To promote reading of the paper in advance of the presentation, add a comment after your name discussing the paper, supporting or objecting to part of it, describing a new citation that confirms/challenges the arguments, etc.
Blue, Ryan DavidDaken, Abigail Ann
I've used 3D displays in the context of engineering design, which I guess is in the realm of scientific computing (as opposed to info viz). There, the reason to use 3D was to get an idea of what an object that did not yet exist would be like if it did exist. For that, 3D is essential, as noted in the paper. Some controls that made it more usable included:
- Very flat command structure, with a multitude of toolbars making commands available with a single click. User customizable.
- A series of commands to control the view, including 1-click access to top, bottom, side, front and back views. Also 1-click commands to move between wire-frame and solid rendering. Partial transparency was available, though harder to access. Generally, it's use is to make individual parts partially transparent as a way of understanding how they interact with other parts.
Even with all this support, there are individuals who specialize in the use of these tools, who are at least twice as fast at building models as a reasonably experienced non-specialist, such as an engineer. Viewing a model is easier than building it, but still takes practice to keep oriented. Many of these tools would be improved by an overview to keep people oriented when zoomed in, and understanding the 3-D rotation controls are a problem for every user I've ever met.
I've also used 3D surface plots to understand moderately multidimensional data, such as spatial data over time. Occlusion is a problem, but the value of being able to visualize the dependence of the data on more than one variable was helpful enough to make up for that. It can suggest multivariate functional forms, for one. Also, my internal understanding of continuous multidimensional data is of n-surfaces in high dimensional spaces, which makes 3D visualizations "natural". That internal representation comes partly from living in a 3D world and understanding physical objects within it, but perhaps if I had grown up with more effective tools for visualizing high-D data my internal representation would be different. Also, as Inbal mentioned, having interaction is essential - it reduced the occlusion problem significantly.
It makes sense to me that the social & gaming sites would be where 3D has been successful. Many of our most important social reactions are keyed to experiences in a 3D world. For gaming, part of the game is occlusion - it wouldn't be fun if you could see everything!
Dogruel, Eylul Do you think that a sculpture is a lesser than a painting because one can never view it as a whole at the same moment. Is classical painting less of an art form because it doesn't have literal depth and just an illusion of one? Why are we insisting on comparing apples and oranges?
What do we mean by 3-D exactly? Krist pointed out an important fact: our ability to focus on a layer, rather than taking our 3-D view as a whole. It seems like we expect from 3-D technologies to show us all range of all of the 3 dimensions at once and then complain when we are unable to deal with the density of information that comes toward us. We blame occlusion when we are equally unable to view all sides of a model at once in real life. Then we compare all of this to games, where 3-D is defined as being able to navigate spatially in order to explore through many 2-D images. First one defines the image as 3-D, second one the conceptual model. Not to mention the fact that all of the interfaces noted up to here are all based on a 2-D screen surface and is not even considered to be 3-D for another definition. Yet another aspect of 3-D goes unnoticed: the 3-D environment doesn't have to be created out of one component. Isn't an office with multiple screens and equipment all connected through a network a 3-D environment Why not? You interact through it daily. Is there any reason why it could not be potentially used to visualize data as a whole?
Maybe we should consider treating 2D and 3D as two mostly independent aspects of an interface rather than dividing ourselves into camps. Why bother labeling as if we are really chosing one versus another?
Krist mentioned earlier the difficulty of navigating 3D environments using devices built for 2D interactions. While haptic feedback devices and many other 3D interface devices are usually well out of the consumer price bracket, some are already commodity products. For example, the NaturalPoint TrackIR allows for 6 degrees of freedom tracking of user head movements for under $200. There is also a proprietary SDK available for game and simulation developers to incorporate TrackIR input. (On a side note, because of problems with tendinitis, I've been using a similar 2D product, the Origin Instruments HeadMouse Extreme for some of my mousing work.)
Cheaply available 3D interaction devices could usher in a new standard for 3D interfaces, replacing complicated mouse and keyboard shortcuts with things like moving your head left, right, up, and down to pan, tilting your head to rotate, and moving your head forward and backward to zoom. Naturally, the limitations of the devices and software will be an initial roadblock, but if these devices become more widely accepted the interfaces will follow.
I found another interesting link exploring the FSV page Michael linked to: Light Speed! is a 3D representation of how special relativity affects the appearance of moving objects.
Fuchs, Adam Phillip
Literacy for most of the world means the ability to read and write, generally on a two-dimensional surface. We learn from an early age how to deal with flat representations of information, so we are naturally biased towards 2D interfaces. However, our eyes and brains have a natural ability to process 3D information as well or sometimes better than 2D information. We're only just starting to build 3D languages, as technology is only beginning to support 3D graphics at a reasonable resolution and frame rate. I think it's possible that 3D information visualizations will be much more intelligible after years of our learning 3D languages. We shouldn't discount 3D as less useful for information visualization today, at least until we've had time to build good 3D languages.
3D interface is thought to be interesting looking, helpful and supportive. 3D representations are definitely useful in many computer-based/reality tasks - some familiar examples to me are architectural design, virtual medical medical help, and displaying molecular 3D structures for education.
One very important application of 3D is in entertainment, for example, games. Surely many game players love the 3D trend. 3D would be very impressive in stereo scences, playing lighting effect. However, in many cases, I believe 2D representations is good enough. 2D gives more details while 3D does not give sufficient details. The speed/display/control of 2D are also more smooth. The 3D rotation always makes me feel dizzy.
I would prefer 2D in many other situations. That is, 2D interface would be my first attempt to simplify a problem. On the one hand, 2D interface is more familiar and easier to control. It might take more time to understand in a 3D visualization, although it looks really cool. It's likely to get confused/lost if the design is not well planned, for example, 3D navigation through a website. On the other hand, 3D desktops etc. are more like a toy to me, cool but not that practical. The 3D interface is still young, if the guidelines for 3D designers are followed, we can see a more promising future of 3D.
I noticed one thing is not mentioned, 2D and 3D motion pictures. 2D motion pictures are a simplified and bombastic form of telling stories. Comparely to photos or normal films, it captures people's attention, especially children.
With advanced of computer graphics, producers tried to make few 3D objects projected in 2D motion pictures. This made a trade off between computation and cost. It is also like the 3D button in a 2D computer desktop environments.
While computation cost going down, true 3D motion pictures become realistic, such as Toy's story.
However, if you ask me which motion picture is pure 3D or not, I really don't know. It implicates that we semi-3D may as good as true 3D.
In Gaming, before pure 3D game, 2D can cheat human eyes to have by using multiple layer scrolling. Ex. Japanese Game 真魂斗羅 We can regard this as another example of not pure 3D.
King, Kyle Jameson
Knudsen, Kenneth B.
As I read this, I couldn't help but think of that extremely painful scene in Jurassic Park where the girl uses some kind of weird 3D flying browser-thing to "hack" a computer system ("This is a Unix system ... I know this!"). I dug around a bit (read: "simple Google search") and apparently there's an open source version of the SGI program used in that scene called FSV. That aside, it seems to me that the fundamental problem with all this 3D stuff is that the monitor is not really a 3D device. There will always be a loss of some kind during the 3D-to-2D translation, and I'm still doubtful that there's any novelly useful application of 3D navigation outside of games. I look forward to being convinced otherwise at the presentation.
-- comment Jsutin Grimes http://www.youtube.com/watch?v=bhOk2H2Mv6U <- is that the scene you were talking about
- Yup, that's it. I just love how she "identifies" Unix based on the 3D visualization--as if it was a standard part of _any_ Unix-based OS. Lam 21:14, 13 April 2008 (EDT)
-- comment Cody Dunne: One screenshot of FSV looks like a treemap!
Some additional thoughts on why I think 3D interfaces are less suitable for many tasks (than 2D).
1. It increases the complexity of the scene or interface.
It adds an extra spatial dimension, making the rendering more complex. In order to make a scene 3D, some information of the third dimension must be visible, thus using more screen space. This makes it difficult for 3D interfaces to support complex interfaces.
2. It also requires extra thought on how the interactions should be performed.
This is mostly because both the display and interface (usually mouse), are based on 2D. The user must 'learn' and 'think' more about how 2D interactions map to 3D interactions. The paper mentions navigation complexity, saying that a fight over number of clicks seems more reasonable than 2D vs 3D, but such extra requirements add to the navigational complexity, and should not be ignored.
The paper mentions that 3D interfaces are successfully used in 3D games. This is true, but even in such games, some kind of 2D interface is usually used to perform actions that require dealing with organizing data, and are routine jobs. (shops, inventories, auctions, mail systems)
Lieberman, Michael David
Lotze, Thomas Harvey
Mir Rashed, Fatemeh
There is no question that 3D reality presentation is the winner where the physical objects that are visualized need to be understood in their “real and solid” form. Some instances of such applications include computer aided surgery, medical imagery or tomography, scientific visualization of the shape of a molecule, and virtual touring of a building or architecture. In addition, 3D visualization comes to the top where creating/presenting “occlusions” and “ambiguity” are essential to the core such as in video games.
However, as stated in the paper, there are many applications or interfaces where 3D representation versus 2D not only is not any value add but also in many cases can slow down user performance in terms of speed and/or accuracy in comprehension of the reality.3D prototypes of air traffic control systems, digital libraries, or file directories, as we saw, were some examples of such 3D usability failures.
But why despite the fact that we live and perceive our surrounding world in 3D (or 4D including sound), this format of visualization fails to outperform the conventional 2D projections? Part of the answer might be because the screens are 2D and therefore our interaction devices such as mouse and keyboards are primarily designed for 2D motions and not 3D navigations.
Olea, Andreea Victoria
I agree with the central theme of the paper - interfaces that attempt to mimic reality make life difficult for the user both visually and in navigating through the interface. Rather, good design principles can be used in making 3D displays better than reality, thereby making life easy for the user, even if reality is not accurately depicted.
One of the factors involved in developing 3D displays is that a considerable amount of effort is required in doing so. Therefore, the benefits obtained from 3D interfaces should be weighed against the time and efforts required in developing them. In the fields of surgery and architecture, the third dimension proves to be extremely useful. In contrast, consider the example of a digital book library - a 2D visualization is probably as good (maybe even better) than a 3D interface.
In conclusion, while 3D interfaces should be made better than reality, it is important to consider whether 3D is required in the first place.
Schulman, Aaron David
Straszheim, Troy Donald
In my opinion, 3D visualizations might not be useful for information visualization; however, they can be effectively used in the scientific applications as well as designing cool and entertaining interfaces like games. As we know, information visualization is meant to provide the users with the dataset information in a simple and fast manner. I believe 3D visualization and sophisticated graphical features may not necessarily improve the way that a dataset is visualized; however, they can make the visualization look cool. In most of the cases, a simple 2D visualization can convey the information faster and easier than 3D one. A good example is a 3D computer desktop which conveys no more information than what a 2D desktop provides us with. Although it looks cooler than a 2D desktop, working with it is more complex (since for example it is difficult to read sentences on the floor and ceiling).
On the other hand in scientific applications such as medical imagery, 3D visualization can enable scientists to look at complex surfaces and analyze them. For example, there is no way other than 3D visualization to observe the shape and volume of a tumor in the brain and look at it from different angles. Moreover, a stack of 2D magnetic resonance images (MRI) of the brain cannot be as informative as a 3D view of it.
VanDaniker, Michael Robert
As several people have already pointed out, 3D suffers because monitors are 2D. Data can take on multiple dimensions, and the human mind has proven that it is capable of handling 3D data with ease (driving would be quite an obstacle otherwise). Data structures have been developed that help computer's efficiently process 3D structures. Both entities on either side of the monitor, the human and the computer, can handle 3D data, but the monitor itself cannot. The computer has to apply a matrix transformation so the monitor can make sense of its input, and the human brain has to translate the pseudo-3D image on the monitor into an actual 3D representation. Going from human to monitor to computer is even worse because the translation that the human needs to perform can't be done automatically. The user has to think about how to interact with the computer's peripherals to create the desired effect.
Nicholas Chen's work on e-book readers presents interesting solutions to this translation issue. The hardware responds to the user's movements, so when the user makes a page turning motion, the computer knows it should turn the page. The act of turning a page when you reach the bottom is almost natural, so the user won't feel like he's using a computer. E-books aren't 3D, but if advances in hardware can make interacting with 2D data feel natural, perhaps similar advances for 3D visualizations can make interactions less clunky.
Hyunyoung Song's work in this area also shows promise for 3D. In her project the user makes markings on a 3D object in the physical world and those annotations are translated by the computer into operations to perform on a virtual copy of the object.
I would separate interface into content part and control panel part. Content is the main thing being shown. For game like Doom, content is the simulation of how you are walking around, carrying rocket launcher and looking for demon. Control panel are controls and information on the screen, for example, what is your gun, how many ammo left, hit points, menu button, etc.
I agree that 3D is being used successfully in some kind of tasks, for instance, game or simulation environment. 3D can do a great job for the content part when the content is reality. Game or simulation environment are trying to make their contents look real so it is a perfect match.
In other kind of applications where reality is not the goal, I still think 2D is more suitable than 3D because of the following reasons.
- Our eyes can see 3D but when we want to focus on something, we still focus on only one plane. 2D is better for people to keep focus on.
- I agree with Joonghoon that most popular interaction devices are in 2D. For example, a mouse, it can move only in x, y plane at the same time. Navigating 3D with 2D device makes brain work harder to bridge the gap between 2D coordination and 3D.
- When using 3D, you cannot use all (x,y,z) position at the same time. However, when using 2D, you can use every (x,y) position on the screen to display the content without worrying about occlusion. It is easier to utilize screen space efficiently.
- Those applications provide 3D content but they still use 2D control panel. Seems like when it comes to interaction, 2D interface is still preferred.
In general, I agree with the concept of the paper. 3D visualization is very useful in many applications, especially when applied to 3D objects (i.e., not multi-dimensional data). I think the main takeout of this paper is noted in the last section and I quote: "provide overviews so users can see the big picture".
Few comments about the paper:
- In page 1 and 2 there are several comments about applications that are more intuitive to use, or friendly, with 2D visualizations (e.g. online libraries, desktop). I believe the reason for that we are used to navigate in 2D spaces, as we do it on a daily basis. Apple offers a very nice 3D CD library, which do not resemble reality, yet much more intuitive, in my opinion, than the standard 2D visualization. (I used to teach elderly people the basics of using MS office tools. Those are NOT intuitive tools at all, to an unpracticed eye)
- I think the goal of the visualization plays a main role. I would not use 3D visualization to explore or mine data, yet I would prefer using it for entertainment (e.g., games) or exploration of 3D objects (e.g., x-rays)
- Also, I believe 3D visualization is meaningful only when it is interactive.