U.S. Highway Routes, Subway Maps and Magnification
The Numbered US Routes as a Subway Map (by Cameron Booth)
Portland-based Graphic Designer Cameron Booth has produced a very nifty rendition of the US Highway system, shown in a style similar to the London Underground and other subway maps.
I am seriously considering buying a copy of the poster just so I can spend some more time searching through the details in this work of art. It is difficult to make out all of the details from the higher-res online version (linked from the image here), however some patterns are certainly evident.
I may be biased, but to my eyes Chicagoland certainly stands out as the major transportation hub of the nation. In general there is a lot more action in the nation’s midsection than on the coasts (especially the west coast). The map certainly makes it appear that “fly-over” country has a pretty rich network of roads and truck stops. Kind of the opposite effect of the famous “New Yorker’s View of the World” images that show Manhattan island as a huge shape filling out more than 50% of the national map. This map is the latest in a series, the designer has many interesting detail images and description of the process on his web site. If you like maps and/or infographics I recommend spending a few minutes at the site to check out his work.
Selective magnification of an updated DC metro map. Original map design by Cameron Booth.
I was also interested to see Cameron Booth’s redesign of a map for the D.C. Metro System. Many years ago I had worked on a focus+context tool for allowing selective magnification of the old D.C. Metro map according to where the viewer is in the system. The resulting image was probably my most popular one ever from that period of my research, and got republished in quite a few journals and books. Just for fun, I loaded Mr. Booth’s updated map into my PhotoXform iPad App for nonlinear magnification to see what it would look like. The results are as you see here. While the magnification is effective in showing local details, the presentation could be enhanced by treating label size more independently of the spatial magnification function. That would allow the labels to be readable throughout the image, and simply spaced out better (or de-cluttered) in the region of interest. In addition, although the radial magnification function has a nice correspondence to the fisheye lens concept, it also disrupts the orthogonal line placement that was no doubt an intended feature of the design.
Orthogonal Selective Magnification of DC Metro Map (original design by Cameron Booth)
To address this last issue, I tried revisiting the map with an alternate magnification function from PhotoXform. This time I used a transformation that would preserve the orthogonal line layouts in the original design, and produced a very natural looking map. The selective magnification effect is so non-distorting as to be almost unnoticeable, yet at the same time it does provide significant resolution enhancement in the region of interest. I would love to do an user study one day to ascertain if this type of presentation is as intuitive to the uninitiated viewer as it is to my expert eyes. My hunch is that this type of selective magnification (perhaps with some additional subtle visual cues) could be understood without additional explanation.
If you have thoughts or comments on this, I’d love to hear them. You can leave a comment here, or connect via one of the channels at the top right of this page.
Lytro Light Field Cameras
Photographers and computer graphics geeks alike have been looking forward to the release of the Lytro Light Field Camera early this year. I pre-ordered mine last year and am hoping to get it in March if everything goes according to plan. The light field camera (also sometimes referred to as a plenoptic camera) has the potential to open up a new generation of techniques for capturing, viewing and exploring images.
Tracing the direction and intensity of light in an artificial computer generated scene has been a well researched topic over the years, involving such techniques as radiosity and ray tracing. These techniques have been used to drive advances in realistic computer graphics that are now routine to see in video games and movies. In essence the light field camera brings these theoretical computer generated graphics techniques into the real world and “reverse engineers” a captured scene by recording both the direction and intensity of the incoming light across the entire sensor array.
From my initial read of Lytro CEO Ren NG’s Ph.D. Thesis, the new technology creates a 4D mapping of the scene as a set of “ray pixels” holding light intensity and incoming direction vector at each sensor location. Through software treatment of the sensor data it is possible to then change how the scene appears in a 2D image. The most frequently given examples shown for interacting with the scenes are by changing the focus points and creating 3D type effects. This technology represents a fascinating fusion of photography and computer graphics, and I am sure that I will have much more to write about this technolgy in the future.
For now though, I am interested in the consumer perspective for this new technology. Lytro seems to be focussing (hah!) on a fairly broad range of customers at this point, with a simple to use device that can be enjoyed by complete novices. The device appears to have relatively low resolution compared to DSLRs and even consumer compact cameras. That makes complete sense for their current startup status; once they get the first version out the door and the revenues start flowing in then they can work on creating tighter sensor array densities and higher image resolutions.
From a photographer’s perspective the Lytro camera offers a huge advantage in not having to attend too much to finding the best focussing distance for the shot. In addition, the camera does need not to spend time autofocusing before the shot, which is a big cause for the slowness of consumer cameras with their slow contrast detect focussing technique. From the examples shown on their website it appears that the device is able to capture a fairly large depth of field in a single shot, but not an infinite depth of field. This has the potential to be a boon for certain types of photographers who require a fast reacting and highly mobile camera, street photographers come to mind immediately. This could potentially also be useful for other types of photography where a large depth of field is desirable, such as landscape and architectural photography. I think the usefulness in these latter types would be more limited however, as the use of tripods and smaller aperture settings with longer exposures can already provide a fairly large depth of field. It is pretty easy to simulate out-of-focus areas in an in-focus image using Aperture or Photoshop.
After the scene is captured, the interesting question becomes what are the cool and useful ways of allowing the viewer to see and interact with those images. ”Cool” and “useful” aren’t always the same thing when it comes to new technology. The Lytro site has several flash examples allowing the user to click on regions of an image and have the focus point shift to that part of the image. It isn’t completely clear to the outsider what is happening under the hood, but it appears that the interaction is telling the software to choose a single focus plane from the scene and render the image with that plane in focus. My guess would be that it uses a contrast detect algorithm to find the plane of maximum focus in the 2D region of the touch, and then apply that focusing plane to the entire image.
This is really cool to watch, but is it actually useful? Lytro is making a big effort to make it easy for people to view and share their images online, pointing out that the Facebook community is projected to publish 100 billion (thousand million) photos online in 2011. Clearly online is “where the eyeballs are” so it makes sense to go after that market aggressively, but will the average facebook member — faced with dozens of photos in their stream every day — really want to have to click and explore individual photos to get the most of them? Pushing focus presentation choices out to the viewer empowers them, but with more and more online photos to view and attention spans even shorter on-line than in the real world, there is a potential conflict. Once the novelty wears off there is a need for a value proposition that makes it worthwhile for both the producer and consumer of the scenes to bring the additional complexity to the viewing process.
In my opinion this is the truly critical area for Lytro to address, and the richest opportunity for creating a paradigm shifting sharing and viewing experience. My guess is that the best value propositions will arise not from allowing viewers to interact directly with the field data to shift focus fields around, but rather from intermediate software layers that allow creatives and producers to produce semi-guided presentations of the scene to the viewer. For example an interactive ad creator could use animated focus shifts to guide the viewer through a sequence of impressions in a single image in a subtle fashion. The need that I imagine for guided tours through the 4D light field dataset is somewhat analogous to what we saw in the graphics and visualization community in the 1990′s and 2000′s: users given free reign to navigate unconstrained through virtual 3D spaces could quickly become lost or fail to observe the critical features. A variety of constrained navigation techniques were introduced to allow novice and expert viewers to more easily move through the 3D worlds. I see a similar set of techniques arising to deal with the 4D light field camera sensor data.
It remains to be seen whether Lytro’s light field camera will be just a cool gadget or something more enduring in the market. My guess leans more towards the latter, but there are substantial software and marketing hurdles to got over before that will happen. My hope is that Lytro will open up a Developers SDK for their file formats that lets them tap into a rich community of developers on different platforms. They have indicated some tentative support for this idea, but nothing is on the table yet. Of course when these light field principles are applied to video, a whole set of ripe opportunities opens up, in addition to some formidable engineering challenges. I’ll write more about that in a future post.
One thing is for sure, this is going to be an interesting few years as the technology enters the market, and I will be following it fairly closely. Please select one of my feeds from the “Stay in Touch” area at the top of this page for more updates and examples when I get my own copy of the camera to experiment with.
What Exactly is a Holistic Sofa?
Excellent Question. The answer has many layers. Let’s peel back one or two of them.
Holistic Sofa has been an evolving concept for me over the past 20 years, when my interest in computer graphics really kicked in. I’d played around with graphics on computers before then, including programmed text and sprite graphics, and QuickDraw on the Macintosh. Things got more serious for me when I took Bill Kocay’s Advanced Linear Algebra class in my junior year at the University of Manitoba. I had a good aptitude for math as a young man, but not much passion for it. Dr. Kocay is an excellent teacher however, and his enthusiasm got me kind of enjoying matrix manipulations. A month into the class he showed how one of the matrix operations he had been teaching us could be used to create a perspective projection from 3D space into a 2D computer display. To this day I remember him standing at the overhead projector twisting his hand to illustrate the transformation and thinking to myself “OMG, this is so cool!”.
OK, I wasn’t actually thinking “OMG” literally, because texting wasn’t even around back then, but you get the point.
You’ve got to understand that back then there was no OpenGL or graphics cards on desktop computers. Heck, my Mac Plus didn’t even have a floating point processor. There were some primitive 3D graphics in games, but they were all hand-rolled deals (Microsoft Flight Simulator is maybe the best example, Doom and Wolfenstein 3D didn’t exist yet), and there were no APIs of any sort to help program 3D graphics on your machine. I drove home quickly after class and worked late into the night (actually, early into the morning) trying to create a 3D rendering of a simple tetrahedron using my new linear algebra to convert the 3D data model into 2D QuickDraw bit drawing operations on my Mac Plus. When I finally got the little 4-sided shape showing up on my screen it was a magical moment for me. I was fully and completely hooked on computer graphics, only I didn’t yet realize it fully.
“Fascinating”, I can almost hear you not saying to yourself, “but what’s all this got to do with a Holistic Sofa?”.
The following weeks were filled with many late nights as I fine tuned the code so that it would run efficiently enough to show an animated tetrahedron rotating and bouncing through 3D space, then multiple tetrahedra, and eventually even cubes. I spent hours writing integer based lookup tables for the calculations so that I could avoid the use of decimal points that would kill the interactive performance on my Mac. Eventually I released a shareware FKEY (remember those?) screensaver to the Mac community called “Cubist“.
Cubist was well received and a lot of fun to make, but I wanted more. More geometric complexity and more meaning. Tetras and Cubes were too generic. There were no online 3D model markets like TurboSquid back then, all of my 3D models were painfully constructed by hand entering the 3D coordinates for the objects in text. The chances of my being able to construct a decent model of anything interesting by hand were about the same as the chances that Joan Jett and the BlackHearts would show up at my birthday party to play I Love Rock and Roll.
I didn’t realize it back then, but I was encountering one of the most difficult and persistent hurdles that occurs throughout the computer graphics and visualization fields: access to good data or models. This has been a constantly recurring issue for my colleagues and I over the years. Sample or toy models are clean, simple and boring. Interesting models are difficult to find and import, filled with exceptions that require special treatment and a general pain to deal with. That’s not 100% the case of course, but it is generally true.
This was the start of my long, dark search for a more interesting 3D model that I could render on my Mac Plus. It had to be easy enough that I could actually stand a chance at hand coding the coordinates, but also have some intrinsic interest greater than a simple geometric shape. The answer came to me while reading Dirk Gently’s Holistic Detective Agency by Douglas Adams (I was a huge fan of both his Hitchhikers Guide to the Galaxy trilogies). There’s a recurring theme in that book where a moving crew gets a sofa stuck in a staircase, and a computer tries unsuccessfully to find a way to get it unstuck. A bolt of lightning struck me right then.
Holistic Sofa Screen Shot
Within a few days I had created a blocky looking sofa model and had it spinning around on my tiny little 9″ grayscale Mac screen. I dressed it up a bit with some rotating number counters, a crude staircase and a huge rejection “X” that would pop up from time to time to signal failure. Holistic Sofa was born! It was somewhat mesmerizing and mysterious to watch. Friends would come over to my house just to drink beer and watch the screen saver do it’s thing on my Mac. Holistic Sofa was eventually released as a shareware After Dark screen saver in 1992. I had managed to find Douglas Adams’ email address and sent him a note asking if he was OK with the release, and to my great surprise he replied not just once but twice to give it his blessing and ask for a copy for himself. Getting a personal email from my Sci-Fi author-hero Douglas Adams was definitely on the same level with having Joan Jett play at my birthday party!
Holistic Sofa at Can You Digit
Over the next several years the Holistic Sofa screen saver became a sort of a cult hit in the Mac community; and in 1996 I was contacted by Tamas Banovich, who wanted to display Holistic Sofa in the upcoming Can You Digit digital art exhibition at his Postmasters gallery in Soho. You can see it in the image here being projected on the wall above the other displays, to the left of the column. These images are the only ones that remain from Holistic Sofa. The program no longer runs on modern macs. A few years ago someone posted a YouTube video of Holistic Sofa running on his 7600 Mac, which is as close as you can get to running it on your own machine these days.
Getting back to the original question then; Holistic Sofa marks the beginning of my journey into the fields of computer graphics and visualization. The Holistic Sofa graphics were primitive by present day standards, however they still resonated with a diverse community for decades beyond their creation. Over the past 20 years I have spent many years studying, conducting research, creating products and building companies in pursuit of the interest that was kindled during that period. That interest has also broadened into many related fields, about which I will write more later. From time to time I have gotten side tracked on non-graphics oriented work, however I always find myself coming back to making visual images with bits and pixels. It’s what I love.
More broadly, Holistic Sofa (Blog Edition) is my new platform for exploring advances in graphics and visualization and sharing them with the gentle reader. Part of this exploration will involve sharing products and techniques that I have developed myself, but it will also include pointers to new technologies that I see coming out on the horizon.
Hello, World. Hello, Me.
Welcome to my new website!
< long silence >
Of course, there’s no one reading yet. So I’m really just welcoming myself to my own website.
< yippee! >
Stay in Touch
