Bob
> If having three imagers is desirable for a better video camera, are
> there any three-imager still cameras? If not, why?
keeping three chips perfectly aligned is *very* difficult and it would
require a beam splitter which makes most lenses not work, there is
three times as much data to move which means storage and cpu
requirements triple, it's a lot more expensive and it doesn't offer
much benefit.
> If having three imagers is desirable for a better video camera, are
> there any three-imager still cameras? If not, why?
It's probably just too bulky and expensive considering the benefits.
Low-light performance is not as important in still cameras as in video
cameras, because the shutter speed isn't limited by the framerate.
http://en.wikipedia.org/wiki/Three-CCD
--
Regards, Robert http://www.arumes.com
There are cameras from Foveon which have special sensors which capture
all three colour components at each pixel, as if the camera had three
sensors.
--
Alfred Molon
------------------------------
Olympus E-series DSLRs and micro 4/3 forum at
http://tech.groups.yahoo.com/group/MyOlympus/
http://myolympus.org/ photo sharing site
Back in 2000, Hasselblad planned a medium format 16 MP DSLR with three
separate Foveon-made CMOS sensors, one each for red, blue and green.
It was to have been called the Hasselblad DFinity.
http://everything2.com/title/Hasselblad+DFinity
The project was never completed. Some prototypes were shown to a
select few journalists, and that was it.
Foveon went on to design and manufacture a three-layer sensor that is
used in Sigma DSLRs and Sigma P&S digicams.
> There are cameras from Foveon which have special sensors which capture
> all three colour components at each pixel, as if the camera had three
> sensors.
here we go again. foveon does *not* capture rgb. it measures three
overlapping ranges and calculates rgb from that. it has all sorts of
problems, and at the end of the day, the output is worse than competing
bayer cameras.
> Name one video camera that uses a Foveon sensor? Canon offers the XL-H1 with
> a 3-chip 1/3", approx. 1.67 megapixels per CCD, CCD.
the sigma dp1 and dp2 do video. it's only 320 x 240 though.
> Back in 2000, Hasselblad planned a medium format 16 MP DSLR with three
> separate Foveon-made CMOS sensors, one each for red, blue and green.
> It was to have been called the Hasselblad DFinity.
>
> http://everything2.com/title/Hasselblad+DFinity
>
> The project was never completed. Some prototypes were shown to a
> select few journalists, and that was it.
it was on display at the photo plus expo in new york (with a *really*
attractive model, but i digress).
The OP was asking for a still camera.
And in fact I wrote "all three colour components".
By the way, you should stop your ridicolous crusade against full colour
sensors. The Foveon implementation may have its weaknesses, but there is
no question that capturing the full colour information at each pixel
delivers way superior results.
One day all cameras will use full-colour sensors. Nobody will use Bayer
sensor cameras anymore.
> > foveon does *not* capture rgb. it measures three
> > overlapping ranges and calculates rgb from that.
>
> And in fact I wrote "all three colour components".
it does not measure any components, it calculates them. if bayer
'guesses' missing data (according to foveon fans), so does foveon. in
fact, foveon guesses much more than bayer does (and is measurably less
accurate).
> By the way, you should stop your ridicolous crusade against full colour
> sensors.
i have nothing against full colour sensors. what i have a problem with
are the lies from sigma/foveon. some of the stuff they claim is
physically impossible and even contradicts their own patents. deception
doesn't even begin to describe it.
> The Foveon implementation may have its weaknesses, but there is
> no question that capturing the full colour information at each pixel
> delivers way superior results.
only *if* they don't compromise other more important aspects, such as
resolution, noise, dynamic range, etc.
right now you have a choice between 4.7 megapixels on foveon or 18
megapixels bayer (of similar crop factor). the image quality of the
bayer sensor is so far ahead it's not even funny.
if someone comes up with an 18 megapixel full colour sensor, then it
might be worth a second look.
> One day all cameras will use full-colour sensors. Nobody will use Bayer
> sensor cameras anymore.
maybe, but it won't be that soon. physics gets in the way and bayer is
actually very good.
It actually sucks a lot, it's just that you don't know.
> > bayer is
> > actually very good.
>
> It actually sucks a lot, it's just that you don't know.
right, all those gorgeous images with lots of detail and accurate
colours had me confused.
The loss of resolution is the problem. But you already know this.
> > right, all those gorgeous images with lots of detail and accurate
> > colours had me confused.
>
> The loss of resolution is the problem. But you already know this.
what loss? an 18 megapixel canon 7d has far more resolution than the
4.7 megapixel foveon could dream about. even the most diehard foveon
fans admit that.
even the 12 and 15 megapixel slrs blow away what foveon can do and
those cameras will probably be replaced with higher resolution sensors
within the next few months (pma is in february).
step up to a full frame slr and it's even worse. 21-24 megapixels,
versus 4.7 and the price of full frame keeps getting lower with every
generation.
maybe one day there will be a full colour sensor that is as good or
better than bayer, but as i said, it won't be very soon.
>> If having three imagers is desirable for a better video camera, are
>> there any three-imager still cameras? If not, why?
> keeping three chips perfectly aligned is *very* difficult
... though maybe someone will come up with a method to use a
sensor-moving IS technology to allow self-adjusting the
chips.
> and it would
> require a beam splitter which makes most lenses not work,
Put the beam splitter where your classical DSLR has it's beam
splitter (half mirror) and forego the classical viewfinder
(and fast phase detection AF).
> there is
> three times as much data to move which means storage and cpu
> requirements triple,
JPEG doesn't care if 2/3rds of the values for each pixel are
interpolated or from a real sensor. Only RAW would triple in size.
And you'd not need the interpolation routines (which have to be
quite clever --- and think how much more time they need if you
do RAW processing on your PC) so while I/O speeds would need to
triple (by using 3 paths for 3 sensors, instead of one from one)
CPU needs might actually drop.
And since you don't need interpolation, you'd make do with less
pixels for the same resolution, so I/O doesn't need to triple,
it needs only slightly more than double.
You'd be able to use a much weaker AA filter, too.
And with proper beam splitting you don't need to loose more than
small amounts of light to the splitting-by-colour, which could
increase sensitivity especially for the red and blue channel ...
> it's a lot more expensive and it doesn't offer
> much benefit.
That's probably the problem.
-Wolfgang
Yep, we did, and the 4.7 M pixel positions Foveon rivals a 6 or
7 MPix classical bayer filter (RGGB or BGGR) sensor for resolution.
As any Google search would show. As any Google search would
show, there are plenty cameras with 8 or more MPix bayer sensors.
As any Google search would show, Foveon has bad colour separation
and no higher ISO qualities to speak of.
Shall I go on reading you what you should have googled yourself?
-Wolfgang
<snip lots of good points>
>> it's a lot more expensive and it doesn't offer
>> much benefit.
>
> That's probably the problem.
Plus it would be huge with three APS sensors and perhaps even need
special lenses with a longer flange distance.
It seems they overcame the speed issue with video though that's pretty
much solved for bayer FF 35mm now.
The point isn't whether it _can_ be done. It's whether it's worth the
trouble and expense.
--
Ray Fischer
rfis...@sonic.net