pouët.net

http://www.ted.com/index.php/talks/view/id/129

<< Using photos of oft-snapped subjects (like Notre Dame) scraped from around the Web, Photosynth creates breathtaking multidimensional spaces with zoom and navigation features that outstrip all expectation. Its architect, Blaise Aguera y Arcas, shows it off in this standing-ovation demo. Curious about that speck in corner? Dive into a freefall and watch as the speck becomes a gargoyle. With an unpleasant grimace. And an ant-sized chip in its lower left molar. "Perhaps the most amazing demo I've seen this year," wrote Ethan Zuckerman, after TED2007. Indeed, Photosynth might utterly transform the way we manipulate and experience digital images. >>

what a typical m$ rubbish idea. Look I'm hip ! Happy sore neck assholes..

i wonder if you'd reacted different if the guy had happened to be working for google instead..

I personally view images by typing:

type gargoyle.jpg |more

The idea itself looks pretty amazing to me.
The execution and maintaining the vast amounts of data is somewhat an issue to me. Who is responsible for keeping the links up-to-date? Up-and-running? A classic example of "This looks too good to be true - I will believe it when I see it".

I was very impressed when I saw this thing, mostly by how fast it ran. Maybe it turns out to be useful too :P

Check it out here:

http://labs.live.com/photosynth/default.html

I'll have to say, I love that technology.

hehe the MS offensive against Google innovation has finally taken off :>

btw this technology seems very useful for crime or accident scene reconstruction or for online pro pictures bought by journalists or publishers, but do we really need it for our personal picture collections or even the ones on social networks ? kinda overkill :/ but it's prolly nowadays trends, creating new geeky fake needs...

probably takes a bit of precalc

Yup, but once the initial precalc is done it should run smoothly. People don't think *that* fast, so I doubt they'd be able to coordinate the cognitive and motor skills they'd need to out-zoom the computer meaningfully. By meaningfully I mean not by clicking the mouse as fast as they can, i.e. every stage is a considered appraisal of where they are and where they want to go - by which time the computer should have done it's automatic grunt work for the surrounding scene.

this+gps in cameras (already reality) will be interesting in 10 years.

zest : I agree with ya. out of the 2 companies I have more time for google. Due to having to wait THREE months for my vista upgrade, which then in turn was 32 bit and not 64 bit, for my 64 bit pc amongst other things.

remember years ago when the net was starting flurishing, netscape etc was ahead of the game and MS play catch up. reminds me of that again all over again. Google are doing good things.

I do like when a company comes along with the latest thing, then others are pretty much forced to compete, if it makes things better for the end user thats the good thing.

Skrebbel: right, also a gyro will help.

I'm waiting to see the "solid-3D" version on wich they're working (seems the one visible in the presentation video) :))

I must've missed all this google-innovation.

MS software does eventually leave beta, after all. ;)

Ok, I was misguided by your avatar, Stefan :)

The first cool thing we see "from Microsoft" in years, and they just bought it from someone else. Figures. :)

Really cool indeed. Chaos already showed me that video last week, I totally dig the Cathedral part.

This is undoubtedly an amazing achievement from an impressive collaboration of talent, even if some of that includes M$ (oh noes!); I think we all agree with that. Personally though, and I'd be interested to find out what you guys think, I can't see Joe public creating their own models (e.g. hey come see my room, I placed each of these photos in exactly the right place, it took me three weeks) but rather if some kid wanted to share photos of their room they might use a technology that would shape the pictures into a ball and spin them around or whatever; a bit like what you can do on a PS3 right now.

What might be more plausible is image sharing facilities such as Flickr, as already mentioned by the creators, adopting this technology and hiring dedicated staff to build models where they have a lot of photos or models made by requests from the community (at a price possibly, business opportunity perhaps???) so eventually instead of visitors to the site just viewing individual's collections they'd be looking at models of many user's photos.

I think this meta-data is going to throw up the most problems. People are going to want to know when photos were taken for example (perhaps to see changes in a location) and by relying on amateur material, it's going to be difficult to ensure reliability and accuracy of data like this. I think people will want to search for their own photos too and at the moment the tech-demo isn't showing user-names or anything similar.

But damn is this cool...

i dont get it.. its just picture slabs 3d referenced.. what are you all so excited about? o_O

Not impressed at all. It looks really messy and awkward and is effectively just a 3d scene with photos texture mapped onto it.

It's actually not very entertaining or useful either. I'd much rather see a basic 3d render.

in all respect, the speed of rendering of the engine seems impressive for the quantity of data. i still wonder how they connect markers between pictures to get the 3d data though, must have some manual work involved for sure. O_o but the technology aint anything new or groundbreaking. its applications through a stable engine might still be though if ms does their shit properly.

gloom: that's MS strategy since the very beginning to catch the market or stay state-of-the-art (uh) : wasn't MSDOS just a relabelled version of QDOS, the Quick-and-Dirty Operating System, bought by MS to be quickly sold on brandnew IBM PC ? ;D

PS: there's no manual work. That's the significance of it. You take a bunch of pics, and it correlates where they all came from.

Now come on guys, figure it out already. This will obviously be integrated with Live maps. MS drove a picture-van around seattle last year taking pictures in all 4 directions, but you didn't care about that until google put those pics into a cube and let you spin around. When google starts integrating cell-phone pics into google maps idea a year from now (and the UI sucks because they don't have professional testers) you'll be singing it's praises and you'll have forgotten this.