OSDev.org

I'm trying to consider some alternate user interface designs. Currently, for XANA I have two, however XANA is meant to be a testbed for all these new ideas, so it'd be great if people could suggest other designs that haven't really been implemented before. This might also serve others writing similarly experimental OSes.

I'm trying to stay as far away from traditional WIMP and CLI designs as possible, and thus far I've favored the joystick and keyboard over the mouse. My basic designs can be found in the Project XANA thread in the OS Announcements forum (I will not repaste here). Other ideas I had were direct neural interfaces with low-power EEG machines, and using a 3d interface in conjunction with an HMD and a setup using dual RF transmitters and a single reciever to find the location in 3d space of a hand (using the body's electromagnetic field to disrupt the signals and then using triangulation on the disruption pattern to find the point).

Additionally, the Z2 will probably have a NeXTSTeP-like minimized applications bar, and a circular selector for various saved states.

A classic one: http://www.youtube.com/watch?v=PLhMVNdplJc

using a 3d interface in conjunction with an HMD and a setup using dual RF transmitters and a single reciever to find the location in 3d space of a hand (using the body's electromagnetic field to disrupt the signals and then using triangulation on the disruption pattern to find the point).

Tried a Wii controller? (It's not limited to the corresponding nintendo console)

The multitouch, I feel, is a little too relegated to a specific space. I'm hoping for the ability to have it relatively portable, which can come in two forms: 1) portable devices; ex. I may be porting XANA and its Z2 interface to the gameboy and perhaps making a PDA (near future ability), and 2) ambient data spaces; ex. the 3d interface can be built into a room without having one necessarily be aware of the medium as an artificial thing -- sort of just moving one's hands in 3d space, and the computer picking it up, without one needing even the level of training or forethought of multitouch. Also, the user interfaces would likely be reprensenting n-dimensional spaces, and not be relegated to 2 or 3 dimensions (except in serialization for display purposes).

On the other hand, the multitouch, when combined with the n-dimensionality of the Z2 and the user interface concepts pioneered by the Apple Knowledge Navigator and Sun StarFire projects, might be useful if, say, it was placed on a wall, or on a piece of paper.

Oh, and the Wiimote would be useful if it wasn't so damn expensive. I can't afford it, and just using one's hand produces the same effect without having to search for a controller.

I had an idea once to use a webcam to track a pointer laser. In this way, moving the laser would also move the cursor. I figured it wouldn't be too hard as the laser is probably the brightest thing in the room, so all you would have to do it find it that way.

I never implemented it though........

That might be tough. Plus, in an OS you'd need to interface it with USB.

Alboin wrote:I had an idea once to use a webcam to track a pointer laser. In this way, moving the laser would also move the cursor. I figured it wouldn't be too hard as the laser is probably the brightest thing in the room, so all you would have to do it find it that way.

I never implemented it though........

That only works until you let somebody else test it who doesn't keep the curtains closed. There are ways to do that though.

I just installed Ubuntu on my desktop a few days ago, and it didn't take me long to realize how much I'd missed the usability of Gnome desktop while I was using Windows. If you've never used the "spatial" Nautilus in Gnome, you should definitely try it. But that's not what I want to talk about...

...because what I want to talk about is visual feedback. By now everybody probably knows that a window manager called Beryl gives you enough eye-candy to make pretty much every other system look old-fashioned (go search Youtube if you don't believe). But I just realized that it's not just about eye-candy as fancy animations help track what's happening on screen and stuff like shadows, slightly translucent inactive windows (with everything behind them slightly blurred) or virtual desktop arranged as a rotatable cube make it (at least for me) easier to keep track of everything I have open.

It's actually pretty surprising how much more pleasant a fairly traditional desktop can be made with minor adjustments (and a lot of eye-candy ).

I have a bit of an idea for you..what if in a filebrowser, you could hide certain files easily at will, like
you could have 2 files, "my_file.html" and "my_otherfile.jpg"
and there be a button to do something like "only show this type of file" or
"hide this type of file" and this being only a view thing, nothing is modified in the filesystem..

Also, a bit of a more revolutionary idea would be if you could group files together, but not abstract them to a folder..(you don't have to do any ../ crap to access the other files)
Like my_otherfile.jpg be put in a group where my_file.html is the owner, as the jpeg is used by html, and so is owned by the html file..

And these somehow be sorted out into small squares or something like that..

Here is an idea (taken from LCARS and modified): just show on the screen whatever information the user is currently working on, without any short of other functionality, and use context menus for whatever operation needs to be done.

Let me give you some examples.

Let's say you are doing word processing. With current UIs we see myriads of little buttons and options, menus, etc around the edited document. An alternative design would be to just view the document, as it is going to be printed. If you want to do anything, use a context menu which brings up a panel of operations relevant to what was clicked. If you have lots of documents and you want to alternate between those, just click somewhere on the screen and select another document.

Unfortunately my idea does not involve any peripheral other than the classic mouse and keyboard, but I can not really imagine anything else being used. 3D guis don't really work, in my opinion.

Another side of this idea is to avoid exposing the filesystem to the programs...treat the filesystem as a database: give each file a unique type (no extensions!) to each object, and let the user organize his information with views. The file organization screen should allow the user to manage views organized by the user's criteria.

For example, let's say the user has photos, songs and letters in his computer. Instead of exposing the filesystem to the user, let him express a query which results in a view, for example "all my photos taken between 1/1/2007 and now" or "all rock songs of Bono".

Finally, do away with the concepts of 'open' and 'close'...just save whatever the user does, when the user does it, keep the changes around, version objects, and expose to the user a functionality where he can activate any old version of an object.