You’d be forgiven if you perhaps hadn’t heard of Lytro before now. A relatively new company in the world of photography, Lytro manufacture consumer targeted - and indeed relatively affordable - light-field cameras.
Lytro’s first camera, the aptly named ‘Lytro’, perhaps didn’t generate the sales and reaction the company were hoping for. Liberties with the camera’s design, ostensibly to allow the light-field technology to work, didn’t endear consumers to its merits.
Their second camera, the ‘Lytro Illum’, has seemingly remedied this issue with a much more conventional and arguably user friendly design. A traditional style lens sits on a compact form factor body, making it strikingly similar to the Sony Alpha series. The only indication that the Illum has anything groundbreaking to offer is its awkwardly angled body - something which has potentially been avoided by other manufacturers for good reason.
That being said, the Illum packs a rather unique punch. The Illum (and the initial Lytro) use modified CMOS sensors and some clever physics to create a ‘light-field sensor’ instead. So what on earth is a light-field sensor and how is it different to a CCD or CMOS sensor? We’ll start with old money: the CCD.
A CCD, or Charge-Coupled Device, is a sensor comprising an arrangement of pixels in rows and columns. As an example, open a spreadsheet and imagine the spreadsheet itself is the camera’s sensor (a CCD in this case). Each cell of the spreadsheet is a pixel in the sensor. The megapixel count of your camera is related to this. A 20 megapixel camera would have 20 million pixels on its sensor. In the case of the spreadsheet example, it would contain 20 million individual cells - that’s a big spreadsheet!
Each pixel is essentially a bucket that collects light incoming from the camera’s lens whilst the shutter is open. When light strikes a pixel, it generates a small electrical charge in that pixel (for the real world analogue, it begins to fill the bucket). Once the shutter has been closed, the sensor is ‘read’ so that the amount of charge (or water in the bucket) in each pixel is analysed. At this point the camera’s processor interprets the information it has read and turns this into an image which is saved to the storage unit.
Most smartphone, tablet and indeed DSLR image sensors are not CCD based. They are in fact CMOS, or Complimentary Metal Oxide Semiconductor sensors. The CMOS style of sensor is, essentially, a more technologically sophisticated version of a CCD that uses a different process to read each pixel. The process uses transistors mounted to each pixel making CMOS sensors much more cost effective, as well as more efficient, hence their uses in devices that need to use power sparingly like smartphones and DSLRs. Telescopes, interestingly, use CCDs.
Regardless of sensor types discussed, the images they produce will always be 2-dimensional. Whilst we may perceive depth in the image, that depth is manufactured and interpreted by the brain and not embedded within the image. To understand why some theory is required.
We live in 3-dimensional space. Things have height, width and depth. Light is similar, only it is has many more dimensions. It has the 3 spatial dimensions (height, width and depth) but also further dimensions that describe associated quantities that make up the literal light-field. A conventional CCD can only capture information based on height and width as the CCD’s pixels all lie on a flat surface, hence why the image it produces is also flat (2-dimensional, as it lacks depth information). The Lytro’s light-field sensor cannot directly detect depth information as, much like the CCD, its pixels are once again on a flat surface. It can, however detect further information from the light-field that it then uses to ‘fake’ depth in the image using both clever hardware and clever software.
In terms of clever hardware, the light-field sensor uses an array of incredibly small lenses (Lytro market this as a microlens array) in front of the pixelated sensor so that the incoming light is broken up into individual light rays. This allows the sensor to analyse individual, rather than multiple, rays of light as would have been the case with a CCD. The microlens array also allows the camera itself to take images through each individual microlens, therefore generating information about each megaray. As a consequence of this Lytro market their cameras with a ‘megaray’ value rather than a ‘megapixel’ value. The megaray value is the number of individual light rays the sensor can process and doesn’t necessarily correspond to the megapixel value of the sensor. As such the Lytro could use a relatively small number of megapixels but still generate a fairly high megaray value.
In terms of the clever software, the light-field engine does something quite spectacular. Because a traditional camera cannot detect depth information, it relies on its user to define the position of the focal point prior to taking the image. The Lytro Illum has a fixed focal length and aperture (resulting in an f-number of f/2) so its focal point cannot be manually defined by the user. Thanks to the information that the light-field sensor captures however, the light-field engine can combine each microlens image and adjust the focal point and aperture after the image has been taken. On the topic of focal points, the Lytro Illum does allow the user to change the focal point using the integrated touch screen. The user simply taps the location of the desired focal point and the light field engine mimics the change in focal length required.
The light-field engine does not output standard JPEG or RAW files, preferring a proprietary file type called LFP: Light Field Picture. Through Lytro’s desktop application, Lytro Desktop, these files can be read and edited to allow the full functionality of the Lytro to be realised. These images can then be uploaded and shared through Lytro Web (https://pictures.lytro.com), Lytro’s web interface to allow rich interaction with its unique images.
All of this technology comes at a price, and a fairly hefty one at that: $1499. Moreover the sensor, although advanced, is limited to an effective resolution of 4 megapixels meaning that the dimensions of a conventional format Lytro image are small in comparison to its competition. Whilst bigger doesn’t always mean better, the rather limited functionality outside of the Illum itself and Lytro Web mean that your interaction with Lytro images is severely restricted. Coupled with the small image size reducing the likelihood that Lytro image JPEGs would appear in print format, the Illum may just be another expensive novelty item. It deserves to be more, but only time will tell if it gets what it deserves.
Author: Tomas James - Twitter