[LINK] Demo of photosynth

[Adam Todd]

At 06:21 PM 8/06/2007, Kim Holburn wrote:
>>>http://www.devilducky.com/media/62817/
>>>Video demo of Seadragon and photosynth including a reproduction of
>>>Notre Dame cathedral, constructed entirely, digitally, from photos
>>>got from a Flickr search.
>>>Pretty nice.
>>
>>Bloody amazing!
>
>It is isn't it.

Ditto ...

>>I found it hard to grasp the concept that the only display/interaction
>>overhead is the number of pixels on the screen. Surely there is a lot
>>of overhead going on in the background. The entire contents of Bleak

[SNIP]

>>text.  But the zoomed out view effectively uses all the vectors of
>every letter in the book, they must store different sized bitmaps,
>thumbs if you will, I don't think you could render a whole book that
>size in real time.

Actually it's not that big a deal.  Given the background of the 
coders, it's really part of the 3D world, being integrated.

I do far more complex work in 3D with real time video, and scrolling, 
zooming, panning, tilting, merging and compositing comes easy.

Some of my 3D projects presently are absolutely HUGE.  To render them 
out for production takes several machines and several hours.  (Good 
thing I'm not doing a feature film like Lord of the Rings)

It's actually all very straight, take out the complexity.  Sample 
tine info, display only the amount of information necessary to fill 
the pixels on the screen.  There is no point sampling a 10 MB image 
in 300 dpi and "squishing it" down to size and displaying it, like 
most applications do.

I'd definitely agree that there is some kind of thumbnail cache built 
at some stage, it didn't appear that the demo was "live" and seeking 
images at the time.  So there is a processing overhead somewhere in 
the pipeline, but it's really not that big a deal and I'd doubt it 
really slows a user down.

As your search finds images, it caches them, you don't see them 
before they are downloaded anyway, and processing the inbound data is 
easy to do as you save it to disk, as the full image and as a cached 
interpretation.  It's all quantitative :)

Photoshop for example would run a zillion times faster if it used the 
engine that SeaDragon is based upon.


>>One has to ask: does a non-innovator and market debasement like
>>Mickeysoft *deserve* to acquire technology like this. I can
>>understand the speaker's apparent dischord with working at Microsoft
>>but at the same time receiving an appreciative applause from a
>>geek-filled audience.
>
>I think the question is also what is MS going to do with it?

A good question and I can't even begin to speculate, other than 
creating (Hmm) a desk top 3D search tool.  But MS is trying to get 
people off the desk top and paying to use remote applications, so who knows.

Maybe it's "cheap" and can be bundled as a bonus?

>Microsoft spends a lot on R&D but where are the results?

Often in places the ordinary person does not see.

>Just what sort of processing power does it require, he seemed to be
>saying that the main grunt work is done on the server.

Yes, I was thinking the same.  However I'm going to stick with my 
view that it's running on an ordinary average PC, the coders aren't 
"heavy" unlike MS applications, so I can see it being slick.

The Demo was probably on a X GHz Dual Core :)  I don't think it was 
server driven.