Disney Algorithm Builds High-Res 3D Models From Ordinary Photos 80
Zothecula writes "Disney Research has developed an algorithm which can generate 3D computer models from 2D images in great detail, sufficient, it says, to meet the needs of video game and film makers. The technology requires multiple images to capture the scene from a variety of vantage points. The 3D model is somewhat limited in that it is only coherent within the field of view encompassed by the original images. It does not appear to fill in data"
Time Saver (Score:2, Insightful)
This is great for scenery, it is amazing how much effort goes into the background scenery that no one will really pay attention to, but if you get it wrong everyone pays attention suddenly.
"It does not appear to fill in data" (Score:2)
Primitive, useless tech (Score:3, Funny)
The 3D model is somewhat limited in that it is only coherent within the field of view encompassed by the original images. It does not appear to fill in data
Just have the CSI boys zoom and enhance. C'mon guys, they've been doing this for years.
Re: (Score:3)
Just have the CSI boys zoom and enhance. C'mon guys, they've been doing this for years.
Darkman did this in the early 90s, long before CSI was a glimmer in CBS's pocket book.
Re:Primitive, useless tech (Score:5, Informative)
Excuse me?
http://www.youtube.com/watch?v=qHepKd38pr0 [youtube.com] (Bladerunner)
Bitch, please.
Re: (Score:3)
What am I supposed to bitch about? The video quality? The audio quality? The lighting? The blue blinking?
They all stink!
Re: (Score:2)
THIS!
the sky was blue as a TV with no input (Score:3)
Next you'll go blathering on about irrational things like "phone books", when everybody all knows they're called Kindles.
Re: the sky was blue as a TV with no input (Score:1)
So you would misunderstand the meaning of, "The sky above the port was the color of television tuned to a dead channel," thinking that meant RGB blue? Gibson was referring to the grey of random noise.
Re: (Score:2)
Apparently, your humor died in the 80s too.
Re: (Score:2)
Considering I was born in the 60's, you get off my lawn, you whiny punk kid.
Re: (Score:2)
Just say enhance a few more times....
Re: (Score:3)
To be fair, Bladerunner is set in a future world where technology is both far ahead of ours and seemingly behind ours in many ways (almost steampunk like...but forward thinking for a 1980s movie). If the camera that took that picture was more advanced than those today, it would be very possible for this to happen. Imagine a small snapshot taken with an 800 megapixel camera and this is very much possible, especially if one assumes that the actual "photo" uses might also contain an embedded memory fragment
Re: (Score:1)
Jeez. Harrison Ford needs a MOUSE.
New thing same as the old thing (Score:2)
The technology requires multiple images to capture the scene from a variety of vantage points.
So simply the big difference here seems to be synthesis of properly spaced stereo cameras by using cameras positioned anywhere.
But it sounds less impressive than some of the stereo movies released from older 2D movies where there wasn't any additional
cameras, and someone simply assigned depths to parts of the image and put CGI to work on it.
Re: (Score:2)
Cheating (Score:4, Informative)
The technology requires multiple images to capture the scene from a variety of vantage points.
That's cheating.
Re: (Score:2)
I think this has been done for some time now (Score:1)
Re:I think this has been done for some time now (Score:5, Informative)
http://usa.autodesk.com/adsk/servlet/pc/index?id=21350337&siteID=123112 [autodesk.com]
They have a cloud service that can make full 3D models from photos.
Re: (Score:2)
You got some Reddit love awhile back so quit your bitching :)
Re: (Score:2)
See this page [pauldebevec.com]; the Campanile movie is from SIGGRAPH 97. How is Disney's tech different?
I saw similar technology at CMU in around that same timeframe (late 90s).
My memory will be obviously hazy here, but the resulting output was much less refined. A simple box-shaped house, for example, ended up having wickedly jagged walls. The technology showed promise, but it was far from realistic.
The Disney folks, while not inventing the tech itself, seem to have taken it a step further. Their key claim -- "Unlike other systems, the algorithm calculates depth for every pixel, proving most effective at
Kinect (Score:2)
Yes, its fairly easy to build a 3d model given enough input and the right algorithms. Look at all the 3d scanner software that uses kinect and multiple frames to construct a higher resolution model.
Re: (Score:3)
Easy? Not at all. IIRC, to be able to theoretically get the model, no... let me try again: to even determine where your cameras are and how they are oriented, you need to be able to define something like 11 points in 7 photos.
At that, that just gets you to the point of having N equations, N unknowns. It doesn't give you the answer. Nor does it account for lens distortion. Throw in lens distortion, and you have that many more unknowns, therefore that many more points you'll need to define.
Having thought abou
Re: (Score:2)
You can roughly calculate camera extrinsic and intrinsic (including lens distortion) parameters from 3 world coordinates. (Assuming they're all in the photos)
One presumes with the amount of threshold and flexibility the system needs, that'll be plenty.
It's not that bad, just needs a lot of tweaking to get things to err, give decent persistent output
Re: (Score:2)
That's fine, if it's already known. But if it is unknown, then you've got a more difficult piece to chew on.
Re: (Score:2)
Re: (Score:2)
Like adding even more digital extras into the future re-releases of A New Hope.
Reminds me of that Outer Limits episode (Score:2)
...where construction crews are responsible for building our universe every second. Somehow our hero accidentally falls into that realm, and sees some whitespace or voids. One of the crewmembers tells him, when nobody goes there, they don't really have to build it. That when one of their crew is negligent, and forgets to put stuff where they're supposed to be; this explains those incidents when you could have sworn you placed that hammer on that table a second ago...
Re: (Score:2)
Re: (Score:2)
Thanks for the link! My 64M ram fails me...
Affine (Score:5, Informative)
The name of the algorithm is called 'affine reconstruction' and is a fairly well studied algorithm in computer vision. It is great that Disney and co. are releasing software to semi-automate the data input and reconstruction.
Re: (Score:2)
My question is, where is the software? All I saw on the Disney Research page was a Youtube video, and a link to white papers and links to download the original images. Just thinking how much I would love to take some old home movies and generate some semi-3D scenes from them. Could also be useful in 3D film conversions (although I thought that this was a similar approach to what they were already doing).
Re:Affine (Score:4, Informative)
Possibly not for those particular use cases, but there certainly is already freely available software to do the "structure from motion" reconstruction trick; e.g., vSFM [washington.edu] -- an easy(FSVO)-to-use frontend for a couple of tools from different research projects.
Re: (Score:2)
Oh cool, thanks, someone mod that up! I may play with this some when I get home. Now this leads to an additional quesion - both this, and the software from Disney, seems to focus mainly on buildings. I guess I can try myself, but I wonder how well this works with people? It would also be nice if I can feed in a video, but I guess I can always take my video, feed it into some video editing software and export as a jpeg or bmp securence or something.
Re: (Score:1)
I think you're confusing Disney with some other company which does open source graphics software...
AutoCAD has a service for this (Score:4, Informative)
http://www.123dapp.com/catch [123dapp.com]
Re: (Score:2)
It's called 123D Catch. They have an iPhone app and everything...
http://www.123dapp.com/catch [123dapp.com]
123D catch creates actual 3d models. in fact, any 123d catch demo is more impressive than this.. this is mostly useful for post process 3d effect to movies, it's one direction.
Not new (Score:3)
I have a program from the mid 90's that I got from a book about VRML http://www.amazon.com/Teach-Yourself-Vrml-Days-Sams/dp/1575211939 [amazon.com] which would turn say buildings in photos into 3d objects. I think it was only a demo so never really tired it out to see if it worked.
NOT Ordinary Photos, paraller moving video (Score:3)
http://www.robots.ox.ac.uk/~gk/PTAM/ [ox.ac.uk]
did it 5 years ago
Dead actors, new movies (Score:2)
Real scientitsts... (Score:1)
...use 4D light fields...
Subliminal messages? (Score:2)
Did anyone else notice flashing images on the screen while playing the video?
I can't get it to stop exactly on the image but it looks like some sort of white and red striped thing near 7 seconds in to the video.
Odd.
Re: (Score:2)
I hate responding to my own comment but I finally got it downloaded and stepped through the video.
It's just a person in the shot wearing a white and red striped shirt.
Re: (Score:2)
photosynth (Score:1)
Re: (Score:1)
Revisiting Stereoscopy (Score:2)
http://en.wikipedia.org/wiki/Stereoscopy
Low-cost, accurate CNC machines? (Score:2)
Re: (Score:2)
Usually stepper motors are used. These *can* lose track of their position but it is typically not an issue. It would mean that something had jammed or bound.
Remember Canoma? (Score:2)
From Wikipedia:
Released by MetaCreations Corp. in 1999, this application allowed users to create 3D models based on one or more photographs taken from various angles.
Great program. Never understood why it died. Assume it was corporate hijinks.
disney algorithm (Score:3)
Yeah, great news.
The future is now? (Score:2)
So art imitating life, or life imitating art? Just saw the Futurama episode Forty percent leadbelly [wikia.com] in which Bender takes a 2-D photo of a guitar, and gets it duplicated by a 3-D printer. Seems we are in the future already!