Stanford's New Website Converts Your Photos to 3D

An anonymous reader writes to tell us that Stanford has a new website that not only shows you how cool their new 3-d modeling system is, but actually allows you to give it a try with your own photos. The system can take a 2-d still image and estimate a detailed 3-d structure which you can navigate. "For each small homogeneous patch in the image, we use a Markov Random Field (MRF) to infer a set of "plane parameters" that capture both the 3-d location and 3-d orientation of the patch. The MRF, trained via supervised learning, models both image depth cues as well as the relationships between different parts of the image. Other than assuming that the environment is made up of a number of small planes, our model makes no explicit assumptions about the structure of the scene; this enables the algorithm to capture much more detailed 3-d structure than does prior art (such as Saxena et al., 2005, Delage et al., 2005, and Hoiem et el., 2005), and also give a much richer experience in the 3-d flythroughs created using image-based rendering, even for scenes with significant non-vertical structure."

Comments

Slashdotted

[d3ac0n]

Aaaaaand it's already slashdotted.

Wow. That was fast.

Re:Slashdotted

[Amorymeltzer]

Damn - I was hoping for someone to upload a picture of a pair of breasts to see how well it worked.

Re:Slashdotted

[Perseid]

Yep. I wonder if we can get a 3-D image of their server room on fire.

Re:Slashdotted

[alx5000]

An anonymous reader writes to tell us that Slashdot has a new website that not only shows you how cool their new hardware melting system is, but actually allows you to give it a try with your own servers.

Re:Slashdotted

[Rude Turnip]

And what, dear Slashdot reader, would your reference be to see if they look real enough or not?

Re:Slashdotted

[Amorymeltzer]

High Speed Connection + p2p program = lots of references

Hell, you need a reference? I've got a few gigs of references here for ya...

Re:Slashdotted

[Anonymous Coward]

My mom's just upstairs.

Re:Slashdotted

[alta]

No kidding, who would have ever thought that putting a link to /. to a service that does IMAGE PROCESSING was a good idea. Image processing is intensive on any server. Hell, lately /. can't even handle /.'s loads. It took 2 minutes to load this comment page, and earlier I was getting 300 errors when trying to read comments!

Re:Slashdotted

[nmg196]

Well no matter how bad they look, they'll still look better than mine.
Brian.

Games?

[webword]

Wow, can you imagine how cool this would be with respect to video games? Drop in some photos, crank up the customized first person shooter, and zoooom! You could even take photos or shots from movies and do the same thing (e.g., using Star Wars stills).

Re:Games?

[Stripe7]

About 20 years ago when they colorized Casablanca an office mate of mine was complaining about their ruining a perfectly good movie. I told him that he would be complaining even more when they used technology to make it in 3D. Seems that won't be too far away any more.

Re:Games?

[_KiTA_]

Wow, can you imagine how cool this would be with respect to video games? Drop in some photos, crank up the customized first person shooter, and zoooom! You could even take photos or shots from movies and do the same thing (e.g., using Star Wars stills).


There can be NO END to the verys to describe how much of a very, very, VERY bad idea making a CounterStrike map of your school/mall/town/etc would be.

Re:Games?

[Toonol]

I bet that the majority of schools in the nation have been converted to first-person shooter maps already. I know every school in my town had been, and that was back during the Quake/Duke Nukem days.

But don't tell the media that.

Re:Games?

[argent]

Right, convert it into a set of sculpties in SL instead, and add prim parodies of neighbors that bug you.

Re:Games?

[mwigmani]

Student Arrested for Making Videogame Map of School

Re:Games?

[Scrameustache]

Wow, can you imagine how cool this would be with respect to video games? Drop in some photos, crank up the customized first person shooter, and zoooom! You could even take photos or shots from movies and do the same thing (e.g., using Star Wars stills).


There can be NO END to the verys to describe how much of a very, very, VERY bad idea making a CounterStrike map of your school/mall/town/etc would be. The crazy old men from florida have won :(

Used for navigation systems?

[pwnies]

Could this type of technology be used for robots to allow them to identify what the 3d layout of the world around them is? Seems like a pretty powerful tool in that area.

Re:Used for navigation systems?

[grub]

Could this type of technology be used for robots to allow them to identify what the 3d layout of the world around them is?

Some (most?) robots already use dual cameras for true depth perception.

Re:Used for navigation systems?

[moderatorrater]

The problem is that binocular vision get's less accurate at longer distances. Also, for whatever reason, the robot might not be able to use two "eyes". Either way, another method of approximating distance would come in useful for anything that gets a lot of every day use.

Re:Used for navigation systems?

[disckitty]

Judging from the google cached pages, it looks like that's precisely what his research is for. Google cached pages: here, and here, and here

Re:Used for navigation systems?

[uglyduckling]

I can think of a very good use for this: laparoscopic "keyhole" surgery. One of the difficult things about that sort of operation is depth perception: until you've tried to do it, it isn't at all obvious just how difficult it can be to get a 5mm scissor blade over e.g. a blood vessel at the right angle. If a computer could analyse the image and add some depth perception cues it could really speed up the surgery and make a difference when something's going wrong and needs to be sorted out fast.

For those that can't get in

[eclectro]

I tried it - it converts your face into a Mars flyby.

Blade Runner anyone?

[virgil_disgr4ce]

Dammit, and all this time I've been decrying the impossible magical 3-d photo processing in Blade Runner! Curse my skepticism!

--Tedb0t

Inside Story

[jd]

• "What do you mean, you can't give a demo to the President?"
• "Errr, we posted the link to Slashdot and the network melted."
• "And what's this requisition order for a 24 gigabit campus network?"
• "We need the extra bandwidth."
• "And if I don't approve it?"
• "We post a link to your cat's facebook page."
• "Nooooooooooooooooooooo!"

!Slashdotted

[Anonymous Coward]

It is not slashdotted. The server crashed after I gave it an image of the impossible triangle.

Re:!Slashdotted

[howdoesth]

This is why we can't have nice things.

Re:!Slashdotted

[Chrutil]

>> The server crashed after I gave it an image of the impossible triangle.

Actually - that one is really easy to do in 3D: http://autodesk.blogs.com/between_the_lines/2004/08/a_really_cool_3.html

Re:!Slashdotted

[General Wesc]

Um...because one is a simple 3-d shape and the other cannot exist in a normal three-space, in the same sense that 'these two parallel lines meet' just won't happen. Hey, I can cut a spiral out of paper, so I must be able to make a square circle too, right? How would that follow?

(I expect the reply to this to be 'whoosh', but it looks serious to me.)

Photosynth

[blankinthefill]

While I know you're all Microsoft haters, bear with me for a minute. This sounds a lot like this Photosynth demonstration. The relevant part of the video starts at about 3:50, but the whole video is really interesting and I would suggest watching it.

Re:Photosynth

[nguy]

Photosynth takes multiple shots, this apparently takes a single shot. And although Photosynth is some nice engineering, (1) it wasn't all developed at Microsoft, and (2) it relies on decades of research work done elsewhere.

Microsoft does invest a lot of money in research. But what they are spending pales in comparison to all the work by other people that they are building on.

Re:Photosynth

[pcgabe]

Photosynth doesn't make anything 3D. It combines flat photos, and while you can move around and see photos attached at different angles, each of those views MUST be a photo on its own. The more pictures you add, the more angles you can look at, but Photosynth isn't making anything 3D.

These two packages are quite, QUITE different.

radar? lidar? either are superior

[Anonymous Coward]

granted, radar doesn't work so great for transparent surfaces to get the depth cue from -behind- that surface, while lidar gets a little iffy if it's -too- transparent to get the depth cue of that very surface. Combination of both - voila.

Re:radar? lidar? either are superior

[Hays]

Radar and Lidar are good for some applications, but they're fundamentally quite different. They're both active sensing technologies- they send out energy in part of the electromagnetic spectrum and then look in that narrow range of the spectrum and see what bounces back. This means that you have trouble seeing things farther away since you'd have to throw more and more energy to keep your samples uniformly bright or uniformly spaced. And it means your power requirements are much higher.

I think the most interesting part of computer vision is that which deals with passive sensing, such as this work. It senses the electromagnetic radiation that comes from our sun, or moon, or man-made sources. By using the same spectrum that our eyes use it should be able to get a qualitative understanding of the world similar to what humans can achieve.

Also, as humans we've built the world to be visually interpreted at the EM frequencies that we sense. This means our signs are readable in those frequencies, our indoor lighting works in those frequencies, etc... By sensing in those frequencies you make sure you don't miss anything that humans can see.

Google's next toy

[pauldy]

This would be sweet if they took all the imagery from google maps/streets and build out little virtual cities with no headed pedestrians and 5 legged dogs.

Not so new

[dfunked]

Several years ago I worked at a german university where recognizing of human faces was researched. We also did 3D reconstruction of faces, which was useful for training some algorithms. Although the technique is very different, 3D reconstruction from 2D images is not that new. Some examples can still be seen here: link

Re:Not so new

[yuriyg]

3D reconstruction from 2D images is not that new

That technology is of course superseded by 1D-to-3D reconstruction

Sorry, couldn't resist... uh-oh there goes my karma!

Example input/output, anyone?

[Kozz]

Since both the processing engine and the article are hosted on the same server, I can't even read about it. Anyone got a mirror to some sample input/output?

(No goatse renderings, please)

porn will never be the same again

[Rakkis]

nt

It's not slashdotted.

[Half-pint HAL]

I'm sorry to say that us geeks have been usurped by young hipsters in the website-disabling stakes. This site has not been slashdotted, it has been YouTubed. Someone at Stanford has been uploading videos of this to YouTube and inviting the plebs to go to their site before us. How ungrateful. The swines. Harumph.

HAL.

Re:It's not slashdotted.

[longacre]

I'm a \. subscriber and saw this story about 2 minutes before it went live to the masses. In that time, I was able to successfully visit the site and register. By the time I logged in, however, the \. post had gone live, and the Stanford site stopped working altogether. So it was indeed \. that crashed the site, not YouTube.

Re:It's not slashdotted.

[FormOfActionBanana]

I never thought I would see anybody spell /. wrong!

Re:It's not slashdotted.

[bob.appleyard]

It's not a spelling error. He's obviously a Windows user.

seems limited

[Anonymous Coward]

Other than assuming that the environment is made up of a number of small planes, our model makes no explicit assumptions about the structure of the scene;

Darn. My photos tend to be mostly of helicopters and boats.