Instrumented GIMP To Identify Usability Flaws

Mike writes "New users of the GIMP often become frustrated at the application's unwieldy user interface. Now Prof. Michael Terry and a group of researchers at the University of Waterloo have created ingimp, a modified version of the GIMP that collects real-time usability data in order to help the GIMP developers find and fix its usability problems. Terry recently gave a lecture about ingimp and the data it collects. During each session, ingimp records events such as document creation, window manipulation, and tool use. A log of these events is sent to the ingimp server for analysis. The project hopes to answer questions such as 'What is the typical monitor resolution of a GIMP user?' and 'Is the GIMP used primarily for photo editing or drawing?'"

Re:GIMP will always be a gimp

[Aminion]

Or they could just call it GIMPshop [wikipedia.org]....

Re:GIMP and Photoshop

[croddy]

For a while, I actually believed the folks that repeated ad nauseum the mantra that GIMP's user interface was difficult compared to their beloved Photoshop. Then one day I sat down to try to do some quick photo edits on a Photoshop box. Two hours later, I gave up on its bizarre layer model and just installed the GIMP so I could get some work done.

The ease-of-use of a graphic user interface, in general, correlates far more with the user's pre-existing familiarity with the interface than it does with any design decisions of the interface itself. There are certainly areas where GIMP's user interface could be improved, but let's not pretend like it's some kind of embarrassment -- because it's not.

Re:representative ?

[Soulfry]

Because involvement in human-subjects research is voluntary, there will always be a self-selection bias. However, we can still estimate the representativeness of the population by understanding the types of people likely to download and install ingimp, and those who are not. If you fall in the latter camp -- you'd never want to use ingimp -- we really want to talk to you. Send us an email at the email address given on the site: http://www.ingimp.org/contact [ingimp.org].

In any case, having some data is better than having no data at all. Currently, there is a very active and vibrant group of individuals working on GIMP usability issues (see http://gui.gimp.org/ [gimp.org] ). ingimp's data complements this other data to help quantify the ubiquity of behavior/activity/computer hardware setups in the wild.

Michael Terry

Re:GIMP and Photoshop

[ajs]

The GIMP tries to do so much that it's quite hard to make it anything less than daunting for new users. The number of windows and dialogs needed for just the simplest edit are a horror.

That said, there are some basic problems that surpass the complexity.

• The arbitrary split between "filters" and "script-fu" which places important items randomly (from the user perspective) into one of the two areas.
  
• The number of dialogs needed to save a file (in some cases: filename dialog, replacement confirmation, export confirmation, and finally format settings).
  
• The selection of color in multiple places (many plugins and tools do not use the default FG/BG)
  
• The role of channels in the UI is not at all intuitively clear
  

However, most of this pales to the limitations that are inherent in the functionality. One of my biggest gripes is that the anti-aliasing code is sloppy in non-uniformly implemented. Try this: select a circle, and then use Edit->Stroke Selection. Select a 2 pixel stroke line and go. You will get absolutely HORRID aliasing. The same thing happens (though not quite as bad) with the paint tool stroking.

Overall, the GIMP is an amazing and powerful tool, but it has some serious warts.

Re:Scary

[Soulfry]

There is a very active group of individuals who are currently doing things like expert walkthroughs and observational studies: See http://gui.gimp.org./ [gui.gimp.org]

Our data is intended to complement this data by quantifying the ubiquity of tasks/activity/system setups. For example, what are typical resolutions of monitors? This type of information can help focus design by indicating what types of interaction designs are feasible and not feasible given the hardware of current users. What we've seen so far is a far greater number of 1024x768 resolutions than anticipated. Breaking these numbers down to see where in the world these resolutions are being reported is one of the next steps we plan to do to better contextualize the data. See http://www.ingimp.org/stats/monitors [ingimp.org].

We also have some emperical data to support the notion that the multiple windows design choice is not the best. Our data indicate that the percentage of the monitor covered by the document window is typically about 50% for most users (again, see http://www.ingimp.org/stats/monitors [ingimp.org] ). Most Photoshop users seem to maximize their document windows; with GIMP, this seems to happen much less frequently, probably because doing so obscures GIMP's other windows.

Michael Terry

Re:representative ?

[tknd]

Anything is better than nothing.

But even just by examining a few users, you will learn a lot. We went through this exercise in an HCI course I took. We were divided into groups of 4 students and we were required to observe 4 students (no in the class) while they used predetermined website they had never seen before (usually small online stores selling furniture). The total man-hours in the assignment would have been 1 hour pre-user * 4 * 4 observers = 16 hours. The operations were simple: find a bed and matching night stand, find 4 chairs and add them to the cart, etc.

With only 16 hours of work and 4 subjects it was immediately obvious that there were significant flaws and things that could easily be fixed. For example, there were many times where the user sat there and stared at the screen because they were trying to figure out how to do what they wanted to do.

I imagine with this GIMP project you could do two things: collect data about users of gimp (distributing the tool to anyone) and hand selecting users of the tool and examining the results on a case by case basis. That should provide a wealth of information.

Re:Representative?

[Soulfry]

Because involvement in human-subjects research is voluntary, there will always be a self-selection bias. (That is, we cannot force people to use this software any more than we can force people to come into a usability lab to participate in a usability study.) However, we can still estimate the representativeness of the population by understanding the types of people likely to download and install ingimp, and those who are not. If you fall in the latter camp -- you'd never want to use ingimp -- we really want to talk to you. Send us an email at the email address given on the site: http://www.ingimp.org/contact [ingimp.org].

Michael Terry

Re:GIMP will always be a gimp

[Anonymous Coward]

Here ya go:

GimpShop [plasticbugs.com]

Re:GIMP and Photoshop

[Rei]

Don't want sharp edges, is that what you're saying? Then stroke using a smooth paintbrush. It looks great to me. I've never used that tool before, and I figured that out in about five seconds.

Re:GIMP and Photoshop

[EvanED]

("waaah, it's not the same menu, I have to use the second button of the mouse, the horror")...

My main complaint about the Gimp UI is the multiple windows. It's cliche, but it's still true.

2.0 helped a great deal because they group things together so there are less, but it's still a pain. Now instead of raising just one window (Photoshop) I have to raise three (the photo, the toolkit, and the navigator/layers/other stuff panel). On Linux this isn't horrible because I can dedicate a virtual desktop to it, but (1) on Windows it's a downright pain in the ass and (2) I shouldn't have to dedicate a virtual desktop to a program to make it usable!

Re:representative ?

[Solra Bizna]

The 'G' in GTK stands for "gimp," FYI.

-:sigma.SB

Re:representative ?

[Celeron1point2ghz]

I thought the 'G' in GIMP stood for "GNU".

Re:representative ?

[eternalnyte]

It does, GNU="GNU's Not UNIX", GIMP="GNU Image Manipulation Program", GTK="GIMP Tool Kit". IE, they can't really "drop GTK", GTK was created for GIMP!

Re:representative ?

[yotto]

So GTK is "GNU's Not UNIX Image Manipulation Program Tool Kit" ?

Re:Then why not KIMP?

[akita]

They (KDE) tried, but the gnome fundamentalists didn't let them. This was in the middle of the gnome/kde wars, when Qt wasn't pure enough for some.

You can still find references on the web:
http://dot.kde.org/1096230607/1096270511/ [kde.org]

I always think, what if they did it ? It would be better than gimp for sure, seeing the quality of other kde apps at the time.

Re:Then why not KIMP?

[akita]

A better url:
http://lwn.net/Articles/101709/ [lwn.net]

Re:Is GIMP still being developed?

[BigSven]

GIMP does support variable brush width based on tablet pressure since version 1.2.

And yes, it is still being developed and we are very close to finally releasing GIMP 2.4 which will bring lots of new features and usability improvements.

Better yet...

[drakaan]

..at least for those of us using Windows XP or (shudder) Vista:
Paint.NET [getpaint.net] is getting better and better (and has an active user community creating plugins, etc). I tried it about 2 years ago and wasn't all that impressed, but as of my latest inspection, it's pretty useful software. Just make sure to check out the forums for effects and tutorials.