gjl711 wrote in post #12644000
That assumes that all the camera is doing is focus stacking with a single lens the way it is done today. From what I understand, that is not how this is works and I'm thinking that this would work in a SLR as well after reading up a bit
This is a plenoptic camera
and it not only is capable of recording the intensity of the light but via an array of micro lenses over the sensor, the direction as well. The image is then generated via some heavy duty processing. So in essence an infinite number in interim steps are generated.
The more I'm reading, the more I wish that they had gone the license route. This could really be interesting.
Yes, if we had sensors to cope with this.
Looking at his PhD dissertation, this effectively is a variant on faceted lenses, behind the lens rather than in front of it. A bunch of pixels/sensels on the sensor is serviced by a single lens, at a appropriate distance from the sensor to focus a sharp image. The image used is the one projected by the lens.
This also means that although by heavy computing and ray tracing it is possible to "focus" at different areas in the object field projected on the sensor, resolution will be significantly lower of what a sensor normally is capable of. Effectively, a 25 sensel array behind each lens will cut down resolution by a factor 5 (square root of number of elements), and this also means that in order to get good enough resolution for large size printing, 100 MP+ sensors are required.
AFAIAC, the reason to start with consumer cameras first, is because it is possible to get an ~2MP image from a reasonably sized smaller sensor this way, good enough for web viewing and small comsumer prints, certainly not enough for high quality printing purposes.
The shots used so far in their examples, were created with a MF back AFAIK, and an experimental microlens array to replace the original one, which explains the fairly shallow DoF in the first place. With the Lytro approach, effective apertures will become approximately F/4, due to the theoretical design limits imposed by this system, which also means that with compact cameras, almost everything should be in focus already.
Interesting concept, but I think we need to wait and see how this really turns out.
BTW, with some special software it is already possible to do this type of thing in the high quality realm too. However, this requires taking several pictures at different focusing distances, hence no moving objects allowed in the time required to take all shots. Here is where the Lytro has its advantage - 1 exposure for all of those shots.
Kind regards, Wim