I don't think there is any current technology to put sensors so close together that there are no gaps between them. If there was it would be a much cheaper way to make large sensors. Arrays of sensors are used in for example astronomy, but then there are gaps between them.
Its done. I saw the writeup this morning (I'll have to find the webpage) it was ~96 compact camera sensors put together in this thing that looked like a WWII rotary aircraft engine. The wires looked like the heat fins. It was a single shot 50 gigapixel camera that stitched it all with custom software.
Well...no. It [butting sensors together without gaps] is not done, at all! That is not how the "50 gigapixel" device works to achieve a gap-less mosaic.
Read
the article linked to by Tim yesterday in the
LuLa thread here carefully. The giveaway is in this statement:
"We arrange for some overlap, so we don’t miss anything.”Even if you could butt the sensors so perfectly together that you completely eliminate any gaps, they would each give you distinct, non-overlapping images.
To get
overlapping images on different sensors - in the same snapshot exposure - requires that either:
* the very same photons are intercepted and recorded by two or more sensors positioned in the same focal plane (in other words, we enter the realm of magic, wizards and dragons
)
- or
* the sensors are not spatially coincident at all, but rather, different photons from different sub-beams within the entrance pupil are diverted to them.
Occam's razor favours the second explanation
.
You can indeed produce multiple or overlapping images of the same thing simultaneously, if you separately image different sub-pupils of the pupil plane - this is what a Shack-Hartmann wavefront sensor does, for example. A beam-splitter is an alternative way to do it, but that's not the technology used here.
What the Duke/Arizona gigapixel folks do, is they let the image pass
through focus, past what we would normally term the focal plane (here, it's a curved surface, rather than plane), and then re-image it as it fans out. This gives them the spatial clearance to space their re-imaging optics and sensors
far apart, even though some of the light that different "micro-cameras" are sampling originated from the same point in the field of view.
Also - it appears that
this design will only work with a single, fixed focal length objective lens. Their "Gigagon lens" reminds me of something like an old Hologon design lens, but it has with a strongly curved primary focal surface. I can't see that being interchangeable with anything else - they admit as much themselves: "
FOV is also strictly a matter of adding more cameras, with no change in the objective lens or micro-optic design". It's a marvellous design, but there won't be a selection of bayonet-mount interchangeable lenses for it
.
Ray