Slashdot Mirror
Lytro Illum Light-Field Camera Lets You Refocus Pictures Later
Iddo Genuth writes "Earlier today Lytro introduced a new light-field camera called Illum. This is the second camera with this innovative refocusing technology from the California based company founded in 2006. The new camera is a more advanced version of the first camera introduced in 2012. It has a much larger sensor with four times the resolution (Lytro still uses the term megarays instead of megapixels), a much larger and longer zoom lens with a f/2 constant aperture and of course the ability to refocus after you take a picture (the new Illum can refocus on many more points in the image compared to the older version). Users will also have more control of the camera, a larger screen, and the ability to create regular JPEG images or videos made from the refocused images they capture."
Is that IIIum, Illum, or IlIum?
The font slashdot uses makes it impossible to tell.
Seems like site is having database problems.
It's mostly a solution in search of a problem.
Photographers choose what to focus on very intentionally, it rarely makes sense to focus on anything else. Of course it's possible to misfocus, but in that case it makes no sense to let the user play with it.
It's still going to be low res, because you get a small fraction of the "megarays" the sensor provides. The spec for this camera was 40, IIRC, so it might get around 4MP, which can't really compete with a modern DSLR. While resolution isn't everything, having some margin for cropping and large prints is a very good thing.
The control for the interactive photos is still clunky. I can't find a way to for instance get the whole image in focus, though that should be possible. It does it while changing perspective.
It doesn't fix the other problem that leads to blurriness -- camera shake. It's all well and good to be able to refocus, but most people learn to focus right pretty fast. The problem is with low light environments, and this isn't going to save you if you handhold and shoot at 1/10.
The sample images still looks low res and blurry.
It costs $1600 and doesn't seem to have interchangeable lenses -- what, are they insane?
Overall interesting toy, but doesn't seem to have a practical use.
I just got the Gen 1 version of the camera. I like the small package size and the small price. You can use it to just take regular pictures, but you can have a lot of fun composing creative photos that takes advantage of the refocus capability to tell a story in the photo using the foreground and and the background as distinct photo elements. For example, a foreground subject tells one story, but refocus on the background element and the meaning of the story suddenly changes in a surprising way. Fun.
Since the short excerpt doesn't mention this I thought to mention: their forums say Illum produces a 4 megapixel image once it's exported in a regular 2D format.
You don't change lenses to change focus; modern lens assemblies move the lens to change focus, and move the aperature to adjust aperature. You change lenses to change the focal length (what zoom does) beyond the optical limits of the lens on the camera.
doesn't seem to have interchangeable lenses
For still photography, focus isn't a terribly hard problem to solve. Autofocus works, and DSLRs let you compose, focus, and shoot manually as well. Easy peasy.
On the other hand, for movies shot using large-format sensors, focus is a huge issue. The amount of work spent following focus on a movie is significant, and it fails more often than you might think. Modern lenses are incredibly sharp, but they have such a tiny range that is in perfect focus that they are hard to use. Admittedly, the people who use these cameras and lenses are professionals with years or decades of experience, and they do well... ...But -- if we could focus our shots after the fact, it would be a true game changer for movie making. We could chose just what part of the scene should be in focus, and change that throughout the shot easily. Yes, this moves yet another part of the movie making process into post, but that's not a bad thing. As other people have suggested at other fora, editing/coloring/framing and visual effects are all done in post, and it helps make better movies. This would help too. Having the depth maps automatically generated would make visual effects easier and better as well.
I recognize that the amount of processing that goes on to make these images makes a motion picture camera a challenge, and the number of high-end motion picture cameras is probably a tenth of a percent of the DSLRs that are made, at most. Still, we could just capture the 40 MRays and do the processing later; storage and networks are getting faster and larger all the time.
Come on, Lytro! Make it happen!
I love Mondays. On a Monday, anything is possible.
I think all crime scene photographers should get these cameras.
Get free satoshi (Bitcoin) and Dogecoins
I was considering getting the first version of their camera, but they use a proprietary image format for the original data and requests to open it are unanswered so far. Not even a SDK is provided to access the original data even though it was promised. Kind of disappointing and enough reason for me not to buy.
I assume you can post-process to get super depth of field without needing to stack images (which is obviously a problem when the subject is moving). Pretty cool product.
The lens is F/2, so you can't get the equivalent brightness of an F/1.4 (though you might be able to get the depth of field in post-processing).
The lens is 30-255, which is pretty good range, but you can't swap it out to go wider/longer.
Tilt-shift type effects (angled focal plane) should be doable in post-processing, but it would depend if they've added that functionality to their software.
It's a basic sampling problem. Instead of dedicating all your pixels to a single image, you're basically splitting them up and sampling many different images simultaneously. This will result in lower resolution in the final image than if you took that image with a "standard" camera. On the other hand, it makes it less likely that you'll miss a shot due to focus issues.
It's a tradeoff between resolution and flexibility.
Making the sensor much bigger would allow for more pixels, but would also be more expensive.
I thought that white niggers were in Texas?
Shoot at f/8+.
Apply dynamic blur via photoshop or plugin.
Done.
http://lewiscollard.com/camera...
I was hoping to read the comments here and find other people asking "Wait, /. didn't get the Lytro joke the first time around?" Where are you, people :-(
The "advertisement" video they posted on youtube actually delivers all the reasons you need to know why not to buy this camera.
The resolution is way too bad even for display on an ordinary 1920x1080 display. Stair steps visible all over the place. The color rendering is horrible, like in some old mobile phone camera. Plus there are artefacts to see where details should be.
Seriously, this is still nothing more than a party gimmick. "Refocus" your first few snapshots, enjoy for a minute, then the "something new" effect is over and what stays is the terrible image quality.
The only resonable use cases for lightfield cameras are currently industrial applications, and those are addressed much better by the Raytrix product line.
The capabilities sound very similar to the dual-sensor setup in the HTC One M8, right down to the 4MP resolution. HTC managed to get that in a $650 smartphone...
So we have a $1500 camera with a pretty piss poor design ergonomically, that takes picuteres that can only be viewed via propriety software? Yep, sounds like an apple product alright.
It's still a one-trick pony, and not a trick that many people need to do very often. Sure, a professional may invest in any number of specialized $1,200 tools to get images under special situations. It's just the idea that this revolutionizes the field of photography, or that _everyone_ needs this to get good pictures of Tommy blowing out the candles on his birthday cake, that's crazy.
I cannot think of a single time in my life when I wanted to press the button once and get two different images, one with subject A in focus and subject B blurred, and the other with subject A blurred and subject B in focus.
If this camera could take "deep focus" pictures, a la _Citizen Kane_, in which all objects at all distances were in focus at the same time, that would be mildly useful and a lot of amateur photographers would like it, even if the effect were a little boring. But, as nearly as I can tell, it can't.
"How to Do Nothing," kids activities, back in print!
May I remind you all that the so called 3D movies are not 3D but only stereoscopic movies.
You only have the depth perception, but you won't turn around the scene as you would be able to do with a 3D volume display.
There isn't any issue of focus with a real 3D movie (volume display), since the spectator focuses his eyes on the part of the scene he watches.
Of course I agree with the focus issues that the stereoscopic movies have.
I thought we've had this technology for years. I mean, every time you see them zoom-and-enhance on CSI they're taking some blurry out-of-focus element of the picture and rendering it in sharp high resolution. And those aren't even special cameras, they're usually just crappy 320x240 black & white security cams. It's all in the software, baby.
Chelloveck
I give up on debugging. From now on, SIGSEGV is a feature.
It's quite common to see sports photographers using something like a 100-400mm zoom.
Finally, I can fix the focus on all my street photography showing girls walking in the street! Absinthia Stacy