At a recent event in France, Adobe showcased a prototype 3D lens that could essentially capture a scene from 19 slightly different angles simultaneously, giving photographers a lot more to work with when they return home for post-processing. Essentially, the firm boasted that by using this lens along with software designed to understand the 3D nature of the image, individuals could utilize newfangled tools such as a "3D healing brush" and make perspective shifts based on the different viewpoints originally captured. Dave Story, vice president of digital imaging product development at Adobe, called the technology "computational photography," and suggested that it could open up an entirely new window of image transformation opportunities. As always, these type of things are better explained in motion, so be sure and hit the read link to check out the video.
[Via
CNET]
posted Oct 9th 2007 10:47AM
Certainly an interesting idea.
But, where's my Star-Trek style holography where I can take "pictures" of the far side of objects?! Truly the holy grail of photography.
It'd be really cool to be able to take pictures of the far side of objects.
Put a mirror on the far side of the object, come back and shoot.
>"It'd be really cool to be able to take pictures of the far side of objects."
Just attach a long stick to your camera and put a mirror on the end.
holy moly - exciting times.
A solution looking for a problem? I can't imagine this getting beyond the gimmick lens threshold... then again I lack imagination.
Bleh! It's just a compact version of those multiple camera rigs Hollywood use for those 'spinning-still' shots and it's effectivness is reduced by a scaleable amount.
The Hollywood setup that you refer to is only more expensive because they are using more hardware to cover a 360* radius. This is one camera trying to render a 3D image. So about a 180* radius in one camera. Maybe not 180* but if this technology works great you wouldn't have to have a 3rd of the cameras needed to complete the shot.
Someone that knows more please let me know if I'm wrong. This is strictly speculation based on what I've read.
um...this was in popular science 2 years ago.
old news guys. old news.
I seem to remember reading about a similar camera somewhere, probably in pop sci. As far as i can remember the camera i read about had slightly different focus on each "lens." But after watching the video it seems like the editing software part is new.
Bringing porn to a whole new level!
Ah yes! Porn in HD3D, add smell-o-vision to that and it's like being there ;)
dont believe the hype :)
nice gimmick tho
Sounds more like the image enhancer in BladeRunner
awesome.. so now I can fake nudes of my cousi... er... some celebrity even better!!!
As someone already said this is old news. AFAIK this is exactly what is knows as bullet time photography, but pointed at one direction rather than surrounding the object.
However what I dont understand is how they shift focus in postprod without getting perspective distorsion? Surely this is only a prototype, but still seems to me the functionality will be very restricted.
Some time ago something much cooler was achieved from a group of students at some uni (think I read it in engadget) they made this crazy cam (just one lens) where focus was not determined at time of exposure, but could be set at any point in postprod! they even showed examples of it in work.
Sweet! Those guys successfully violated the laws of Physics! Does it use a perpetual motion machine as its power source?
http://graphics.stanford.edu/papers/lfcamera/
hehe is he using a Mac with winxp?
:-P
With the perspective/depth information provided from processing the multiple perspectives together into a perspective-corrected voxel graphic, making silhouettes would be a crapload easier, against even the most crowded background; just select the depth range you want, instead of cutting out the complex 2D projected shape.
I can imagine using it, but it better have pretty good resolution for it to be useful. We're talking at least 5 *crisp* megapixels per perspective, at the 19 perspectives they claim.
Just google for "plenoptic camera" to find all sorts of places where Adobe has had nothing to do with this.
I'm not saying they won't be awarded the patent, but, you know...
I thought it was light field photography, but yeah, what he said. It's like taking a bunch of pinhole camera images (like a fly's compound eye), then stitching them together to have an image that has everything in focus, or computing an image focused at any depth and field range.
Can't wait to get one for my Mamiya. 3D on medium format. Sweet like jelly.
This is good news for movie makers, but if you thought you would fix a bad picture taken with that camera, then you have to be a really bad enough photographer to get your images messed-up in the first place, so just hire a professional to do your work.
Good to see Adobe presenting some research, though computational photography is not a new term or concept Recent SIGGRAPH 2007 papers here:
http://www.siggraph.org/s2007/attendees/papers/17.html
And a course:
http://www.siggraph.org/s2007/attendees/courses/1.html
Nah, fuck that. Computational photography is a field completely independent of Adobe. You can't come at me like that.
Kind of like Blade Runner photo tech. The ability to look a little around corners and stuff. Interesting.
unethical image manipulation!!! OMGwhere can I watch the rest!!!
This is definitely old news and it was is PopSci. However, what's cool, for those not in the know, is that you can choose focus on anything from close to infinity...AFTER you take the picture and in post production. For example, if you want everything in focus then you can have it. If you just wanted an object in the foreground and maybe 6 feet away in focus you can do that. If you wanted the background at infinity and something 3 ft away in focus but everything else not in focus you can have that too. It's actually pretty innovative. Check out the PopSci article online (if they have it).