Message #1274

From: Melinda Green <>
Subject: Re: [MC4D] Re: Announcing MC4D for Android
Date: Sat, 27 Nov 2010 23:29:21 -0800

On 11/27/2010 9:57 PM, Andrey wrote:
> So it could be the following way:
> it you press left bottom corner of the screen by one finger, and drag picture with another finger, it works as 4D rotation of the picture, and if you press right bottom corner, then dragging up/down will work as 4D sliding, and left/right as Z-rotation? Like virtual ctrl and shift buttons on the screen.

I like your basic idea but remember that this would ideally be
discoverable without instructions. I’d provide instructions but I have
to assume that most people will not read them and simply decide whether
to delete the app based on whether they can figure it out quickly on
their own or not.

For this suggestion, I suppose it might be made to work if there were a
button with some sort of text that implies "For 4D rotation, hold while
dragging" but that’s probably not worth the screen space. I guess it
could be implemented exactly as you describe so long as it’s not
required to successfully use the app. Those few who do read the
instructions will find a treat. Similar to right-click options that we
can’t assume most users will even try.

> And face rotations… Catch one face with one finger, keep it in place and twist with another finger… It may be 90-deg rotations again (6 movements) or free 3D rotations with positioning to the closest possible orientation after the twist. It could be great interface! It’s pity that screen is too small, may be no room for the second finger movements.

We’ve fantasized about this sort of "snap-to closest twist" idea before.
Ideally this would include a stereo display and a data-glove or haptic
feedback input device to more physically express your intent. Smartphone
screens may be large enough for such a two-fingered approach. It could
certainly be worth a try. The biggest problem appears to me that it
could confuse many users who expect pinch-zooming to be the only
two-fingered gesture. I think that a Z rotate could be added to
pinch-zoom without violating user’s expectations, but it seems that your
two-fingered gesture would need to violate user expectations. They may
generally forgive the unexpected behavior but it is a risk.

> Equivalent in the desktop application: click some face to define that you’ll work with it, then rotate it (as 3D object with fixed point) to the proper position. It may be more understandable for common users than current "sticker-based twist definition. But I’m not sure.
> If I find multy-finger events in C# for tablet screen, I’ll try to do something like that. But not sure that it will work in the hyperbolic space…

Hm, I suddenly wonder why I’ve never seen anyone try creating a
two-mouse UI? You’d feel like you had grasping hooks for hands but you
sure could do a lot more than with one hook. Wild thoughts aside, you
could avoid adding a mode by instead pressing a control key to establish
a center-of-rotation at the middle of whichever face center is closest
to the mouse and then dragging that face with that key held down. Let go
of the mouse or control key to "snap-to twist".

> I don’t think that it’s good idea to hide controls of the central face. If you can’t twist both central face and invisible face, it gives the next level of difficulty - and in this case i’m not even sure that all scrambles of "unrestricted" cube are solvable.
> To solve the invisible face only is the same as solving of the first layer of 3^4. May be it will give interesting puzzle - I’m not sure. For me it doesn’t look elegant enough.

Yes but solving a layer that you can’t see would be a new challenge. You
could attempt the 3D analog by solving one layer while keeping that face
pointed away from you at all times. You get to look at the side faces
but not the furthest one. Would the added challenge be fun or just
annoying? It seems like this would be a close cousin of blindfold
competitions which some people enjoy.

You may be right about not hiding the central face but that’s easily
tested both ways and doesn’t need to be decided up front. These
constrained puzzles look the most interesting things for me to fool
around with right now but there is definitely a lot of room for creative
interaction design here! I’m really starting to enjoy Android
development for many reasons and this is a great one.

> But 2^4 (with recentering!) may be good puzzle for Android - not very large and not very difficult, easily can be solved in one session :)

Oh, I’m sure you could solve it without a recentering option, Andrey. ;-)
The 2^4 definitely works better on these small touchscreens. My only
concern is with the discoverability of the polygon-based twisting UI.
What is the likelihood that someone with only some familiarity with the
3D puzzle will figure out how the twisting works? Is that likelihood
greater or less than for the 3^4? These are my main questions, and
without some focus testing, I doubt that I can guess with any accuracy.
If anybody has opinions on these two questions I would love to hear them!