Microsoft Demos "Deep Zoom" Technology

Barence writes "Yesterday, during a presentation for this year's Imagine Cup, Microsoft's Mark Taylor demonstrated the company's Deep Zoom technology to appreciative gasps of admiration from the computing students present. It's pretty impressive stuff, and you can try 'deep zooming' for yourself at the Hard Rock Memorabilia Site." Unfortunately the demo requires the Silverlight plugin and the story is pretty thin on technical details. I would be interested to see how they captured the image data to that level without massive pixelation.

Comments

Multiple resolutions

[clarkn0va]

My understanding is that you use different resolutions of the photo. Just speculating here (I don't anticipate installing Silverlight for another 24 years or so), but I think you're on the money. It should work something like Google Earth, where the resolution is improved progressively as you zoom in.

db

This is not new...

[ImNotYourFriend]

Typical Microsoft.. they bought out a company that created the technology. It's called Microsoft Photosynth and this video explains how they do it... http://www.youtube.com/watch?v=jEcHcRqxmj4

Re:Imagine Cup

[dave420]

No, this is the software they use on CSI (NY, at least). You can read a few articles about it starting here.

Deep Ripoff

[jdb2]

This is a lame "embraced" and "extended" version of an old demo effect that was first demonstrated in the early 90's, if not earlier. Obviously the entire "zoomed-out" image is not stored. The "zoomed-in" images are stored, but to make the effect work a series of intermediate images has to be stored between each "zoom stage". For example, in one implementation, when "zooming" through the intermediate stage between a "larger" and "smaller" image, for each series of frames an "outside" or "boundary" image is stored in full resolution and that image is zoomed ( and clipped against the view port boundaries ) until it is outside the view port while at the same time the "internal" image is enlarged until it fills the viewport and then the process is repeated again with the "internal image" now consisting of the next "boundary" image surrounding another "internal image".

Go to Pouet and you'll find many demonstrations of this effect.

jdb2

This has been done. Better.

[Animats]

See Charles and Ray Eames' Powers of Ten. Now that's a zoom.

As for doing it in real time, Keyhole (bought by Google and renamed Google Earth) was doing this on PCs five years ago. Any decent GPU can do this today, and you can download Google Earth to see it.

I saw one of the first systems able to do this in real time about 25 years ago. It was inside a classified tank at a major aerospace firm, and required a rack of special-purpose hardware. The user interface was beautifully simple - a big trackball (for pan), a lever (for zoom), and a knob (for rotation).

Even Microsoft's little film isn't original. That technique has been used a few times in commercials.

So Silverlight doing this isn't exactly a big "wow" development.

Re:Unfortunately?

[dodgedodge]

The beauty of Silverlight is the number of languages you can use. Even Ruby is coming to it.

Silverlight is a browser plugin. It takes all of about 10 seconds to install.

Flash needs to just go away.

Re:Haven't you ever..

[Eudial]

It's the actual software they use on CSI. Read more here. I think he refers to the software in which they miraculously rotate a single two dimensional image to see stuff from other angles, or enhance gritty 320x200 CCTV images into uber-high resolution with no artifacts or fuzziness.

(Might have been in some other forensics/cop show they did that, though.)

No, but I've seen GigaPan

[awtbfb]

I don't watch much TV, but the functionality is awfully similar to GigaPan.

Re:Installing Silverlight

[lilfields]

I would like to see Silverlight content able to be indexed on search engines...that is one HUGE disadvantage that Flash has...it would really help push this product with web developers. Otherwise you have to create two versions of the site, one for search engines and one for users...though I think 100% flash sites are stupid...but people use them, and like I said it could help Microsoft boost it's market share quite substantially. Silverlight does look pretty impressive...

Re:Crashed FF 3.0 on my Mac

[plover]

It consistently crashes Firefox 3.0 RC2 on Vista 64, too, after installing the Silverlight plug-in. I disabled the plug-in and no crashes.

Of course the Silverlight and the zooming works as advertised in IE 7.0.6

Re:WTF?

[Dragonshed]

The earth-shaking innovation is in the form including deep zoom as part of a plugin featuring a fast 2d compositor with video decoding and animation support, common RIA application components and controls using a small .NET Runtime, packaged in a 4.3mb download, "installed in 20 seconds or less", and all of it designed to run on multiple platforms.

MS Devs have done some amazing things within their allotted size quotas. /perspective-and-koolaid

Re:Installing Silverlight

[Dragonshed]

Similar hurdles exist for indexing silverlight content as they exist with flash. Silverlight is mainly for media and data/info visualization.

It's technically possible to index silverlight 1 content, because it's content is "loose Xaml files", which means the site has xml files alongside html/js/etc, that is rendered by the silverlight 1 engine.

Silverlight 2 has the same capabilities, but noone will use them, because using C# for application/interaction logic is way more productive than using Javascript. Silverlight 2 sites using C# have the following structure

SomeSite.XAP (zip file containing all code and assets)
- AppManifest.xml
- ApplicationCode.dll (.NET Assembly containing Entrypoint and embedded assets)
- SomeResources/ (compressed folder)
- SomeResources/SomeImage.jpg (...)

AdditionalContent.XAP (supplemental resources and code)
- AppManifest.xml
- SupplementalCode.dll

This makes silverlight 2 apps and content updates really easy to, but are a barrier to extract information.

In both cases the information gained isn't nearly as useful as textual html content, and completely different heuristics would be necessary to analyze the importance of one unit of textual content vs another. Indeed, nearly all the visual cues (The relative position, color, highlights, animations, and reactions to the user) would likely be lost in the process. Perhaps the search engine that can index flash and silverlight content is one that analyzes both visual and textual content.

Re:SeaDragon

[Dragonshed]

So they're essentially recreating Apple's Quartz + OpenGL + standard image formats with Photosynth + DirectX + WMPF. Simply put, apple does an incredible job visually representing itself, it's technology and providing a user experience that is very hard to match.

That said, I disagree that microsoft is recreating any preexisting technology. You could argue that DirectX is just like OpenGL, but that's likely grossly oversimplified.

Photosynth and Seadragon are demoed here: http://www.youtube.com/watch?v=vHsYnkLnepk

Neither of those are similar to things that apple has done.

Re:I don't get it

[prockcore]

what you're missing is that seadragon constructs these things analytically from a collection of photos.

Basically I can run around taking random photos some zoomed in, some not and seadragon will automatically stitch it all together.

Re:Uses gigapixel imagery as source

[markana]

And exactly how is this different from the wavelet-compressed MrSid format? LizardTech was doing this sort of "download-what-your're-focused-on" multi-resolution zoom *years* ago. Six years ago I could zoom in smoothly and deeply to an area of a multi-GB image, and the plugin would grab only those pixels needed to show that area at that resolution.

So what exactly is new here, except for the use of Silverblight?

Re:Imagine Cup

[ozmanjusri]

they just interpolate the color values between each pixel "point" instead of drawing huge square pixels.

It's not a new interpolation algorithm.

It's a live version of the The shift-and-add method or image-stacking technique used by astronomers for decades. It's just that now computer hardware is fast enough do it seamlessly.

Basically, the zoom is made from hundreds of still photographs taken from different vantage points. There was something similar being done with tourist destinations, if I remember correctly.

It's an interesting toy, but the practical applications are limited by the lengthy production process.