April 17, 2014

Confessions of a Hired Gun

Sam B Avatar Sam Billingsley

"Confessions of a Hired Gun" is just that, the true tales spun by a guy who's been in the field, scanning for a living, as long as anyone in the industry. What works and what doesn't work? What does the client want and what should the client want? This is a place where you'll find advice and commiseration if you're in the scanning business, and a place where you can learn about best practices and what to expect from your scanning provider if you're an engineering firm or asset owner/operator.

Want to pretend you are Tony Stark?

If you said, “Yes, I do want to be like Tony Stark,” I just might have the computer interface part solved for you. As for the Audi R8, the ever present supermodels and the whole Iron Man thing … you’re on your own. Many of us have watched with interest as the Microsoft Kinect was used as a motion controller for applications outside of the Xbox. However, if you have tried this you know that while it may be fun, it’s not as productive as a traditional mouse or anywhere close to a 3D mouse. The issue seems to be a combination of a lack of point density and low accuracy returns. While there have been Kinect knock-offs that have improved on this, the new Leap Motion Controller is in a class by itself.

According to the forum boards on LEAP’s website, the technical specifications are as follows: 

  • 0.01mm precision in detecting position in Field of View
  • Field of View (range) 0.2 cubic meters  

Unfortunately, that’s about it. By looking at some unboxing videos you'll find out that there are three sensors and that the release version will be version 6. Early versions were just a couple of raw circuit boards without a case. In these you could see two lens like sensors and three LEDs being used to create the sensor field. For a full demo by the CTO of Leap, check out this video posted by tech site The Verge. The demos are quite impressive; especially for those of us that are already accustomed to working in a 3D environment. The best news is that, delays aside, the product does not appear to be vaporware. 
 

LEAP ControllerLEAP Motion is the new name for OcuSpec, a startup that raised more than $14.5 million from 2011 to 2012. So, funding doesn’t seem to be a problem. While they have announced that they will have an App Store (or whatever they will call it so that Apple doesn’t sue them out of existence) for LEAP specific apps, the design of the motion controller is stated to be fully cross platform and OS agnostic.  In fact, they have been sending out units to developers since May 2012 in order to front-load the app store with third party apps.  That should open up a wide user base. 

LEAP tweeted last month that Autodesk was making plugins for the motion controller and said it would begin shipping pre-orders the week of May 13. Best Buy has announced that they will have units for sale the week of May 18. Best of all, units are listed as retailing for $79.99.

Now, about that Audi R8 …


Permanent link

I-Site 8800 Field Test

I have had the privilege and in most cases the pleasure of working with most every major 3D laser scanner on the market over the past decade. Despite knowing John Dolan of Maptek for most of that time, the I-Site scanners remained a blight on my record. I think that the fact that Maptek markets themselves so heavily toward mining has something to do with this, but to be perfectly honest I had always viewed them as sacrificing accuracy for range. Not necessarily a bad choice, just one that did not align itself with my client base.

Earlier this spring, I finally had the chance to use an I-Site 8800 in the field thanks to Chris Little of Maptek Australia. We were capturing a racetrack for a video game developer and for once the accuracy required was in the high centimeter range as opposed to millimeters. I had not planned on using the 8800 but three mornings of heavy fog along with an aggressive track grade put us a bit behind schedule. In the end I was extremely happy to see Chris and the 8800 on-site!

ISite8800

The rig consisted of an I-site 8800 (also sold as the Leica HDS8800) mounted on the roof rack of a Toyota FJ Cruiser. The unit is controlled by a Toshiba tablet PC running Maptek I-Site Studio. There is an internal camera for RGB capture and most of the settings for point density are easily selected via menu presets. For this application we used a GPS/RTK unit and took a shot on top of the scanhead and then another on the bull bar of the cruiser. By locating that same spot on the bulbar in the point cloud you assign your back sight position and thus roughly geo-locate each scan. I was concerned that such a short baseline would cause accuracy issues but Chris assured me they could be resolved through post-processing.

I only got about 10 days with their office software package (I-Site Studio 3.5.1) but the work flow is quite a bit different from the Leica Cyclone package that I use most often. I was not able to produce registered datasets with the level of accuracy that Chris was able to achieve, but I did find the user menu to be very helpful (with the exception of noting that selections for “drag & drop” functionality require the scroll wheel button!). The application allows for most of the standard export formats (no PTX or e57) and has a wizard for creating custom export fields for ASCII files.

Overall I liked the system a lot. The negatives are few ($250,000 price tag, lack of 3D view during scanning with tablet) with quite a few positives. The added range was fantastic, but the biggest benefit from this range increase was the high intensity return on black asphalt. The work flow in the field was very familiar to anyone that has done any land surveying and fairly quick given the vehicle mount. While Chris achieved a good registration, if I found myself using the system again I would set the back sight up with a much longer baseline to increase the accuracy of the initial registration. I really like some of the visualization tools like being able to automatically assign random colors to each point cloud during registration. This makes spotting misalignment much easier. I also liked the way that you could select minimum and maximum ranges (relative to each scanner position) to be used in cloud to cloud registration calculations. If you get a chance to play with one of the 8800s, I  would definitely take the time to do so. I’m of the opinion that every scanner gains some attribute at the expense of another. We all bought/rented/borrowed the scanner we have with the idea that it hit that sweet spot in terms of attributes that best fit our needs. 

However, it’s nice to try on someone else’s idea of that sweet spot on occasion!


Permanent link

Layar Creator

As some of you may know I am crazy about Augmented Reality. I have been developing for the Layar platform for the last two years, but this past week they really took it up a notch with the release of Layar Creator. This is an additional tool that has been added to the Layar A.R. Browser that accesses A.R. content via image recognition as opposed to spatial placement (i.e. Lat/Long). I have a lot of clients that have been waiting on a way to easily attach metadata to an object without needing GPS data or a QR code or anything else 21st century in view. Best of all it is currently free! After the initial launch, Layar will begin charging for premium content but they will continue to offer a free version that is ad-supported. 

Layar Creator Image

To give you an idea of what Layar Creator can do I used the mouse pad that we gave away at SPAR 2012 as an example. If you did not get one you can download the image and still try out the app. First of all you will need to download Layar (free) from the App Store or Google Play to your android or iOS device. Afterwards, launch the application and point the device’s camera at the mouse pad or image. Use the “Tap to View” button in the center of the screen and the app will perform an image recognition. Once the image is recognized it will download and display the A.R. content.

Currently, you can add buttons that link to web content, Twitter, email, Facebook, Pinterest, e-commerce, and YouTube. Best of all, Layar provides use statistics for each campaign similar to Google Adwords. You can see the number of unique users, total page views per day, content clicks, and the country each user is in. Another point to make is the backward compatibility of the application. Seeing as Layar Creator was just released last week, I did not plan to use the mouse pad that I gave away at Spar in this way. However, all you have to do is upload the image you want to use and those images then are recognized regardless of when they were created. This makes all sorts of legacy objects good candidates for A.R. content. 

Now the caveats. I’ve tested this with several devices (iPad 2, iPhone 4s, Droid X2, Droid Bionic, Samsung Galaxy Nexus), but obviously not everything out there. Even within these I noticed quite a bit of variation. The email function does not seem to work on iOS devices, the video plays in AR mode on iOS and Motorola phones but only in full mode on the Samsung, and I’m certain there will be more issues with other devices. Also data speeds can be an issue. WiFi works best, followed by 4G then 3G.

According to the developer forums, Layar is planning to add support for 3D models in an upcoming release. Due to bandwidth considerations they will be limited, but the fact that you can link back to any web address really opens up some options for tying datasets to 2D documents. I’m already working through a long list of possible deliverables using this technology and I hope you will take the time to check it out for yourself. The more users we get the better the company does and the greater the feature set becomes! 


Permanent link

Deliverable Concepts Part 3 (AR all the way)

For those of you that missed the last two posts here's fair warning that this is part 3 of our 3 part series on deliverables. In case you want to catch up you can read Part 1 or Part 2 first. We'll wait...

So, here we are at one of the most interesting deliverable types in the industry: Augmented Reality (AR). In fact, this is interesting enough to have become an industry unto itself! This is great news for all of us creating virtual environments through 3D imaging hardware and software of all kinds. Because AR is taking off as its own industry we are seeing similar advances and cost reductions as the video game industry. More users equals lower cost for those users. Add to this the fact that many AR platforms are using off-the-shelf hardware such as smartphones, webcams, and tablet computers and we have the makings of some very powerful viewing opportunities without six figure investments in a viewing system.

AR Stairs
 
3DS model projected on a marker with the AR-media plug-in.

To my great disappointment, I have not yet found an AR system that allows for the import and display of point clouds. As a result, we are once again forced to create models. Many of the technical details required to get modeled objects into an AR platform are the same as those that we discussed last week for video games. The key is to reduce the model size as much as possible for faster rendering and data transfer (especially if it is moving over a cellular network). Additionally, you will find that some AR applications require the models to be encoded in a proprietary format. This is typically accomplished through a Java applet or BAT file. It's rarely difficult, but be prepared to spend a fair amount of time reading through the Wiki associated with the AR platform in question.

Let's look at a few of the platforms currently available.

  • LAYAR: Layar is an augmented reality browser that allows smartphone users (iOS or Android) to load Layar layers and view them through the camera on the smartphone. Layar also has a player that can be used to insert AR media into other applications. For those of you that caught my AR presentationat SPAR2011, you may remember that the biggest complaint was that the placement of objects was relegated to the accuracy of the phone's GPS positioning ability. However, since then Layar has introduced Layar Vision. Vision is essentially a photo recognition version of Layar. You upload an image and associate it with a model so that viewers of the image automatically gain access to the model, hotlinks, metadata, etc. Best of all, unless you are getting more than 10,000 matches per month the service is free!
  • AR-media: Inglobe technologies makes the AR-media platform for developing AR applications. My experience has been with the plug-ins that they make for multiple applications including 3DS Max, Google SketchUp, Nemetschek Scia Engineer & Vectorworks. For this example I'll refer to the 3DS plug-in. AR-media provides a PDF of a marker that you print out. You load a model in 3DS and launch the AR-media plug-in, which turns on your webcam. When the camera sees the printed marker it places the model on the marker in the camera view on your monitor. Here's a video of me using it, forclarity. The cool thing about this is that animation in the 3DS file is included in the AR view. The full-service application allows for stand-alone viewing of AR content without having to have the modeling software that was used to author the model.
  • Total Immersion: This platform has both a Flash plug-in and their flagship D'FusionSuite. The suite allows for much more control of AR media and export to multiple end user devices. While I have less personal experience with this platform, I can say that they are set up for more general use than the previous platforms. Compared with the other platforms, D'Fusion is similar to a video editing suite in that you make the scene and then render it out for various end user devices.
  • Metaio: I have not used any of Metaio's authoring software, but I have used their mobile platform, "Junaio." My understanding is that Metaio has a larger following in the advertising industry and is the format of choice for some large advertising firms. However, their Engineer platform is of great interest to me as the model sizes are much larger than most AR platforms support.

Please understand that this short list is, by no means, complete. There are many more applications and the development is faster and more varied than the 3D imaging market! It makes it difficult to keep up but very exciting at the same time. I encourage you to keep up with AR development even if you are not producing AR deliverables at the moment. There is no more natural an environment for accessing data via a NUI than the real world that we live in everyday. 


Permanent link