Monday, April 16, 2012

Lytro - It's Aliiiiiive!


The most succinct description one could give of the Lytro Light Field Camera is this: it's not a camera, it's a purchasable tech-demo.  That was my impression when I placed my preorder last fall, and six weeks of shooting with the device hasn't changed my mind.  There, you can stop reading now.  You're welcome.




Oh, hi.  You're still here?  As you may already have noticed, you can click on the picture above to change the focus.  Seriously, try it.  Focusing after having taken the picture is a neat trick enabled by the Lytro camera.  Lytro calls this a "living picture", and their site is steeped in a thick broth of marketing copy explaining how much more engaging these photos will be for the viewer, compared to the static, fixed, presumably "dead" pictures of yore.  I suspect I'm not going out on much of a limb when I say that's a bunch of crap. Viewers will find clicking on photos to refocus them to be fun for exactly 17 seconds, after which they'll be bored, not to mention annoyed that they're being made to do work the photographer should have done already.  As for the folks behind the camera, placing the focus is a creative decision, and in general, photographers aren't going to want the end viewer making their creative decisions for them.  Lytro knows this perfectly well, as evidenced by this picture from their website:




You'll notice that this is not a "living picture."  You cannot click it to change the focus.  Why ever not?  Because the photographer has a specific intent, an individual element of the picture on which they wish to focus your attention.  I'll give you a hint, it's not the chick in the background.

Let me be clear, I'm not skeptical about the value of being able to set focus after the fact, I'm only skeptical about the value of letting the viewer set focus after the fact.  Letting the photographer set focus after the fact is fantastic.  Any technology that shifts creative decision making away from the crucial moment of capture, and instead lets you make the decision later from the comfort of your couch, is hugely valuable.  If we didn't think so, we wouldn't bother shooting raw.  As such, it's easy for me to imagine that this functionality, if not necessarily this specific technology, will make its way into a wide variety of cameras in the future, and I suspect that Lytro's goal is not to become a consumer electronics manufacturer, but rather to license their technology to established camera makers, or perhaps be purchased outright, à la Foveon.  The camera is a proof of concept.

So, does it work?  Yeah, sure, I guess.  It has a tiny little sensor, so like any small sensor camera, pretty much everything is in focus pretty much all of the time anyway.  If you actually want to see the power of refocusing at work, you essentially need to compose your shot with that in mind.  You need to put a humming bird feeder a few inches from the lens, with San Francisco Bay at infinity in the background.  Lytro talks about how this camera eliminates the need to worry about focus.  But of course you want to take advantage of the cool thing that your camera can do, so you end up thinking more about focus when shooting with this camera, rather than less.  It would be a different matter if shooting with the Lytro were like shooting with a 50mm f/1.2 on your Canon 5D Mark III, where your shot of that beautiful woman can be ruined because you focused on the tip of her nose, rather than her eye.  Besides the fact that the Lytro can't achieve a depth of field that shallow, it doesn't have sufficient resolution to reveal missed focus that subtle.  That's not soft focus you're seeing, that's the maximum resolving power of the 1 megapixel image.

And both the camera and software have enough quirks, flaws, and oversights that shooting with the camera isn't always quite the experience you might hope.

For example, the camera has a tremendous zoom range, 8x, roughly 35mm-280mm equivalent.  But in order to avail yourself of the full zoom range, you need to put the camera in "creative" mode, which then requires that you, get this: focus before shooting the picture.  That's done by tapping on the screen, and is quite slow.  Positively glacial, actually.  You can still refocus after the fact, but only across a narrow range, thus the necessity to specify the center of that range at the time the picture is taken.  Why is this required?  I'm honestly not sure.  DP Review's article on the Lytro makes it sound as if, in non-creative mode, the camera is simply set to hyperfocal distance, everything is captured in focus, and portions of it are then presumably defocused later.  But maybe I'm not understanding them correctly.  Certainly Lytro CEO Ren Ng's dissertation does not make it sound like this is how it works, though my reading of his dissertation was admittedly cursory.

What else?  The screen on the camera is terrible.  You haven't seen an LCD screen this bad since the mid 90's.  Which is a real shame, because this camera attracts a lot of attention, and people always want you to show them its trick.  But the screen is so incredibly low resolution that you can't tell focused from unfocused except in the most extreme conditions, thus making it essentially impossible to demo the magic for interested bystanders.  Simply from a word-of-mouth marketing standpoint, it would have behoved Lytro to put a better screen in this thing.

The form factor of the camera, including the placement of its touch sensitive zoom strip (hyper-touch sensitive; I'm constantly zooming by accident) and its tap-screen-to-focus creative mode mean that shooting with this camera is pretty much a two-handed operation, despite its diminutive size.  And the lens cap attaches magnetically, which seems very slick until you realize how weak the magnets are and how easily the cap comes off.  I'm quite confident I'll lose it at some point.  (Update: yup, lost it.  To Lytro's credit, they sent me a replacement lens cap for free!  I'm going to tether this one.)

The software is extremely limited, which is fine, this is the first go, and presumably it will be expanded going forward.  But even allowing for its minimal functionality, it has some inexplicable usability flaws.  For example, you can't view your pictures at 1:1, or full screen (though you can once they're uploaded to the website - what?).  And it's twitchy.  Sometimes it fails to hold focus on the point I specify, and it's crashed on me a handful of times.

And here, perhaps, is the most telling thing of all: a week after I got the Lytro in my hot little hands, after having eagerly anticipated its arrival for almost six months, I took a weekend trip to New York.  Of course I brought the Lytro with me and guess what: I never took it out with me, not once.  I wasn't carrying a bag, it's too awkwardly shaped to fit in a pocket, and perhaps most importantly, I had a great little camera, the Fuji X10, slung on a strap around my neck.  When judged on conventional criteria, the image quality of the Lytro can't even compete with your phone.  Resolution is low.  Noise is high.  Dynamic range is poor.  I knew that if I ended up capturing something nice with the Lytro, I would wish that it had been taken with the Fuji.

And yet.

Beyond the obvious wow-factor of "shoot first, focus later," there are a lot of reasons to love, or at least be impressed by, this camera.  First of all, it's a beautiful little object, a characteristic I actually place a lot of value on.  The software is also attractive.  They've taken something that could easily have been overly complex, and made it overly simplistic, which is really quite an accomplishment.  And this whole "light field technology" thing enables some not-so-obvious things too:

The camera is really fast.  Shot to shot time, when you don't have to focus, is essentially instant.  It's liberating to have a point and shoot that shoots as fast as the quickest SLRs I've used.  It boots fast too.

While the lens does exhibit mild barrel distortion and pin cushioning, there's no chromatic aberration.  Why not?  Heck, I don't know, probably something to do with micro lenses.  I don't notice any vignetting either.

The 8x zoom lens is an f/2 across the entire zoom range.  A 35-280 f/2 zoom in a cheap little point and shoot.  Are you still shooting with that 5D Mark III?  A 200mm f/2 prime will cost you six thousand dollars.  I don't mean to suggest they're equivalent, but any technology that enables fast, small, cheap, long zooms with no chromatic aberration is pretty cool, never mind being able to focus later.

The reach of the optical zoom can be seen here when we go from full wide:


to full telephoto:


And there really are instances where focusing later is handy.  For example, it can often be hard to capture the right focus when shooting small, fast moving things close to the lens, such as hyper active dogs that charge whenever you hold something down at their level.  Picking focus later makes this shooting situation much easier.  I know I want the dog in focus, I really don't want you clicking on the house instead, so I'm not giving you a "living picture."  Instead, after setting the focus on the dog in Lytro's software, I export a JPEG, bring it into Lightroom to touch it up a bit, and then post it here.  Just like a regular picture, except without a blurry dog.


Lastly, I guess I have to admit, there are times when being able to to rack focus back and forth across a shot is kinda cool.






Then, of coures, there's the future.  And I'm not talking about improved hardware and better cameras, though surely that will come.  I'm talking about improvements to the Lytro software, which will enable you to do more with the pictures that you've already shot using the first gen Lytro you own today.  Lytro has already demoed the ability to subtly shift perspective in a picture.  Not just from one set view to another, as with the two images taken by a stereo camera, but continuously, albeit across a very small range.  I assume the same shift could be made along the Y axis as well.


This animated gif is from Lytro's blog, and it illustrates some intriguing possibilities besides just perspective shift.  Most obviously, the ability to shift perspective like that also means that 3D images could be created from single exposures.  Although I regularly boycott 3D movies, I'm a big fan of 3D photography.  Life is full of contradictions.  There are some other, less obvious, things about this photo as well.  Note, for example, that everything is in focus, the depth of field is infinite.  Currently the Lytro software only allows you to place the focal plane at the point of your choosing along the Z axis.  The software does not allow control over depth of field, but presumably it could.  The ability to create photos where everything is in focus, like the one above, would be cool, but cooler than that would be the ability to use a small sensor point and shoot to create a photo with as shallow a depth of field as you're getting at f/1.2 on your full frame SLR.  Either way, decoupling depth of field from aperture would open up some interesting possibilities.  And is there any reason that future versions of the software couldn't allow you to swing the focal plane, as with a tilt-shift lens?  If Lytro really does continue to enhance the software, this camera will be the gift that keeps on giving.

Despite the camera's shortcomings, shooting with it has been a lot of fun, and that's really the only reason I shoot with anything.  If you think of it as a regular camera, it's a disappointment.  If you think of it as an amazing photographic novelty, it's a blast.  Heck, if nothing else, it's a great conversation starter.

8 comments:

  1. Lytro showed some 3d images at the entertainment gathering last week... so, looks like you're right on that count.

    ReplyDelete
  2. I have a Lytro, and like you I'd preordered it months ahead of time. The best comment I've seen from anyone on the camera is "Owning a Lytro is like being a beta tester for the future of photography." The camera has room for improvement (and the software much more so), but it can, and possibly will, be a game-changing technology.

    I love mine. I've found plenty of places to get good shots, and if you want another place where the Lytro is surprisingly capable compared to expensive alternatives - the camera takes *fantastic* shots into and through glass and water. I've taken some stunning photos of things inside my girlfriend's saltwater reef tank; I don't think the camera lenses do any kind of polarization but zooming in and shooting through a half inch of glass and 6-10" of water, I get shots that look like they were taken with the lens just right up against the subject.

    Also, if you haven't experimented with the Lytro's macro capabilities, it's also a good feature. At 1x zoom in Creative Mode, your minimum focusing distance is just about the surface of the lens of the camera. I use my Lytro for macro photography a lot more than for the refocusing aspect of it.

    The camera image file format is also extremely hackable, and there are a few tools out there for easily disassembling the image formats. I would expect there will be a few tools to manipulate images outwide of Lytro's software. In theory, you should be able to do the same focus calculations on the raw image to be able to redefine the depth-of-field in software to be able to, say, convert that f/2 image into an f/8 DOF simply by refocusing the light rays in the distance range around the desired DOF.

    ReplyDelete
  3. This tweet by Ben Horowitz says that there is a lot more to the product than "what meets the eye" :

    https://twitter.com/#!/bhorowitz/status/83579227533873152

    "@om @orenjacob @mraccah @cdixon consumer hardware with mind blowing viral pictures as the marketing tool. Yes, new market. Gradual build."

    ReplyDelete
  4. A & H are investors in Lytro.

    ReplyDelete
  5. The poor stereo depth in most of the images of Lytro's 3D demo confirms that they haven't found a way to rewrite the laws of physics.

    Here's the problem, for those who aren't stereo buffs like Benjamin: 3D depth perception depends on your brain analyzing two different images, as seen from two different points of view that are physically separated in space. The brain expects that separation to be 65mm, because that's the distance between your eyes. If a stereo camera's lenses are closer together than that, the stereo effect is weakened--perceptually, the picture flattens out.

    Now, how far apart can any two images from a Lytro camera be? Well, the lens is about 20mm wide, so that answers the question. *At best*, a Lytro-generated stereo pair will have less than one-third the amount of depth that your eyes--or a conventional stereo camera--would see. There's no way around this inconvenient fact: you can't get good stereo if your images are taken from viewpoints that are close together.

    The only (partial) exception is in macrophotography. Macro stereograms need closer lens spacing, because otherwise you could end up looking at opposite sides of the object (like crossing your eyes and looking at opposite sides of your nose), so there'd be nothing the brain could fuse into a single image. Special cameras such as the old Macro Realist are designed to do this kind of closeup work using 16mm lens separation, but they are unusable for normal photos.

    Lytro is well aware of this limitation, although they aren't talking about it. All the stereo photos in their demo reel involve objects that are very close to the camera, starting with a row of Nixie tubes and continuing with closeup shots of flowers, babies and so on. See any landscapes there? Normal family snapshots? Nope. They'd look flat. And even what's there doesn't look too great from a 3D perspective.

    Bottom line: the Lytro camera is capable of low-res ultra-closeup stereograms, but when shooting scenes and objects at normal distances (more than a foot), the stereo effect is very weak, due to the unavoidable limitations of the single small lens. To pretend that this can be used as a general-purpose stereo camera is like pretending that a single microphone can be used to record stereo sound.

    ReplyDelete
  6. Wow! This will revolutionize the adult entertainment industry!

    ReplyDelete
  7. Sorry for the person above, but it doesn't even match up with other camera in terms of that kind of entertainment. You can just take pictures and select the focused area on that picture. I think that's what Lytro can do overall.

    ReplyDelete