dan's face  
Daniel C. Robbins


Related Pages:

Focus + Context

People typically spend most of their time concentrating on one task at a time, whether on the computer or engaged in more tangible pursuits. This concentrated focus, though, is often interspersed with attempts to gain overall awareness of context. Sometimes this happens serially: intense focus while writing a document and then a brief and temporary switch to check email. Sometimes it happens in parallel: taking in peripheral information while driving. Current computer-based productivity applications do not support this basic need: the need to combine focus and context. By placing existing productivity applications in a 3D environment, the TaskGallery design lets the user easily and dynamically choose a balance between focus and context. A user can:

  • Choose to have a single document fill nearly the entire screen. 
  • "Step-back" and compare several documents.
  • Gather several related documents together into a Task
  • Spatially arrange Tasks to suit a particular working style

The screen becomes a room for the user to work in, a long gallery with paintings on the walls that represent different tasks.

Overall Structure

The overall containment structure of the TaskGallery is fairly simple. Windows are in Tasks, Tasks are in rooms, and the TaskGallery itself is made up of some number of rooms.N

Non-Focused Application (does not receive keyboard focus)

Focused Application (receives keyboard focus)

Ordered Stack  (linear arrangement of Applications on a base) Loose Stack (informal arrangement of Applications) Selected Applications (Applications in the foreground)

Task  (containter of Stacks and Forcused Appllications)

The backwall, or Stage, holds the currently active task. The Stage distinguishes the backwall from other walls and adds a sense of drama!

The sidewalls hold tasks which are not currently being used. 

Room (containter of Tasks)

[NOTE: This third-person view  is not typically seen by users and is provided here for didactic purposes.]

Spatial Memory

Our new TaskGallery 3D Window Manager extends the desktop metaphor, and tailors it to the way our own minds work to identify and track objects in physical space. 

The user navigates the space and manipulates objects with a simple series of mouse and keyboard commands. We kept the controls as simple as possible, reasoning that the less the user has to think about how the system works, the more brainpower she has left over for her work, which is the reason for using the computer in the first place.

Placement of tasks in the gallery takes advantage of human spatial memory – the active task takes center stage on a platform at the end of the room, and other tasks are shunted to the periphery. 

Each task has several related windows in it. This allows users to easily recall sets of related applications and  documents

The user can arrange the tasks any way she wants, and the system will remember where she left them. 

Over time a user may use each task in different ways. Some tasks may be used more than others and some tasks may be related to other tasks. To support this, a user can easily change the location of stowed tasks in the TaskGallery.

To reposition a task, all a user need do is grab it with the cursor, and drag to a new location. 

In user studies, we found that users preferred placing tasks on the walls over using the floor and ceiling

We made it as easy as possible to rearrange tasks, thus addressing the different ways in which a user may want to organize their information at different times.


Switching between tasks is a quick and simple matter of moving back down the gallery to see more tasks, and then selecting the right one.

Here, the user first clicks on the task which is on the floor. The task on the Stage automatically moves back to its "stowed" location. The Task on the floor then moves up to take its place on the Stage. The user can now move back in to the stage  to focus on the desired task.
Switching between tasks is a very fluid action and allows users to easily shift their attention from one detailed view to an overview to another detail view.


As the user adds tasks, more rooms are dynamically created, successively behind the "last" room.

Additional Tools

In the user’s virtual left hand is a palette that contains frequently used applications and favorite documents, which she can add to the current task or use to create new tasks. Users can freely arrange and group items on the palette to suit their own working style.

The StartPalette takes advantage of spatial memory by supporting user specified arrangements.

The user creates a new task, selects a background to help her remember which one it is, and arranges the documents and applications associated with that task as she sees fit. Instead of confusable and hard to learn icons, open documents and running applications are shown as snapshots, small versions of their actual appearance. In this friendly environment, finding a document in a hurry is easy. 

Looking at multiple documents at the same time is as easy to do in the Task Gallery as it is in physical space. Instead of fussing with moving and resizing windows, the user can just select a few documents and the system automatically lays them out side by side, scaling them to fit.

We have a notion of a "selected set" of windows. As users add windows to the selected set, they are automatically arranged to best use the screen real-estate.  Typically this works well for up to four selected windows. It is important to understand that by taking advantage of 3D hardware, we generally can scale windows rather than having to crop them. 

The user can move the application windows within the selected task like cards in a solitaire game – bringing them to the front to work on them, or back to sit with others on an ordered stack where they are clearly visible for easy retrieval later. The movements are accompanied by sounds, and together they create the illusion of an object moving through space. The illusion appeals to people’s knowledge of the way physical space works - they don’t have to learn or adjust to it, they get it intuitively. For more information on this, refer to the Windows page.

The untrained user can easily become confused and disoriented when confronted with a virtual 3D environment that allows freeform navigation. To combat this effect, we made the controls as simple as we could and constrained movement to a series of set vantage points. The user can choose between on-screen mouse controls or keyboard commands to go forward and back, turn from side to side, go to a home position, or gain an overview. It’s impossible to get lost in the TaskGallery.

The Future

As we move forward, we will extend our design by exploring additional visual metaphors and other high-level interaction techniques. These lessons will help inform future versions of the Windows User Interface.