Exclusive: a look at Kinect Fusion and Lightspace from Microsoft labs
Exclusive: a look at Kinect Fusion and Lightspace from Microsoft labs
2011-12-21
what's happening in this build I mean
what are you guys working you got about
850 researchers about 1% of the entire
employee base of the company working on
farther out research this is a system we
call connect fusion I've got a kinect
camera here we used to sort of putting a
kinect camera down somewhere and having
a sort of capture the room but we
thought you know what happens if you
actually pick it up and you move it
around use your kind of a scanning
device right top left we've got just the
raw image that comes out of the 3d part
of the Kinect camera in the middle we've
got the raw image that comes out of the
just standard web camera top right we've
got what's called a surface image which
is how we use vision technology to pull
out different surfaces bottom left is
actually the 30 frames a second we get
from the 3d camera and building a whole
model of them you can see in the bottom
right we actually get a sort of nice
earth surface model all about right
easier a 3d rendering early action we're
building the model of the world by
taking lots of individual frames and
putting together and we can sort of
update with some flesh if I stick my
hand in here you see in the second hand
shows up I pull it out and it disappears
ok I drive back up in my head and yeah
will do even better JoshJepson you let
me sit down I don't I don't like that I
don't like the where this is going at
all oh that is this is disturbing I feel
like I'm having a bad drug experience
right now is that did you slip anything
in my coffee you know if you went and
tried to get by like an industrial
version of this there's sell for about
fifty thousand dollars right and i'm
using $150 camera plugged into a pc and
the crazy thing is this is connect
version one I mean presumably it'll be
future higher resolution yeah faster
versions of it connect so now let's take
the RGB camera and let's add that in so
wait that's a that's a render down there
this is an actual model and what I'm all
have a human is it's out here you can go
sit down there chillin just sit really
now drop your arms yeah see you're in
the model and you can you know given
particularly how social software is
today we need to understand how all
these kinds of technologies change the
way we interact with other people in our
family you know we spend a lot of time
doing research
and doing studies and all of that as
well okay so this is our next stop
pretty excited there's a laser warning
on the door let's take a look we are
researching different ways to use
projectors and depth cameras to make you
user interfaces above us we have three
projectors and three depth cameras kind
of looks like an early a beta version of
Gladys from pork yeah yeah you might say
that and so this is a research prototype
where we've basically testing some of
the interactive possibilities of this
setup now I can do certain things on
this this table it's very much like a
multi-touch or surface like experience
but this is just that this is a regular
tape yeah yeah one of the things we're
playing around with is using the sensing
infrastructure here to enable you to
effortlessly move objects from one
display to another I can now move that
back and I can over to the table top and
move it move it over here and you get
this kind of interesting relationship
now that we didn't have huh that is it
is very freaky so let's say I can't
reach both displays at the same time so
what i can what i could do is pull that
and then if I'm holding it and then I
can do such that and it appears over
here Wow so we're trying to make virtual
objects have a little bit of a physical
presence in the world by using the
projection now the key thing here is
that we can using the death data we can
correct the graphics and we know where
everything is so we can hit the hand
with the with the projected graphics
there we go there we go there we are
this is it's kind of incredible we
believe that a rising tide raises all
shelves are 850 people aren't going to
solve all the problems of computing
right right so we learn from other
people other people learn from us all
the researches about doing things that
nobody has ever done before and a lot of
people tell you it's impossible to do
that right and so we constantly sort of
bands of state of the art today it's
possible to do this or we know it's
possibly this tomorrow we'll take one
step further and we'll get one step
further than that and we'll keep kind of
working for until important point we go
oh wow now this X is whole new scenario
we can do because we figured out how to
do all the pieces
you
We are a participant in the Amazon Services LLC Associates Program, an affiliate advertising program designed to provide a means for us to earn fees by linking to Amazon.com and affiliated sites.