Light field photography

[prestone818]

another interesting concept...

http://www.engadget.com/2011/10/14/pano ... ously-thr/

[ngoodnight]

JayMiller wrote:

Bean wrote
You can set DOF in post

.

Jay wrote: Could you tell me where you found the information? I have checked their website and blog carefully and have only found references to focus and being able to switch beween 2D and 3D. Nothing about Depth of Field. Depth of Field would be a game changer.

Bean wrote: It's the entire point of the technology. It captures all the visible light, the DOF shown in the online images is purely a function of post-processing settings. It may or may not be available at this time but that's purely a software limitation, not hardware.

It may be a problem with software (i'm not convinced of that), the fact remains that apparently you can't currently change DoF with the Lytro and the Lytro is what is under discussion.

Changing DOF is a software problem in that it's possible to do entirely in software. However, that doesn't mean it's easy to do quickly and/or high quality. You can change DOF by collapsing multiple slices of the light field, but that's obviously more computationally expensive than a single refocus. You can also change DOF by integrating more/less of the angular sample information, but that means more image noise (for increased DOF). Regardless, the light field data contains enough information to change DOF.

I was a bit surprised that Lytro's demos didn't have a DOF knob. It might be that they don't want to increase image noise (for the fast algorithm) or pay the cost (for the multiple slice algorithm). Hard to say.

[Bean]

1) Terrible form factor and ergonomic design
2) Touchscreen only and tiny LCD (I personally hate this in a camera)
3) As far as I can tell little in the way of shutter speed control or manual control of any kind (no mention of max/min shutter speeds at all)
4) No ability to expand memory (no SD card slot)
5) No flash
6) No mention of metering modes and how it exposes a scene
7) No mention of what they mean by HD quality but if the examples are any indication your resulting image is very low-res

I could go on and on.

Your complaint seems to have gone from "it can't do what this array at the Stanford lab can" (because it can) to a number of complaints that are mostly subjective and regarding a lack of shared information. There's a ~177 page dissertation on their website from the guy's time in academia that probably has lots of the information you wish were shared.

You're right, this model is pretty limited (in a number of ways) and likely not an ideal choice for serious photogs; to look past the amazing technology behind it however would be absurd. I posted this because the technology is fascinating. Like you, I'm not lining up to buy one. Trivial spec limitations on a version 1.0 model do not concern me in the slightest.

[highpilgrim]

Changing DOF is a software problem in that it's possible to do entirely in software. However, that doesn't mean it's easy to do quickly and/or high quality. You can change DOF by collapsing multiple slices of the light field, but that's obviously more computationally expensive than a single refocus. You can also change DOF by integrating more/less of the angular sample information, but that means more image noise (for increased DOF). Regardless, the light field data contains enough information to change DOF.

I was a bit surprised that Lytro's demos didn't have a DOF knob. It might be that they don't want to increase image noise (for the fast algorithm) or pay the cost (for the multiple slice algorithm).

I was starting to think I could follow these developments...turns out I was woefully overestimating myself.

[highpilgrim]

I posted this because the technology is fascinating. Like you, I'm not lining up to buy one. Trivial spec limitations on a version 1.0 model do not concern me in the slightest.

Bean's right; The first computer, called ENIAC, was as big as a railroad boxcar. Now, any smart phone has capabilities so far beyond that computer that there is no basis for any comparison other than their shared lineage.

They'll work out the bugs. That's how things work.

This is cool technology, at its inception.