Stanford University via YouTube
A YouTube video demonstrating the capabilities of the Frankencamera.
Posted on 10/12/2009 7:20:19 AM PDT by ShadowAce
It's big. It's ugly. And it's made from recycled parts, at least for now. It's called the "Frankencamera" — and it might someday change the way you take pictures.
Computer scientists at Stanford University say the new camera works something like an iPhone: It can be altered in nearly infinite ways, depending on the applications downloaded to it.
Even the best digital camera on the market today has lots of limitations, the professor behind the prototype, Marc Levoy, tells NPR's Guy Raz.
Say you want to take a photo of your child playing with a new toy in a dark room, near a bright window. It's tough to do now because of the variations in lighting within that single frame. If Junior's face is visible and bright, the tree and the sky through the window will be bleached out. If the scene outside comes out just right, Junior will be so underexposed you won't be able to see the glee on his face.
Not so with the Frankencamera, says Levoy. For the prototype, he and his colleagues developed a program that instructs the camera to take two rapid shots if a frame has both dark and light parts. One shot exposes correctly for the dark; one shot exposes correctly for the light. The program then merges the two images into one, taking the best parts from each.
And what if a camera could do the same thing for focus — take three shots, focusing on different things in each frame, and merge them into one crystal-clear shot?
According to Levoy, these are just a couple of examples of how programmers could change the future of photography with the Frankencamera. The key is that the camera uses a Linux operating system. All digital cameras are essentially minicomputers, says Levoy, and they can be modified a bit by the photographer. But the manufacturer determines what features are available. Linux is "open source," which means the camera owner can change everything about the electronic guts.
A YouTube video demonstrating the capabilities of the Frankencamera.
Photography took an enormous leap about 30 years ago, after Willard Boyle and George Smith came up with the "electronic eye" sensor that allowed cameras to ditch film and go digital. Boyle and Smith won the Nobel Prize for Physics this past week for their work. But Levoy says that digital photography has not progressed as fast as it should have since then — and that's what he's hoping to change.
Still, the Frankencamera is only a prototype at the moment. Presumably, it will become smaller and more elegant over time. But Levoy says his team is not planning to try to compete with Canon or Nikon. "Our goal is not to make a product. Our goal is to try and push traditional camera makers to incorporate more of these flexible ways of producing images in their cameras."
And, he says, their goal is to try to make 1 of every 5 of the photos you snap a keeper, rather than 1 in 100.
But why can't you just take your pictures with a standard digital camera and fix them later in Photoshop? "Well, sometimes you haven't captured everything you need to capture to make a good picture," the scientist says.
Levoy is the son, grandson and great-grandson of opticians, so he comes naturally by his interest in things visual. He imagines a future in which the camera is able to make adjustments that better mimic what the human eye is capable of.
But the Frankencamera won't necessarily turn everyone into Annie Leibovitz. "The part that we're not automating is the part that's nine inches behind the lens," Levoy says. "You still have to be a good designer of photographs."
The photos below, taken by Stanford's Marc Levoy with a conventional Canon 5D SLR camera, illustrate how the Frankencamera might combine images to create a single, well-exposed picture.
Marc LevoyA photo of Zurich with the right side properly exposed.
A photo of Zurich with the left side properly exposed.
Combined, as they would be if taken by the Frankencamera, both photos create a photo of Zurich with each side properly exposed.
Very cool...American Exceptionalism at work.
I'd love to see some sort of intelligent, partial-frame compensation that more closely replicates the behavior of the eye. The bright part of the shot could be corrected for in real time, necessitating only ONE capture instead of multiples.
Still, this idea has huge potential.
That is called a bracketed exposure. It’s a snap to combine two, three or four images in PS for a similar effect. I do it all the time.
Seems to me this ‘process’ would only need to be programed into the brains of existing cameras to deliver pretty much the same result.
I had a pic exposed one time.....cost me $200 to get rid of it...*grins*....actually cool photos.
This won’t work on action photos. It’s fine for stills. Unless they can get two frames at about 1/1000 sec within 2/1000 sec. Also, there is lens aperture to deal with. It’s probably going to work out better as a software update to current technology.
My Canon EOS has a shot bracketing feature that I haven’t played with but sounds a lot like what this does.
From what I can tell it takes 3 different programable exposures with one shutter click. The 3 different photos would be easy enough to layer together in photoshop.
My Canon already has a bracketing feature. Guess I need to play with it a bit today.
It is only a matter of time before a lot of the digital enhancements are done on the camera, not just in software on a computer. Powerful stuff.
Bracket usually means adjusting f-stop or shutter speed in successive shutter firings. The camera will take one pic underexposed by whatever stop adjustment you set, then take one exposed at normal and then another overexposed by the stop adjustment.
It take three separate pictures.
Most of the time you can adjust the pic in PS or PSPro to get the same effect. Setting the "highlight, shadow and midtone' settings will, in effect, change the "exposure".
I use Paint Shop Pro which is available for about a quarter of the price of Photoshop.
If you use Photo Shop, you can shoot in RAW at 1 - 2 stops bracketed and combine then for HDR (High Dynamic Range). Be sure to use a sturdy tripod though.
If it can keep me from getting my finger in the shot, I'm all for it.
I've used PSP for years, and find The Gimp to be a pretty good free alternative.
A lot of it already is done on-camera. However, a lot of adjustment and tweaking in post-processing is still desirable. Blanket enhancements are often not what the photographer wanted, or envisioned when he captured the photo.
Thans. Yeah, I know. What am I going to do with PSP? I really don't feel like learning a whole new menu, commands, etc. Besides, I don't think I'll ever need another PSP upgrade. I don't do commercial stuff just my pics.
I really have trouble working with/understanding layers and how to manipulate them. Pointing me to a comprehensive tutorial would be greatly appreciated. PSP's help on layers leaves me confused.
Does "The Gimp" handle RAW images. I looked but could not find out. PSP's upgrade allows PSP to read Nikon RAW. I couldn't bring myself to load Nikon's Capture NX program onto my computer. It seems very heavy on the system.
My thoughts:
1. It’s a cool and dare I say obvious feature - My Nikon digital camera has a whole library of features that it can do - I could see this as yet another digital camera feature. I don’t see this as revolutionary but evolutionary. But that’s just me. I guess the hard part is the registration problem which is lining the two images up exactly right. Maybe that’s hard.
2. I don’t know Mark Levoy personally but I do know him by reputation and he’s a very well respected figure in the computer graphics field - i.e. he’s the real deal.
(IF #’s 1 and 2 above seem contradictory - well I guess that’s cause they are, at least a bit :) )
OK re-read the article. The double exposure trick is just one application of the more general feature of making digital cameras open source. I frankly don’t see that happening, in any case on a large commercial scale. The big camera companies would have a lot to lose and little to gain by going open source I feel. And the small startups would have a hard time competing with the Nikons and Canons.
Hmm, the first pic of Zurich is the prettier.
Check out Trey Ratcliff Photomatix Pro and LightZone are far more cost effective
and more like a photographers "darkroom" than PS.
A similar effect can be achieved in Photoshop using duplicate layers.
I love well done HDR photos......C
Seems like this would work if the cam is on a tripod, but I don’t see how handheld shots could lineup the pixels exactly right.
bookmark
a. Shoot RAW
b. Shoot multiple bracketed exposures
c. Process images in photoshop / lightroom / photomatix
There is nothing that camera will be able to do that is not possible with a good DSLR and some good post production.
A photo of Mendelhall Glacier in Juneau taken last month with my cheap Canon SD1200 using auto feature
It is already being done in software.See Venice in HDRi
Fujifilm has a camara (FinePix F200EXR) that takes two instantanious photos and then combines them as above.
This will take all the fun out of photography, like hunting rabbits with a bazooka!
I think cripplecreek already has Photo Shop.
I’ve played with Photomatrix. Will check out the other link, thanks.
Awesome tagline, BTW.
I’m sure if congress will just provide a trillion dollar grant, some Leftist geeks in a Leftist college will solve this problem when they happen to be sober.
Goobermint solves all problems and has all answers! /sarc
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.