World: r3wp

none ... ditto for ARM in general. We are waiting for the HostKit. 
Carl today blogged, that he is looking for HostKit maintainers for 
certain platforms ...

you said you were going to get an Android phone - do you want to 
take on the ARM hostkit?

no, because I am not fluent with C :-(  .... although I bought two 
books, and I succesfully set-up Extension and try loading SQLite 
DLL in C :-) So - I should be at least ready to test ....

I guess you have to have used a device with motion sensing or accelerometer 
to see how terrible an idea this is.

I have... (iPhone) and I do wish more things where gesture driven 
I'd love to do more things one handed.  with the iphone, you are 
just about forced to use it two handed-for anything... the touch 
screen is quite awkward to use with thumbs I find.


but these gestures have to be user controlable.... cause for example, 
itunes allows me to shake the phone and it randomizes to a new track... 
well when the phone is jacked into my car... hehehe, it can be *interesting* 
 ;-)

you start swerving out of pot holes... not for the sake of the car... 
but just cause you don't want to change tracks  ;-)

Henrik, it's the limited info you get with just the bare accelerometer 
that makes it's use not always so great - this motion processor should 
change that

I would love to have *fewer* things to be gesture driven, and it's 
not very often that I invoke a gesture on purpose, simply because 
I'm shifting in the seat or getting up from a chair or moving around, 
because I can't see the display for sun light. There are just too 
many ways to accidentally invoke a gesture with a handheld device, 
when the state you are changing is on the device itself. This only 
works if you are changing simple states, like a pedometer, but not 
with a "complex" UI as on a phone.


You have two conflicting requirements of precision levels for performing 
adjustments to a user interface, comparable to playing chess on a 
trampoline. It doesn't work.

I think you're not giving it a chance. The trigerring can be complex 
enough that it should just go off with random movement. Say a 45% 
tilt, followed by a double shake, followed by a lateral motion, to 
trigger some task

I have already given it a chance. It doesn't work.

this doesn't exist yet in current devices, so what do you mean by 
this?

Say a 45% tilt, followed by a double shake, followed by a lateral 
motion, to trigger some task

 - so how I'm I supposed to remember that? what if I'm lying on the 
 couch and not standing up?

the iPhone can already do most of this stuff. it just doesn't exist 
for android phones yet, it seems.

no, they have accelerometers too - with this processor it's all about 
the precision and the ability to follow relative motion to a much 
finer degree - there is quite a bit more here than just a simple 
accelerometer

it doesn't matter how accurate it is or even how intelligent it is. 
it's the basic principle of using gestures to change states on the 
device itself that doesn't work.

as for the actions to trigger various things, you could create aliases 
for various situations - as I said earlier, the human body is pretty 
good at memorizing motion patterns

I beg to disagree - I want this! :-)

actually no, because: 1. the actions are 3 dimensional and you have 
to have a 3-dimensional frame of reference to perform the motion. 
2. you have no force feedback, so you have to observe the screen 
while performing the motion. this is not like pouring a glass of 
water.

having more (better) sensors is never a bad thing - they can always 
be combined with other ones to achieve your intent

motions don't have to be 3d - they can be filtered for 2

as said, it doesn't matter if the sensors can sense 1/10th degree 
and milimeter precision. it's the basic principle that fails.

and I don't have feedback on my mouse gestures either - yet I use 
them all the time

your mouse gestures don't alter the state of the mouse, but the screen 
you are looking at.

huh - they alter the state of my browser

yes, the screen is not on your mouse.

no, it's not - the point is that it's a memorized gesture that can 
invoke useful functionality

but whatever - it's not for everyone

on the screen yes, not on the mouse. I already said this is quite 
a significant difference.

same with my motion sensing ps3 controllers - I love them in the 
games that use them well, but other people prefer the joysticks

yes, your PS3 games probably work well, because the display is not 
on the controller, but a still-standing TV in your livingroom. :-)

henrik, each application has to be logical in gestures it uses.  
the "shake" gesture is almost always a bad idea.  but tilting especially 
should be used MUCH more.

the acceleration metrics of the movement can usually indicate intent 
on some more obscure gestures.

gestures can use combinations of actions - to reduce accidental triggering, 
and to be appropriate or maybe mimic abstractly the action to be 
performed

Maxim, it could probably be used, but it fails more than it should: 
I own the Mass Effect game, which allows movement of the character 
via tilting the iPod, but you need a frame of reference to do that, 
hence you must sit very still when playing the game, and you must 
perform calibration, if you change your position.

Another app is a bit more reasonable: A star chart app that I have, 
will change the field of view if I move the iPod over my head, perpendicular 
to my face, but it has limited usefulness.

you keep mentioning current device Henrik

this is quite different

AdrianS, there is no difference.

One where it makes perfect sense is a sleep application, where I 
place the iPod on my bed and it passively registers motions I do 
throughout the night and then records them. Based on the motion it 
wakes me at the correct time in the morning. This requires no feedback 
to the display, so it makes good sense here.

I agree that on the current iPod (and other devices that have just 
this - like my ThinkPad), the usefulness of the accelerometer is 
debatable

well, if you keep saying that there is no difference, there's not 
much to discuss - since the whole point of the preceeding discussion 
is that the new functionality brought in by the processor and gyroscope 
is what makes a difference

There is no difference, because it's the fundamental principle of 
requiring shake, angle or motion gestures to use such a small device 
to change a state on the device itself that is wrong. I don't care 
how accurate the measuring of gestures is.

a viable example is the tilting to slide a list.  


on the iphone, the tilting is VERY slow to react cause its trying 
to guess the tilt based on acceleration and must be filtered.  so 
the lag is annoying.  Full body Motion capture has the same kind 
of problems with accelerator sensors.  with a gyro, the tilt isn't 
"guessed" its actual, so you can easily make it precise.

currently (and I keep emphasizing this), you can't even measure some 
of things this will allow

I agree with yout point Henrik, but I also think that most people 
use gestures as gimicks and haven't yet understood how to use them 
efectively.


I hate having to use two hands to quickly browse through contacts 
and pics.  I'd rather just tilt my phone and shake up& down slightly, 
like if I was letting sand (or cardboard cards) trickle down on a 
flat surface.


the gestures have to mimick real life or be very obvious (like turning 
the phone upside down).

yout = your

by "most people"  I mean developpers

one of the stupid gestures shown is using the tilt to select things... 
like a ball tilt game...   does the guy remember that ball tilting 
is quite hard in real-life?  that shoudn't be used at all.

Maxim, maybe it depends on the size of your hand and possibly a thicker 
iPhone, but I can browse photos, make calendar appointments, browse 
webpages, select music and type with one hand easily on the iPod. 
No gestures needed. I'd think I have average size hands.

Pinching is harder, though.

yeah, or more than one point touching