Advantages and Options of 3D Models

August 11, 2011 By: Art Kalinski

At almost every GIS-related conference I’ve attended this year, there has been a steady increase in the use of 3D models and a growing variety of ways to generate and use the models. It is, however, important to understand why you would want to go through the expense and additional work of getting involved with 3D models when perhaps ortho, street level, or oblique imagery may accomplish the same goal quicker and at significantly lower cost. I contend that the reason 3D is so popular is that it provides continuous oblique views. To prove my point, watch someone using a 3D model. I have yet to observe anyone using a 3D model go to a straight down ortho view and navigate around in the ortho world from that point.

Years ago I started to use geo-referenced metric oblique imagery to supplement our long-established otho imagery library. I frequently use this example of a North Atlanta hospital complex.The building in the red circle is hard to decipher in this ortho view. By comparison, the same building in a high-resolution oblique view is clearly a parking garage. Note that one can also see that the land slopes away from the building and that thin vertical features such as light poles on the roof of the building are clearly visible in the oblique view. These light poles could be real rotor hazards if one needed to land a helicopter in an emergency response.

It is pretty intuitive that oblique views help with perception, but why? Last year I cited some work using MRI brain scans to determine how people perceive their world. The bottom line was that 3D models and oblique views speed the brain's ability to form a Common Operational Picture. Most likely, this is because our real world is an oblique environment. We need the visual cues of angles and perspective to form a quick and clear perception of what we are looking at.

So, what are the options? It depends on what you need to accomplish, how fast, and all of it dependent on your budget. There are three factors to consider: content, 3D model creation, and 3D model viewing and interaction.

CONTENT and MODEL CREATION

Data content and model creation go hand in glove depending on the source data and method used to create the models. The process can be as simple as extruded building foot prints to a defined “Z” value with no imagery or textures on the sides. Many GIS users have done this in ESRI environments such as 3D Analyst. Another popular method is Google Sketch Up, which permits users to manually create wireframes with textures or imagery pasted on the sides. Usually the imagery sources are either cloned textures or actual digital ground-level photos that are stretched to fit the buildings. This has helped Google populate their environment with many 3D models submitted by users worldwide. Many of the models look very good, but there is a lack of consistency and some of the models look imperfect.

A number of firms are now building 3D models using ground-level LiDAR. The models are extremely accurate and detailed, but lack the visual realism of photographic capture. However, the technology used to create the models is improving steadily and may be hard to beat in the near future.

The most sophisticated 3D models I’ve seen are from a small Florida firm called PLW Precision Model Works out of Melbourne. This company developed a semi-automated process to create 3D models very rapidly by taking advantage of geo-referenced and measurable Pictometry oblique imagery. Since the imagery is captured at oblique angles from four directions, each side of a building is imaged. Since each pixel is geo-referenced, the imagery can be used to mathematically build a 3D model through auto-correlation. Once built, the same imagery is then draped on the model with perfect registration. This scientifically consistent method results in 3D models with superb accuracy and uncanny realism at a cost lower than traditional methods.

3D MODEL DISPLAY and INTERACTION

Most users display their 3D models in traditional viewers such as Google Earth, CAD viewers by AutoDesk, or GIS software such as 3D Analyst or ArcScene.

A completely different approach is by Zebra Imaging, a leader in 3D holographic images. Zebra takes 3D models and creates a flat panel. When viewed under a special light, the buildings and other features seem to float up above the panel. If you’ve never seen one, they do have a wow factor, and I can see their value in some public venues. I’m not, however, convinced that they are refined enough or practical for emergency operations. You would have to judge that for yourself.

An equally impressive technology are 3D printers such as from Z Corp. These work by laying down layer after layer of a powder that is solidified during each layer pass by printing inks. When complete, the model is removed, and the powder that was not hit with ink remains a loose powder. Shaking away the loose powder reveals a solid 3D model that is also colored with the printing inks. The 3D models are somewhat durable and can be very intricate. The limitation is the 15-inch maximum size, the slow speed, and cost. These have become very popular with manufacturers to build product prototypes, but have seen limited use in the GIS community.

When I was with the Atlanta Regional Commission we created very rapid physical 3D models of buildings using our oblique imagery printed on our HP plotters, then laminated on foam core, cut, and assembled into a box. Although the imagery of each side had to be edited to be rectilinear and consistent in size when assembled, the process was surprisingly fast. The two-foot building model shown here took twenty minutes start to finish. We used oblique imagery because we had the luxury of a full metric oblique image library. The same technique can be accomplished using a digital camera at street level or higher.

SWAT teams and special operations people are especially strong fans of physical 3D models. In fact, news reports say the Bin Laden compound was completely reconstructed as a full-size 3D model that even used the same materials so the operators could judge the protection offered by walls and doors.

Many people are quite surprised to learn that TerraGo’s GeoPDFs are capable of displaying multilayer GIS data and interactive 3D models in a single document viewable with Adobe Acrobat. Note this screenshot of a 3D model that was created of the Bin Laden compound.

At the high end of the 3D viewers are products such as Lockheed Martin’s TopScene, which is a sophisticated 3D helicopter simulation software. At first glance, TopScene looks like a really good video game, but that’s where the similarity ends. This system is designed to react perfectly to controls and the environment, and unlike video games, the visual databases are real-world accurate, not just pretty pictures and cloned graphics. Since TopScene is used for actual mission planning, the models contain no fake or cloned imagery — only the real images or black no-data areas.

On the same level of sophistication but designed for ground-level work is a 3D viewer from Skyline. This viewer has tools that permit measurement in all directions. There are view shed analysis tools, flood plain tools, tools to show blast effects and radius, protected and unprotected areas, and many other analysis tools that fully understand and can exploit the 3D environment. Both TopScene and Skyline use oblique imagery and 3D models from PLW. They are so good that USSOCOM recently reduced the number of 3D viewers it supports from 25 to 2 — TopScene and Skyline.

But getting back to my original point. I like 3D computer models but they do take time, they are less accurate than the original source data/imagery, and the price is roughly 10-20 times more expensive than oblique imagery. My personal experience with first responders has convinced me that in many cases ground-level photos, oblique imagery, and physical 3D models can provide a Common Operational Picture that is just as good as 3D electronic models at a more practical price.