You can click on any part of this Lytro photo to refocus on that part, from the eyelashes to the reflected image.
This plenoptic technology spun out of Stanford. They put an array of microlenses over the sensor, so in a single shot they grab a spectrum of focal planes. The effect is dramatic with an oblique macro shot like this. So you can choose the focal point after the fact. The sacrifice is that you get 1.5MP images from a 16MP sensor.
Interesting! I’ve been reading about this for months. First set of photos that I’ve seen.
I don’t yet see an application for this. I suppose the instantaneous capture of volumetric state using non-invasive optical methods could get interesting once you have dozens of these things aimed at the same volume so multiple points of view can give enough redundancy to eliminate occlusions. Interesting, but I still don’t see an application for that.
Now for moving pictures it makes more sense – microscopic movies where the thing being captured constantly moves out of the focal plane (like trying to do research into insect wingbeats without having to tether the insect) – being able to set the focal plane per frame of such a movie would be great.
So maybe a movie version will get used for the next Attenborough show if he goes microscopic! And maybe it could lead to new understanding of insect flight and thus even smaller DARPA hummingbirdbots.
I could see this being very useful for microphotography, especially of moving insects, something I’ve been tackling recently. It’s almost impossible to get an entire insect in focus, so the current technique is focus stacking. But that requires the subject stay still, something that living insects don’t like to do.
Does the Lytro web site give you some way to put the entire image in focus at the same time?
I could see this being used for security reasons. Rather than waiting for a camera to auto focus on a face for facial recognition or risk missing a glimpse of a suspect that is out of focus you have complete focus over the entire scene for every frame shot. Might be a stretch but you don’t risk losing any information that way. You also don’t increase the amount of raw data collected. we will have to see I suppose 🙂
If you take a picture with a 16 Megapixel camera without the special optics and then with the optics the filesize should be the same, right? Or am I missing something? Perhaps "raw data" was not the proper term to use….. 😉
But here you end up with a 1.5mp usable image from that 16mp sensor. If you only need 1.5mp for the face recognition app, you could use a 1.5mp sensor to start with. I guess all I am trying to say is that you capture 10x the data to get the after-the-fact focus feature. If you then pick one of those fixed-focus images, and discard the rest, then yeah, it’s probably the same data storage. But it sounded like you were talking about keeping it all so you could pick out different faces in the crowd for future analysis. Then it would be 10x (data or resolution hit depending on comparison chosen) versus a single narrow aperture shot where you hope the depth-of-field is good enough to see all faces.
Ah yes fair enough. There may be an application out there somewhere. It is incredible what people can come up with. Here’s to innovation!
Leave a Reply