-Information, GIStec, 3D Geo; research institutions: Universities of Bonn, Potsdam, Dortmund, Applied Sciences Stuttgart; Research Centre Karlsruhe and Fraunhofer Institute for Graphics Research.

For anyone with even a passing interest in urban modelling and all associated challenges arising from producing more than just 3D geometry for urban spaces, more details on City GML can be found at

Friday, 3 August 2007

Sony PSP SDK available discounted for UK universities

An interesting piece of news for all of us currently studying/working/researching at UK universities in the mobile device field; Sony Europe has made it known that it will be releasing the PSP software development kit to academic institutions at a "vastly reduced rate" (no details on that one yet). SCEE announced the new, non-commercial scheme, via the appointed Academic Liaison at the Games:Education day of the Developer gaming Conference, held in Brighton.

The PSP SDK which has been offered to students/researchers will come with the same ProDG programming tools and debugging tools used by professionals in studios worldwide. Students/researchers will be able to get support online through a cloned version of the professional development website, which will have a dedicated forum to cater for all questions. It's been speculated that the launch partners will probably be the University of Abertay Dundee, Sheffield Hallam and Hull, all of which have their own computer games programming courses.

While ever-so-important pricing tag is not revealed yet, just a few days after this announcement, it is definitely worth keeping an eye on for further developments since the PSP, in combination with the recently launched GPS add-on can offer many new opportunities for innovative research concepts and ideas...