Peephole Displays

benh57 writes "A student at Berkeley has come up with a novel approach for navigating small handheld displays. In effect the display is a "peephole" into a much larger information area. You see different parts of the display by moving the handheld around - no more tiny scrollbars. Check out the DiVX movies to see it in action. It even works in 3D!"

Comments

Text of Proposal document

[Alranor]

As it's been slashdotted already

OVERVIEW

Recent years have shown an explosion of interest in handheld computing devices (such as personal digital assistants, cellphones, and mini-notebook computers). These devices have a form factor that enhances convenience, portability, and durability, and they tend to provide desirable operational features such as instant-on, fast non-volatile storage, and simpler, more direct modes of interaction (touch screens, application-specific buttons, no need to "save" work and "quit" programs).

However, current display technology constrains the size of the display to be no larger than the physical size of the device. This sets up a tension between the desire to make the device small, light, and non-intrusive, and the desire to display a reasonable amount of information and provide efficient interaction.

Accessing a large amount of information on a small display generally requires some kind of selection or scrolling mechanism. Cellphones and PDAs, for example, have "up" and "down" buttons that are pressed repeatedly to scroll through lists of records, but using them is slow and cumbersome.

I propose a new scrolling mechanism based on the metaphor of a virtual window: the information is laid out on a virtual space much larger than the device itself. The device itself is moved around the virtual space to view a small part (a window) of the space. I hypothesize that this will have several advantages:

Scrolling becomes direct and intuitive; one can move to a new region of the space just as fast as one can move the device.
It eliminates the feedback loop of normal scrolling (press "Down", read, press "Down", read, etc.) and replaces it with a single movement.
It replaces discrete control with continuous control, massively increasing the bandwidth of information communicated between user and device.
It frees the hand used to operate the device, permitting scrolling and interaction at the same time. Scrolling moves into the background, occupying little or no cognitive load, producing the illusion that the entire
workspace is available at once.
It yields some of the advantages of two-handed interfaces for free: the non-dominant hand gives coarse positioning information, while the dominant hand does specific pointing and manipulation.

SPECIFIC GOALS

During this semester, i hope to achieve the following specific goals:

Choose a platform that is sufficiently open and fast to support this development (a Palm-based PDA will be a likely first choice if early attempts to interface to it are successful).
Explore and develop at least one method for sensing the position of the device. (Some possibilities to examine include: the use of accelerometers to obtain differential information; the use of a tether with a mechanical encoder to measure absolute position; the use of computer vision to locate a marker that's stuck to the device.)
Devise a task to be performed that requires scrolling functionality. (Possibilities include making a selection from a scrolling list, or locating an object on a large map.)
Develop a sample application that allows a user to perform this task (a) using directional scrolling buttons; (b) using conventional scrollbars; (c) using the virtual window technique (or techniques) developed in this project.
Perform user tests and compare performance and preference among these scrolling techniques.
Submit a short paper to UIST.

RELATED WORK

I've heard of other work on tilting input, but not direct-positioning input. Tilting, in my opinion, completely misses the point: tilt input is still differential rather than direct, and is therefore no better than holding down a scroll button and waiting until you've arrived. Positional input should be much better, because it just lets you put yourself where you want to be.

Joel F. Bartlett. Rock'n'Scroll Is Here to Stay. IEEE Computer Graphics and Applications, May/June 2000, pp. 40-45.

Jun Rekimoto. Tilting Operations for Small Screen Interfaces. User Interface Software and Technologies 1996.

Re:Text of Proposal document

[kah13]

He missed a big reference: this idea was already proposed by Jef Raskin. It's called ZoomWorld, some references can be found in his "The Humane Interface" book. The example he shows is the same idea used to provide information about patients in an ICU at Catholic Healthcare West. There is also a company doing something with this idea in webspace, Cincro. Their product is called Zanvas.

Mirror of the videos

[Kalewa]

Well my web server admin just said today "it can handle a slashdot, easily" - so I guess I'm going to see if he was just talking out his arse ;) These are just the low-res ones. peepdemo-200.avi (15.3mb) peepfig.avi (13.4mb) peep.avi (44.2mb) First two should be done in about five minutes, third may take a little longer.

Re:X windows virtual screens

[Samus]

I was wondering the same thing but I was lucky enough to be able to download the low res video before the site was smashed. Think of the old nintendo power glove or one of those joy sticks that you just hold in mid air and tilt around. Then think about that crossed with a virtual desktop. You are pretty close now. The interesting thing is this guy as attached the motion sensor to the pda and the screen scrolls around when he moves the pda. He took it one step further though. It not only works on a X,Y axis but also the Z axis. You can use your one hand holding the pda and the other a stylus to pick up an object and drag and drop it somewhere else. The demo video was pretty cool. I do recomend checking it out after the /. effect has worn off.