BackgroundIn 2007, our lab released ingimp, a version of the open source GNU Image Manipulation Program (GIMP) that has been modified to gather rich usage data from its users. All gathered data is freely available at www.ingimp.org. The idea was to give HCI researchers and software developers a picture of how this sophisticated software application was used in practice.
To date, we have gathered data from over a thousand users. Recently, we published our analysis of 2-years worth of ingimp data including logs from more than 200 users. We found that the typical ingimp user performs relatively short, targeted tasks with the application, using only a small fraction of the commands and tools available.
With this as motivation, we are building Adaptable GIMP, a modification of GIMP that adds an adaptable interface. That is, an interface that allows the user to easily simplify the application to fit the task that they are trying to accomplish.
In this document, we present our latest design mock-ups for Adaptable GIMP. These are the latest in a long line of evolutions of the previous set of mock-ups posted in May 2009. We welcome (and encourage) comments, questions, and suggestions. Help us make these designs better! (Please see http://www.ingimp.org/adaptable/study for specific information on how we will use comments posted to this blog.)
IntroductionOur adaptable interface to GIMP is based around task sets. Task sets are collections of tools and commands that are used to accomplish a given task. For instance, a task set for painting might include the Paintbrush, Pencil, and Eraser tools, but would probably not include the Sharpen tool. While the user was using a painting task set, only the tools included in the task set would be visible in the interface, reducing unneeded interface complexity. The notion of a task set is somewhat akin to perspectives in Eclipse---a way to organize one's workspace around a specific task.
Task sets may also include rich, user-generated documentation. This documentation may be used to describe a particular workflow with the task set, describe the rationale for including certain commands, or serve as a tutorial for performing a certain task. This is a second use for task sets: communicating how to use the application to accomplish tasks.
In our current design, task sets are created by users. When a user creates a task set, it is uploaded to our server and instantly made available to the entire user community. This openness makes it extremely easy for people to share interface adaptations and information on how to use them.
Given that introduction, let's jump into the mock-ups.
Overall InterfaceWhen a user has installed and is using a task set, its tools and commands show up in a modified toolbox. The rich description document for the task set can also be displayed in a window within the application.
As mentioned above, once a task set has been created, it is shared with the entire community. At this point, both the commands and tools making up the task set as well as the description of the task set, are editable by any member of the community. When a user installs a task set, a copy of that task set is made on the users' system, including the commands and tools that were in the community task set at that time. As a result, that user's installation of the task set may become out of sync with the community version if changes are made. The user may choose to update to the latest version to follow changes made by the rest of the community.
In contrast to the commands and tools in a task set, when a viewer opens the description of a task set, the application always tries to fetch the most recent version. That is, the description document acts like a web page.
In fact, task sets and the description documents associated with them are accessible on a website associated with the project. The website provides a wiki-like interface for editing task set descriptions, and provides information on how the task set is being used by the user community.
Hopefully this has given a high-level overview of the interface. We'll now walk through a number of specific user interactions with the application.
Using Adaptable GIMP for the first timeThe user starts by downloading and installing the application. When run for the first time, the user is presented with the toolbox, which asks for a username and password.
The user clicks create new user and a wizard pops up. Step 1 presents the user with information on the Adaptable GIMP project including a research consent form. Step 2, asks the user to enter a username, password, and password confirmation. The box for username has a “check availability” button beside it, which allows the user to find a username that is not yet being used. Step 3 informs the user: “Your account has been created”, and volunteers a brief tour of how Adaptable GIMP works. The user can choose to dismiss this tour and start using the application immediately instead. Once the wizard completes, the user is automatically logged in to their newly created account.
Navigating installed task setsSelecting the active task set - User selects an installed task set from the drop down menu. The commands contained in this task set are then displayed in the toolbox, along with supplementary information about the task set.
Grid and Detail views - The user may select between two views of the active task set. In the Grid view (below left), commands are shown as icons without descriptions, arranged in a grid similar to how commands are typically arranged in the standard GIMP toolbox. In Detail view (below right), icons for commands are arranged vertically, with their names to their right, and an Info button. The info button for each command brings up a window describing the command. This window is similar to the description document window for task sets, and includes both information from GIMP's help files, as well as information created-by and editable-by the community. Each command has one description across all task sets (that is, the description document for a given command is the same, no matter which task set the user was in when they clicked the info button).
Forward and Back buttons - Navigation through task sets acts like navigation through recently-visited web pages. A back stack stores previously active task sets, and the Forward and Back buttons allow the user to navigate the stack. We decided upon this interaction for navigating installed task sets because it is well suited for working on tasks that have a hierarchical nature, including sub-tasks for which a user might want to use a different task set, before returning to the main task.
We would also like to include keyboard shortcuts for Forward and Back operations, to support fast switching.
Viewing a Description DocumentClicking the Info button for a task set (displayed in the Toolbox) brings up a window containing the description document for that task set. This window is non-modal, to allow the user to keep it open for reference while working. For the same reason, the window, by default, is thin and tall so that it may be placed unobtrusively beside the canvas on which the user is working.
When connected to the Internet, the application loads the current version of the task set's description document from the web. The extra information displayed around the description on the website can be hidden to conserve space. Below is an example of a task set description document that a hypothetical use has used to author a tutorial.
As mentioned above, the application tries to load the latest version of the description document. However, if the user is not connected to the Internet, then a cached version of the document from the last time the user was connected is displayed, along with a warning that the information may not be up to date, and instructions on how to view the latest version.
Installing new task setsSearching for and installing task sets can be done right from the Toolbox. The user types keywords to search for into the search box. While typing, a window pops up listing task sets that match the entered search terms.
The user can use the arrow keys to navigate down and highlight one of the search results, and hit the [enter] key to install the task set. Alternately, the user can click the Info button to bring up the description document for a task set, or click the Add Task Set button to install the task set. When a task set is installed, it is also set as the current active task set.
In the case where a task set in the search results is already installed, selecting the task set will set the currently installed version of that task set as active.
We would also like to include an application-wide keyboard shortcut for moving focus to the search box. Ideally this could support an interaction where, while working, a user could switch task sets entirely through use of the keyboard. For instance, the user could hit a keyboard shortcut, bringing focus to the quick search box. They could then use the interaction described above to search and install a new task set, or select an installed task set. Once the task set was selected, focus would return to what the user was working on, allowing them to begin using the task set.
Adding or removing commands from task setsThere are two scenarios where a user might want to add or remove commands from a task set: 1) for personal use, or 2) to change the community version.
In the first case, the user wants to make small personal modifications to the installed version of a task set. Perhaps they want to add a command that they use frequently, but is not traditionally associated with the task set's task. Or they might want to add a command to invoke a plugin that is only available on their system.
To do this, the user first makes the task set that they wish to edit the active task set. They then click the Edit Commands button below the list of commands in the task set. This brings up the Task Set Editor dialog, shown below.
This dialog displays the current state of the task set on the right, and provides a range of ways for navigating the commands available in GIMP on the left. Commands from the left side can be added using the Add button. On the right, commands can be removed from the task set and reordered.
The second scenario is where the user wants to make changes to the community copy of a task set. In this case, the user first navigates to the task set’s description document page, either on the web or within the application. Once there, the user clicks an [Edit] link in the Commands section.
This would bring the user to a web-based, wiki-style interface for editing the commands in the task set. The intention of this separate interface for editing the community version of a task set's commands is to clearly differentiate between making changes that will only affect the user's local system, and changes that will affect other users in the community.
This community command editing functionality can also be accessed through a hyperlink from various places within the application, including the description document for the task set. These will launch a browser and load the web-based task set command editor.
Merging community changes into an installed task setAs mentioned above, when changes are made to the community version of a task set, they are not automatically pushed to users of that task set. That is, installing a task set installs a copy of the current version NOT a subscription. As well, as mentioned above, a user may have made small changes to their installation of a task set.
The user may want to 1) update an installed task set to the latest version, or 2) view and merge changes from the latest community version with the changes that they have made to their installed version.
Users are notified that the community version has deviated from their installed version from the description box for the active task set. If there are changes in the community version, the date field is displayed in green and italicized, and when the user passes their mouse over this area, a tooltip is displayed which reads “There are changes from your installed version. Click [here] to merge changes”.
To merge changes, the user clicks the link in the tooltip mentioned above. This will open the Task Set Editor dialog, with the “Most Recent Version” option selected in the area on the left.
Note that this is the same window loaded by clicking “Edit Commands” as described previously, and so the user could also see changes in the community version by loading this window using the "Edit Commands" button and then selecting the “Most Recent Version” option in the left hand side of the window.
In this view, the Task Set Editor dialog displays the commands in the current community version on the left. Subtle blue highlighting indicates commands that are in both the locally installed version and the current community version. This makes it easy for the user to see the changes that they have made to their local version, and changes in the community version relative to their local version. Additionally, there is also a Synchronize button displayed below the standard Add arrow button. When clicked, the Synchronize button overwrites any local changes, effectively installing the latest community version.
Creating a new task setTo create a new task set, the user starts by selecting "Create a new task set" from the task set selection drop down in the Toolbox. This loads a modified Task Set Editor window containing a blank task set.
The user enters a name for the task set, and adds the desired commands. On the right, below the commands box, there is a small WYSIWYG rich text editing area for adding an initial description for the task set. Note that, in the Task Set Editor windows that we have seen previously, this area merely displays the current state of the description document for the task set, and does not allow editing. All editing for created task sets is done through a web interface using wiki-syntax, as we describe in the next section.
The reason that this window does include simple WYSIWYG functionality is to provide a gentle introduction to wiki-syntax for users who may be unfamiliar it, and for whom this might be a barrier to creating a task set. Users can use this simple editing functionality to provide an initial description, and see the wiki-syntax equivalent when they go to make further edits through the web interface, scaffolding their learning of wiki-syntax.
Editing a task set's description documentAs mentioned above, when the task set is first created the user can enter a basic description using WYSIWYG tools. Further editing of a task set’s description is done by visiting the task set's description document on the web, and clicking [Edit]. This brings the user to a wiki-syntax editing interface where they can make further changes, and upload attachments such as photos.
That's all for now!Hopefully that covers the most important interactions that we have designed for Adaptable GIMP, and gives a flavour of what our current plans are for the system as a working whole.
As mentioned previously, these designs are constantly evolving, and we value any and all feedback you can provide.
Thanks a lot!
-- Ben Lafreniere (May 31, 2010)