Lytro light field camera Since before the days of Fox Talbot, cameras have worked like the human eye. A lens focuses an image on a plane, be it a retina, silver halide or electronic sensors. The Lytro is different Lytro light field camera Lytro: the box camera for the 21st century Instead of capturing a single image it captures the rays of light, …
Making your first proof-of-concept simple to use and visually appealing, with a clear aim at the most receptive segment of the market... seems like Lytro choose the best strategy available to them!
But yeah, it is a toy. Prove me wrong, kids!
I remember a Mac-only product being released some years back, called the iPod... I wonder what ever happened to it?
I remember follwoing this on dpreview a good few months ago. Lytro have had a good 6-12 months to covert their OSX software to Windows but there seems to be no urgency. The fact they appear to be working on an iPad version instead unfortunately says a certain amount about their mentaility and business sense.
Nice idea that came out of a research project, but potentially with the wrong people running the company and niche product written all over it.
"says a certain amount about their mentaility and business sense"
Actually they have quite good business sense. Make a mediocre product, tart it up in fancy packaging, lock users in to your service, market to a demographic renowned for massive amounts of cash and very little sense = PROFIT!
Industrial light-field cameras already exist - Raytrix.de - and Stanford are continuing research in this area by doing strange things to a Nikon microscope. Doubly cunning, because the illumination system it uses is also a computer controlled light-field.
http://en.wikipedia.org/wiki/Light-field_camera#Manufacturers_of_light_field_cameras
This is just the first generation. If they can build a more egronomic body including things like micro/SDHC slot, a removeable battery and perhaps an electronic viewer (if only to save on battery and not look like you're using a mini-telescope) then it can move beyond being a niche product. Need to find some way to work in low light though. Not so bothered about the 3D though. Video would be a more useful feature to have.
Changing focal point a lot will rarely be that helpful as far-away things will be too small when selectively focussed (or alternatively, close things tool arge): scene size limits win out.
however, were the tech here in a hybrid with a normal camera, allowing a selected focus point in a selected scene to be tweaked after-the-fact, and/or depth-of-field adjusted, then it could be useful indeed.
"were the tech here in a hybrid with a normal camera..."
It seems that Mitsubishi have the same idea as you, with a film that sits in front of a normal sensor:
http://en.wikipedia.org/wiki/Light-field_camera#Manufacturers_of_light_field_cameras
The issue is this: Most compact cameras have lenses that distort the image, a barrel distortion that is then corrected for either in-camera (JPEG output) or in RAW conversion software. AFAIK, this is why the Lytro has a long body - a genuine undistorted 8x zoom. Making the sensor bigger means everything else gets big too.
When or whether anyone would come up with a practical way of implementing this.
I can think of lots of potential uses. As a photographer you choose where to focus in a picture. Later on someone might be interested in a different part of the picture to you and wished to have something else in focus.
Presumably when converting these images to a standard image you could chose different focus points for different parts of the picture. So I love to show faces in really low light, F1.4 on a 50mm or 85mm lens, well at the point the depth of field in measured in millimetres, at least your likely not to be able to get both eyes in focus. Personally I like the effect, but I can see why people might want to generate a print where both eyes were in focus and you then allow the rest of the picture to go soft as it would normally do.
Other reviews suggest there is no shutter lag. Shutter speed varies between 1/15 and 1/250 depending upon light- and entirely controlled by camera.
So yeah, it could be suitable for instant shots. The Nikon J1 takes a different approach to the same issue: very fast autofocus, 60 fps full resolution burst mode, and buffering (so that your series of burst shots starts before your press the shutter release).
It's been many, many, many years since man first thought about how to capture an image - make a hole, have a recording media screen, an exposure time, etc - been stuck like that since the start. This is a crazy progression IMO, sure at the moment it might not be 'great', but jesus, this could just be looked at as a proof of concept. How long did it take digital cameras to lift themselves up to decent quality? Were they great from the start - were they hell!
The form factor will be the biggest issue - will increasing the resolution result in a thicker tube? Can this be rectified by making it collapsible? Etc. Beyond that obvious disavantage the potential advantages are huge. Every item in a photo in focus is surely a possibility with some software cleverness.
Great to see such innovation
"Great to see such innovation" ...and the death of creativiy!
"Every item in a photo in focus is surely a possibility with some software cleverness."
You obviously don't enjoy photography as a hobby, part of the enjoyment is creating something special, a personal one-off using the limited abilities of the kit in front of you and that often includes clever use of "depth of field". You deliberately blur some parts of the image in front and behind the subject, to highlight that subject more clearly. Portraits on busy streets are a prime example of very short DoF., you want people to see the pretty girl's face and get distracted by the dirty black cab, the builders leering from the white-van or the wino pissing in the corner!
Every item in a photo in perfect focus will lead to a massive glut of photos that all look the same not matter who took them, they will have no interest other than to the person who took them and maybe 3 mates on Facebook.
I like the idea behind the tech but sometimes it's good to make some things in life a little tricky as it helps you to learn a skill and by learning something you stimulate ideas and progress your creativity.
"struggle with the finer points of photography"
Since when has focusing been one of photography's finer points? It's either in focus or its not and then its simply blurred. The Lytro simply lets you make the decision of the focal point after you've taken the photo.
I'd guess the Lytro will appeal to women, means they can change their mind after the event....
Further reading on the underlying science allegedly involved:
http://graphics.stanford.edu/papers/lfcamera/lfcamera-150dpi.pdf
The original thesis (all 200+ pages) on which the product is based is at
http://www.lytro.com/renng-thesis.pdf
If you know a bit about optics and/or signal processing, have a look and see what you make of the original work, before the people in Lytro's marketing department shined it up and dumbed it down.
"The original science is fine."
Is it though?
I happily agree a great deal of it is fine (I did a physics degree and have had a subsequent passing interest in sensors and signal processing).
But then (based on recollection from last time I looked at this) some of the papers start talking about the *phase* of light in this picture.
If these folk expect light sensitive devices to be able to work meaningfully with phase in situations using normal (not coherent e.g. laser) light then either I've missed something fundamental or they ARE pulling the wool over people's eyes.
For comparison, working with phase in "sound field" exercises is relatively trivial.
Great to see a low cost lightfield camera on the market .
Negative. Without specifications or API support to see what the camera is actually recording and enable third party software to access the photo files all very limiting. A glimpse of the future maybe but an expensive novelty item ofr the time being.
With $50m invested in the company according to tyro.com, surprising they haven't spent a few quid to develop a Windows UI for their software yet..
I'll quite possibly have to eat my poorly-focussed hat on this, but I would be amazed if this ever became anything more than a novelty. It cannot squeeze down to a practical depth to fit into a phone which is the big market for novel photo apps - and that's a physics limit more than an engineering limit. Even with the absurd rate of increase of pixel density in sensors, you still lose pixels geometrically with this system - crudely speaking, N planes of focus means you only get an Nth of the raw pixels of the sensor in the final image(s) - so you hit the noise limit for tiny pixels N times faster than with a 'normal' camera. And there are far less "clever" but far more practical methods of after-the-fact focussing. The small sensors and lenses on compacts and cameraphones effectively give infinite depth of field already and it isn't rocket science to measure and add a depth parameter to the raw image.
Most annoyingly of all, the focus on most Lytro images is terrible. The focal plane is never truly sharp regardless of where you try to refocus the image.
Light field image capture is a super clever solution to a problem that simply doesn't exist.
That said, five or so years ago when this was first proposed I brashly declared that it would never be anything more than vapourware so I don't have much of a track record here.
I wouldn't worry too much - look how long Duke Nukem was vapourware, and look at the resulting product. Just because it has been in development a long time doesn't mean it's good.
Narrow depth of field images have very limited uses - mostly arty effects in film and art photos, and food blog photos by hipsters.
For the rest of us, we prefer images actually in focus - I can't think of too many times I've actually WANTED to have to sit down and focus my pictures, badly, after the event.
I can imagine doing that to the alleged 750 pics on the camera would result in an afternoon of spectacular tedium. Coupled with only even being able to view them with the special software it looks like too much trouble.
I can see how media-arts-and-food-blog hipsters who find aperture settings hard to remember will love it.
Take two of those, add eye tracking and you get the first fully realistic 3D viewing technology.
I've been thinking where would the next jump in the required video bandwidth come from (because you can't just add more and more resolution - it's getting silly already as it is) and that is a possible answer. A stereo movie where each frame is a pair of these "light fields" will reanimate every holographic storage idea there ever was...
It's not really a world changing device, but the technology could solve the major issue I have with 3D films. I came out of Avatar with a banging headache, because in a 3D (or 2D) film the camera can only focus on one thing (or one distance). Therefore, if you try to look at anything else (like the background) your eyes try to focus as though you're looking at something that's further away, but since the background of the film image is blurred your eyes can never find a point of focus - however hard your eye muscles try - resulting in a serious headache. It is for this reason that I have always said 3D film will never really work, but with this technology (paired with a way of analysing where a viewer is looking) perhaps it could...
Nice to know other, obviously more professional, versions exist. Anything with "price on request" usually has me wondering though. It's not as though I'm working for the Croesus foundation.
It's also quite apparent that this gadget is a version 1.0 for Lyto. I hope to see improved versions soon, once the market shows where the money is.
Re the Apple-only compatibility: It was obviously a conscious choice for Lytro, whatever their reasons. They just might have identified their target audience, or whatever else. I don't care, I like the camera anyway.
On its own, this is a very cool technology. Combine this with high dynamic range, throw enough resolution and field-of-view, and you might get a camera that can do things like the Grand Canyon justice.
But think of this: cameras like this, coupled with the sound field big-array-of-microphones, and you get a tech to make Big Brother cream his pants: a system that could watch over a large public space, recording everything, and allowing a Ministry of Privacy officer to, after the fact, select anybody in the crowd, and have clear video and audio of that person.
And how big would you want this camera to be? : D It uses roughly 150 sensor pixels to record a single 'pixel' output, since it needs to know not only the final destination of a ray, but its direction, too. To get DSLR levels of resolution in the output would require a massive sensor (or else a dense, noisy sensor) and optics scaled-up to match- though this is less of an issue with a fixed wide-angle lens.
For the Grand Canyon - and other stationary subjects - you'll be better off with a conventional DSLR, a tripod and a focus-bracketing script - then some messing around in post-production.
"Works fine for still scenes, very poorly for ones with any movement"
So have multiple sensors taking pictures simultaneously, sharing the same basic lens config, with secondary compensating lenses so the different sensors have different effective focal lengths and therefore different focal planes. If you can make those secondary lenses small enough then the multiple sensors can actually be different areas on the same sensor, and all you need to pick between the various image focal planes is a bit of post-processing of the resulting image.
Hey look, Lytro explained, without all the pseudo-scientific psychobabble. Thank you! [Or maybe not?]
no that's not the point, the surgical distance is the point you set as lens to give the perception of the deepest area in focus. on a wide angle the area in focus is "greater" because of what is called s reduced circle of confusion. to recreate lyto all you would need is to to pictures in rapid succession like. at .5 metres 6 metres and 25 metres, then Mark the images as "one" where all items are in focus and the choose the focal point later in software. the hyperfocal distance is lens and aperture dependent as well as how you view the image because in reality there is only one focal point and this circle of confusion gives an parent depth of field. in lower light you would need more separate images because to get the image the hyperfocal distances would be lower and a telephone lens needs many more shots but then the software could be limited to a primary area like the entire head for a portrait. trust me i was a physicist then a photographer. a low light portrait would be great because you'd shoot the head but then select the eyes and nose tip in software and omit the background and ears from the focal zone.
I think that this is a great idea. However what have only one focus point?. The most annoying thing about focus is having to pick one subject. The possibility of having more than one subject in the picture but both in focus would IMHO be ground breaking. We will have to wait until Lytro 2.0.
By about page 2 I was thinking:
"wow, a GBP$400 camera! So cheap! Why not buy TWO? Shiny colours, square design, super minimal buttons, no apparent use except to wealthy hipsters doing media or arts degrees, can only view pics on special website; surely these guys thought to pitch it to Apple."
Then I read on page 3 - "At the moment there is only a Mac version available"
Close enough.
Why not just encode a z-channel into normal photos?
I bet most decent modern DSLRs with multi-focus sensors could do this, I am sure it could be rigged up with an IR laser too.
The film & special effects industry has been doing this for 20 years or so by encoding the z-channel (image depth) into images. The pictures are rendered with an aperture of infinity (hey, because you can!) which makes everything crystal clear, then you choose what you want your depth of field to be, and where, Gaussian-blur filter by z-channel mask and presto!
Same effect, easier, quicker, all the advantages of DSLR or even PAS cameras, standard image format (TIF) and images only 25% larger (4 channels instead of 3).
Hmmm, maybe I should patent that.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
