Discussion of all digital SLR cameras under the Minolta and Konica Minolta brands
Forum rules
No more than three images or three external links allowed in any post or reply. Please trim quotations and do not include images in quotes unless essential.
Dusty
I cam across this company via a post on DPreview... Lytro
its a small start up company that makes some very bold claims about miniaturising light field tech onto a camera sensor.. the long and short of it is you take the shot first, then focus afterwards in software! I'm skeptical but if it works (and gets refined over the course of a few years / generations of cameras) it could be a real game changer. They claim the photos in the gallery are real shots, not just mock up. Still, its fun to play with - just click on an area of the image to refocus.
Scott - must be 20 years ago I worked out a simple concept for using calculated, not real, reverse optical paths (reverse ray tracing) to recover sharp, grain free images from film exposures made deliberately out of focus. The system relies on having a taking lens of very precise characteristics, a moderate aperture (such as f/8), fixed at beyond-infinity focus.
It is impossible to create an infinite field of discrete point sources positioned behind this film exposure, as it is not possible to restrict each ray bundle (cone) to map itself to the aperture. So an optical light-based solution, even using coherent light induced to scatter at this plane, was never practical. I tried to work out how this could be done as a teenager and concluded that a very slow scanning process using a stepped, orientable laser might work.
Today, there's no reason to work with real light rays. A theoretical cartesian array of point sources can be positioned behind the sensor plane, at the true infinity focus plane position for the lens. A fixed number of rays from each pixel coordinate can then be mapped to the theoretical origin of every ray at the rear nodal plane aperture of the lens. There will be no stray ray crosstalk from other points, as this is not real light.
Modulating each traced ray by the pixel value it passes through at the sensor plane should enable the reconstruction of a sharp image. Moreoever, this sharp image can be refocused and it will be very noise-free.
But it's not my job to devise things which should occur to any imaging system development scientist over a cup of coffee... so I conclude my concepts are flawed. Otherwise it would already have been done and patented half a century ago.
