A future technology for dSLRs?
Ray Fischer
it might not be too far a stretch to include this technology as well.
It would deal with the Bayer sensor's shortcoming and probably end the
use of the Foveon sensor.
Hasselblad has announced a Multi-Shot (MS) version of its H3DII-50
medium format camera. First shown in the H3DII-39 MS in 2008, the
system captures four shots in a row, moving the sensor by one
pixel between each shot to record full RGB values at each
position. THe H3DII-50 MS costs 23,000 euros with less expensive
trade-in prices and a trade-up route for Hasselblad owners.
http://www.dpreview.com/news/0911/09111902hasselbladh3d50ms.asp
--
Ray Fischer
rfis...@sonic.net
Charles
"Ray Fischer" <rfis...@sonic.net> wrote in message
news:4b084ed5$0$1642$742e...@news.sonic.net...
> Given that SLR makers are already incorporating piezo dust shakers,
> it might not be too far a stretch to include this technology as well.
> It would deal with the Bayer sensor's shortcoming and probably end the
> use of the Foveon sensor.
Interesting idea. I looked closely at the armor shot and could see more
detail. But, it was not a drop-your-socks event.
How does one keep the required image shifts to exactly one pixel? In two
dimensions? Over a period of time? After the camera has been thumped once
or twice? The cost disadvantage is obvious (at least at this point in
time).
I'm still rooting for Foveon II, as it seems more realistic to me as a way
to leave (progress beyond) Bayer demosaicing.
Thanks for the post.
nospam
<charles...@comcast.net> wrote:
> I'm still rooting for Foveon II, as it seems more realistic to me as a way
> to leave (progress beyond) Bayer demosaicing.
don't hold your breath on that one, and bayer actually works quite well.
Charles
"nospam" <nos...@nospam.invalid> wrote in message
news:211120091637482900%nos...@nospam.invalid...
Agreed as to both points. However, the Foveon technology intrigues me with
its elegance ... just like nuclear fusion. Probably both great ideas that
will never actually pan out at any practical level.
nospam
<charles...@comcast.net> wrote:
> >> I'm still rooting for Foveon II, as it seems more realistic to me as a
> >> way
> >> to leave (progress beyond) Bayer demosaicing.
> >
> > don't hold your breath on that one, and bayer actually works quite well.
>
> Agreed as to both points. However, the Foveon technology intrigues me with
> its elegance ... just like nuclear fusion. Probably both great ideas that
> will never actually pan out at any practical level.
foveon is anything but elegant. it doesn't actually sense pure red,
green and blue like in the ads. there's a lot of overlap in the three
layers, requiring a *lot* of processing to produce r/g/b, and because
of that, the results are less accurate than bayer with a lot more
noise, even in ideal conditions. it's tough to break the laws of
physics.
ransley
It sounds great, but what does any movement of the camera do to the
quailty of the photo.
Charles
"nospam" <nos...@nospam.invalid> wrote in message
I was not clear enough: the "idea" is elegant. Future vertical processing
of semiconductors offers hope in this arena. There is a slim, but real,
possibility that defined energy gaps could be built into a vertical
photosensor.
Planar to vertical and planar plus vertical might just give us some
wonderful new sensor technology. Or not.
Ray Fischer
>On Nov 21, 2:34�pm, rfisc...@sonic.net (Ray Fischer) wrote:
>> Given that SLR makers are already incorporating piezo dust shakers,
>> it might not be too far a stretch to include this technology as well.
>> It would deal with the Bayer sensor's shortcoming and probably end the
>> use of the Foveon sensor.
>>
>> � � Hasselblad has announced a Multi-Shot (MS) version of its H3DII-50
>> � � medium format camera. First shown in the H3DII-39 MS in 2008, the
>> � � system captures four shots in a row, moving the sensor by one
>> � � pixel between each shot to record full RGB values at each
>> � � position. THe H3DII-50 MS costs 23,000 euros with less expensive
>> � � trade-in prices and a trade-up route for Hasselblad owners.
>> � �http://www.dpreview.com/news/0911/09111902hasselbladh3d50ms.asp
>
>quailty of the photo.
Dunno. The fallback is to do the usual Bayer processing. Another
option is to spit out an error.
Sure, it's a feature with limited need, but I can see it being a cheap
add-on to an existing camera.
--
Ray Fischer
rfis...@sonic.net
nospam
<charles...@comcast.net> wrote:
> I was not clear enough: the "idea" is elegant.
the idea is interesting, but it isn't really solving a problem that
needs to be solved. the eye can't see the added colour resolution.
> Future vertical processing
> of semiconductors offers hope in this arena. There is a slim, but real,
> possibility that defined energy gaps could be built into a vertical
> photosensor.
nikon has a dichroic mirror patent but it's going to be a bitch to
manufacture. fuji was working on organic dyes but i think they gave up
on it. canon had a controllable filter, but it would require multiple
exposures. and nobody mentions that the data processing and storage
needs go up threefold (at least) with one of these systems.
meanwhile, bayer works.
Alfred Molon
> the idea is interesting, but it isn't really solving a problem that
> needs to be solved. the eye can't see the added colour resolution.
difference.
--
Alfred Molon
http://www.molon.de - Photos of Asia, Africa and Europe
Alfred Molon
Hasselblad implementing it, if there is no need for full colour
information at each pixel?
--
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
nospam
Molon <alfred...@yahoo.com> wrote:
> A full colour sensor, an idea some people here hate so much.
it's not that people hate it, it's that there's a lot of drawbacks that
some people blindly dismiss. a full colour sensor would be better *if*
it didn't compromise on high iso, dynamic range, frame rate and colour
accuracy. it also triples the storage and processing requirements.
foveon loses on all of them. nothing is perfect, and bayer is an
excellent tradeoff of what is currently possible.
> So why is
> Hasselblad implementing it, if there is no need for full colour
> information at each pixel?
they're moving the sensor and combining the results. that's not the
same as a full colour sensor and it certainly won't work very well for
anything with motion in it. it's also not cheap.
nospam
<alfred...@yahoo.com> wrote:
> > the idea is interesting, but it isn't really solving a problem that
> > needs to be solved. the eye can't see the added colour resolution.
>
> Nonsense. Zoom or get close enough and you will be able to see the
> difference.
i assume by zoom you mean pixel peep, and if you do that you can see
all sorts of problems. why not look at photos normally? did you grain
peep film too?
the eye's colour resolution is about 1/10th its luminance resolution
and bayer is 1/2. in other words bayer already has more colour than the
eye can resolve, so there's not much point in making it even higher.
Wolfgang Weisselberg
> A full colour sensor, an idea some people here hate so much. So why is
> Hasselblad implementing it, if there is no need for full colour
> information at each pixel?
Well, colour still photography and colour film by shooting through
3 different filters is actually a very very old technology.
It never worked well with moving subjects, why should it work
better for Hasselblad? Do they change the laws of physics?
As for why H is doing it ... because people like you decide that
photographers have to have full colour sensors or you won't use,
buy, etc. their shots. So everyone must buy H.
-Wolfgang
Mr. Strat
<rfis...@sonic.net> wrote:
> Given that SLR makers are already incorporating piezo dust shakers,
> it might not be too far a stretch to include this technology as well.
> It would deal with the Bayer sensor's shortcoming and probably end the
> use of the Foveon sensor.
There's only one company using the Foveon. Maybe Sigma could die also
along with their crappy technology.
Outing Trolls is FUN!
>In article <he9nis$9ki$1...@news.eternal-september.org>, Charles
><charles...@comcast.net> wrote:
>
>> >> I'm still rooting for Foveon II, as it seems more realistic to me as a
>> >> way
>> >> to leave (progress beyond) Bayer demosaicing.
>> >
>> > don't hold your breath on that one, and bayer actually works quite well.
>>
>> Agreed as to both points. However, the Foveon technology intrigues me with
>> its elegance ... just like nuclear fusion. Probably both great ideas that
>> will never actually pan out at any practical level.
>
>foveon is anything but elegant. it doesn't actually sense pure red,
>green and blue like in the ads. there's a lot of overlap in the three
>layers,
About as much overlap in the filters on a Bayer sensor.
Go educate yourself, troll.
Alfred Molon
> i assume by zoom you mean pixel peep, and if you do that you can see
> all sorts of problems. why not look at photos normally? did you grain
> peep film too?
No. Make an enlargement and you will see how much resolution a photo
really has. And if you can't make enlargements, what is the point of
having a high pixel count?
Alfred Molon
>
> > A full colour sensor, an idea some people here hate so much.
>
> it's not that people hate it, it's that there's a lot of drawbacks that
> some people blindly dismiss. a full colour sensor would be better *if*
> it didn't compromise on high iso, dynamic range, frame rate and colour
> accuracy. it also triples the storage and processing requirements.
> foveon loses on all of them. nothing is perfect, and bayer is an
> excellent tradeoff of what is currently possible.
We are not discussing the implementation, only if it would be preferable
to have full colour sensor.
> > So why is
> > Hasselblad implementing it, if there is no need for full colour
> > information at each pixel?
>
> they're moving the sensor and combining the results. that's not the
> same as a full colour sensor and it certainly won't work very well for
> anything with motion in it. it's also not cheap.
It's for static scenes of course, but it is effectively a full colour
sensor, regardless of how they collect the full colour information. The
only thing they have to add is the piezo actuators, so the additional
cost will be small.
David J Taylor
news:MPG.257321408...@news.supernews.com...
[]
> It's for static scenes of course, but it is effectively a full colour
> sensor, regardless of how they collect the full colour information. The
> only thing they have to add is the piezo actuators, so the additional
> cost will be small.
> --
>
> Alfred Molon
.. and probably something which could be easily added to any sensor-shift
camera at little cost, perhaps providing /some/ image quality improvement.
It should have better colour accuracy than the Foveon implementation.
Whether there are many lenses that would justify the potential quadrupling
of the resolution is another question. To get that resolution means that
you would want a less severe anti-alias filter, or even omit the AA
filter, which could then mean that the camera was more likely to see Moirￜ
artefacts when not using multi-shot mode.
Cheers,
David
Better Info
>filter, which could then mean that the camera was more likely to see Moir�
>artefacts when not using multi-shot mode.
>
>Cheers,
>David
If the sensor is shifted in sub-pixel increments, then much higher
resolutions are available for both chroma and luma.
Alfred Molon
says...
> If the sensor is shifted in sub-pixel increments, then much higher
> resolutions are available for both chroma and luma.
Hasselblad are not moving the sensor by sub-pixels, but by full pixels.
Alfred Molon
Taylor says...
> Whether there are many lenses that would justify the potential quadrupling
> of the resolution is another question.
the full nominal resulution of the sensor (as opposed to Bayer sensors,
where the effective resulution is somewhere between ~ 30-100% of the
pixel count, depending on the scene captured).
In other words with their 39MP sensor, with this shift technology they
obtain an image with 39MP of information, while with a Bayer sensor you
are somewhere between 10 and 39MP of information. The numbers are of
course just guesstimates.
David J Taylor
> In article <CJ7Om.7526$Ym4....@text.news.virginmedia.com>, David J
> Taylor says...
>> Whether there are many lenses that would justify the potential
>> quadrupling
>> of the resolution is another question.
>
> Hasselblad are not quadrupling the resolution, they are just achieving
> the full nominal resulution of the sensor (as opposed to Bayer sensors,
> where the effective resulution is somewhere between ~ 30-100% of the
> pixel count, depending on the scene captured).
> In other words with their 39MP sensor, with this shift technology they
> obtain an image with 39MP of information, while with a Bayer sensor you
> are somewhere between 10 and 39MP of information. The numbers are of
> course just guesstimates.
> --
>
> Alfred Molon
So 10 => 39 isn't potential quadrupling? Even as a guesstimate? <G>
My question about lenses stands - how many are actually good enough the
reach the full resolution (rather than the lower "Bayer resolution") of
today's high-end DSLRs, even with ideal taking conditions?
Cheers,
David
Andrew Haley
> "Ray Fischer" <rfis...@sonic.net> wrote in message
> news:4b084ed5$0$1642$742e...@news.sonic.net...
> > Given that SLR makers are already incorporating piezo dust shakers,
> > it might not be too far a stretch to include this technology as well.
> > It would deal with the Bayer sensor's shortcoming and probably end the
> > use of the Foveon sensor.
> Interesting idea. I looked closely at the armor shot and could see
> more detail. But, it was not a drop-your-socks event.
> How does one keep the required image shifts to exactly one pixel?
> In two dimensions? Over a period of time? After the camera has
> been thumped once or twice?
A piezo actuator. This is actually a pretty old idea: Imacon, now
part of Hasselblad, have been doing it for many years.
Andrew.
nospam
is FUN! <ot...@trollouters.org> wrote:
> >foveon is anything but elegant. it doesn't actually sense pure red,
> >green and blue like in the ads. there's a lot of overlap in the three
> >layers,
>
> About as much overlap in the filters on a Bayer sensor.
significantly more. see the foveon patents.
nospam
Molon <alfred...@yahoo.com> wrote:
> > i assume by zoom you mean pixel peep, and if you do that you can see
> > all sorts of problems. why not look at photos normally? did you grain
> > peep film too?
>
> No. Make an enlargement and you will see how much resolution a photo
> really has. And if you can't make enlargements, what is the point of
> having a high pixel count?
since bayer cameras have as much as 5 times as many pixels (in the same
35mm form factor), you can make substantially bigger enlargements than
with foveon. if you step up to medium format backs, it's over 12 times
as many pixels.
do you regularly make 20" x 30" posters?
nospam
Molon <alfred...@yahoo.com> wrote:
> In article <211120091910312644%nos...@nospam.invalid>, nospam says...
> > In article <MPG.257290a73...@news.supernews.com>, Alfred
> > Molon wrote:
> >
> > > A full colour sensor, an idea some people here hate so much.
> >
> > it's not that people hate it, it's that there's a lot of drawbacks that
> > some people blindly dismiss. a full colour sensor would be better *if*
> > it didn't compromise on high iso, dynamic range, frame rate and colour
> > accuracy. it also triples the storage and processing requirements.
> > foveon loses on all of them. nothing is perfect, and bayer is an
> > excellent tradeoff of what is currently possible.
>
> We are not discussing the implementation, only if it would be preferable
> to have full colour sensor.
of course it would be preferable *if* you could do it without
compromising anything else. unfortunately, there's no free lunch.
the reality is you end up compromising on things which are far more
important than having higher colour resolution. you *will* see higher
noise and colour casts, but you won't see the added colour resolution
and a slow frame rate is not desirable for some types of photography,
such as sports.
nospam
Molon <alfred...@yahoo.com> wrote:
> Hasselblad are not quadrupling the resolution, they are just achieving
> the full nominal resulution of the sensor (as opposed to Bayer sensors,
> where the effective resulution is somewhere between ~ 30-100% of the
> pixel count, depending on the scene captured).
completely wrong.
> In other words with their 39MP sensor, with this shift technology they
> obtain an image with 39MP of information, while with a Bayer sensor you
> are somewhere between 10 and 39MP of information. The numbers are of
> course just guesstimates.
yes, wild guesstimates that are completely bogus.
Wolfgang Weisselberg
> .. and probably something which could be easily added to any sensor-shift
> camera at little cost, perhaps providing /some/ image quality improvement.
why not go the whole nine yards, use a monochrome sensor (no bayer
pattern at all) and a slew of filters? In that way you'd not
only have some broad filters which sort-of center on red, green
and blue --- you could use narrow spectrum filters in addition
and IR-pass and UV-pass filters and cyan, yellow, magenta and
emerald and filters for all the difficult to resolve colours, too.
Works perfectly well in our remote probes, in space, currently
on Mars, too --- even worked on the hot and inhospitable Venus.
It doesn't need a sensor shift mechanism but uses well known and
time-tested methods. It also offers much higher effective ISO
values for monochrome shots when used with a clear or missing
filter, as no Bayer filter reduces the passing light.
But if you insist on a sensor shift, make it shift sub-pixel
distances ... 16 (i.e. 4x4) images could quadruple the resolution.
The software to do so is already available, look e.g. here
http://auricle.dyndns.org/ALE/gallery-auto/
In addition, you can use it for many scientific tasks by just
choosing the correct set of filters.
> Whether there are many lenses that would justify the potential quadrupling
> of the resolution is another question.
You get close to quadrupling as often as you win the national
lottery. It *could* happen, theoretically ...
The average case would be about 30-50% more resolution.
Remember that the green filter is broad, and records also red
and blue light, just weaker.
-Wolfgang
C J Campbell
> Given that SLR makers are already incorporating piezo dust shakers,
> it might not be too far a stretch to include this technology as well.
> It would deal with the Bayer sensor's shortcoming and probably end the
> use of the Foveon sensor.
>
> Hasselblad has announced a Multi-Shot (MS) version of its H3DII-50
> medium format camera. First shown in the H3DII-39 MS in 2008, the
> system captures four shots in a row, moving the sensor by one
> pixel between each shot to record full RGB values at each
> position. THe H3DII-50 MS costs 23,000 euros with less expensive
> trade-in prices and a trade-up route for Hasselblad owners.
> http://www.dpreview.com/news/0911/09111902hasselbladh3d50ms.asp
Yeah, I read that. Interesting idea and one that I remember asking on
this group a couple years ago if it could be done. All you brilliant
souls who answered in the affirmative can pat yourselves on the back,
now. (I seem to recall that was nearly all of you.)
I would love it for landscapes, less so for portraiture. No way are you
going to get four identical shots of the bride walking down the aisle.
And a tourist shot of that porpoise at Sea World jumping through a hoop
-- um, right. So, unfortunately, the technology may never migrate to
where it would do the most good -- cameras with small sensors.
--
Waddling Eagle
World Famous Flight Instructor
Alfred Molon
> since bayer cameras have as much as 5 times as many pixels (in the same
> 35mm form factor), you can make substantially bigger enlargements than
> with foveon. if you step up to medium format backs, it's over 12 times
> as many pixels.
Alfred Molon
> of course it would be preferable *if* you could do it without
> compromising anything else. unfortunately, there's no free lunch.
>
> the reality is you end up compromising on things which are far more
> important than having higher colour resolution. you *will* see higher
> noise and colour casts, but you won't see the added colour resolution
> and a slow frame rate is not desirable for some types of photography,
> such as sports.
There is a free lunch and no compromising with this piezo actuator
system, if the scene is static.
In any case, with a Bayer system you throw away 2/3 of the light with
the colour filters. It's just a matter of finding a way to use all
incoming light, then there will be no performance degradation.
Alfred Molon
>
> > Hasselblad are not quadrupling the resolution, they are just achieving
> > the full nominal resulution of the sensor (as opposed to Bayer sensors,
> > where the effective resulution is somewhere between ~ 30-100% of the
> > pixel count, depending on the scene captured).
>
> completely wrong.
True instead.
Alfred Molon
Taylor says...
> My question about lenses stands - how many are actually good enough the
> reach the full resolution (rather than the lower "Bayer resolution") of
> today's high-end DSLRs, even with ideal taking conditions?
nospam
Molon <alfred...@yahoo.com> wrote:
> > since bayer cameras have as much as 5 times as many pixels (in the same
> > 35mm form factor), you can make substantially bigger enlargements than
> > with foveon. if you step up to medium format backs, it's over 12 times
> > as many pixels.
>
> What are you talking about? 5 times as many pixels compared to what?
all of sigma's current cameras (including the vaporware sd15) have the
same 4.7 megapixel sensor they introduced a few years ago. meanwhile,
nikon, canon and sony have sensors as high as 24 megapixels. dividing,
24 by 4.7 is 5.11 times as many pixels.
phase one has a 60 megapixel medium format back. 60/4.7 = 12.77 times
as many pixels.
nospam
Molon <alfred...@yahoo.com> wrote:
> > of course it would be preferable *if* you could do it without
> > compromising anything else. unfortunately, there's no free lunch.
> >
> > the reality is you end up compromising on things which are far more
> > important than having higher colour resolution. you *will* see higher
> > noise and colour casts, but you won't see the added colour resolution
> > and a slow frame rate is not desirable for some types of photography,
> > such as sports.
>
> There is a free lunch and no compromising with this piezo actuator
> system, if the scene is static.
that's hardly a free lunch. it's only free if you aren't moving, which
rules out just about everything other than still life with the camera
on a tripod.
> In any case, with a Bayer system you throw away 2/3 of the light with
> the colour filters. It's just a matter of finding a way to use all
> incoming light, then there will be no performance degradation.
and that 2/3 is regenerated with excellent accuracy. that's why it's so
clever.
Ray Fischer
>David J Taylor <david-...@blueyonder.not-this-bit.nor-this-part.co.uk.invalid> wrote:
>
>> .. and probably something which could be easily added to any sensor-shift
>> camera at little cost, perhaps providing /some/ image quality improvement.
>
>why not go the whole nine yards, use a monochrome sensor (no bayer
>pattern at all) and a slew of filters?
Too slow.
--
Ray Fischer
rfis...@sonic.net
nospam
Molon <alfred...@yahoo.com> wrote:
> > > Hasselblad are not quadrupling the resolution, they are just achieving
> > > the full nominal resulution of the sensor (as opposed to Bayer sensors,
> > > where the effective resulution is somewhere between ~ 30-100% of the
> > > pixel count, depending on the scene captured).
> >
> > completely wrong.
>
> True instead.
nope. it's completely wrong. bayer resolution is *not* dependent on the
scene in normal situations. in synthetic situations, such as a colour
res chart, it does degrade, but so does the human eye. not much point
in capturing what you can't see.
Ray Fischer
When you insist that he's wrong and then proceed to describe how he is
right it makes you look like an argumentative asshole
--
Ray Fischer
rfis...@sonic.net
Ray Fischer
>In article <MPG.2573e402f...@news.supernews.com>, Alfred
>Molon <alfred...@yahoo.com> wrote:
>
>> > of course it would be preferable *if* you could do it without
>> > compromising anything else. unfortunately, there's no free lunch.
>> >
>> > the reality is you end up compromising on things which are far more
>> > important than having higher colour resolution. you *will* see higher
>> > noise and colour casts, but you won't see the added colour resolution
>> > and a slow frame rate is not desirable for some types of photography,
>> > such as sports.
>>
>> There is a free lunch and no compromising with this piezo actuator
>> system, if the scene is static.
>
>that's hardly a free lunch. it's only free if you aren't moving, which
>rules out just about everything other than still life with the camera
>on a tripod.
One need only look at any catalog to see all of the many photos taken
of still objects.
>> In any case, with a Bayer system you throw away 2/3 of the light with
>> the colour filters. It's just a matter of finding a way to use all
>> incoming light, then there will be no performance degradation.
>
>and that 2/3 is regenerated with excellent accuracy.
But not as good as it could be.
--
Ray Fischer
rfis...@sonic.net
nospam
<rfis...@sonic.net> wrote:
> >nope. it's completely wrong. bayer resolution is *not* dependent on the
> >scene in normal situations. in synthetic situations, such as a colour
> >res chart, it does degrade, but so does the human eye.
>
> When you insist that he's wrong and then proceed to describe how he is
> right it makes you look like an argumentative asshole
read it again. it's *only* an issue with artificial scenes that don't
occur in the real world. it's *not* a problem in normal usage,
especially since the eye can't see the difference.
nothing is perfect and *any* camera is going to have an edge case where
it falls apart.
nospam
<rfis...@sonic.net> wrote:
> >> There is a free lunch and no compromising with this piezo actuator
> >> system, if the scene is static.
> >
> >that's hardly a free lunch. it's only free if you aren't moving, which
> >rules out just about everything other than still life with the camera
> >on a tripod.
>
> One need only look at any catalog to see all of the many photos taken
> of still objects.
there are many more photos of moving objects, namely people, who often
are holding the objects in the catalogues you mention.
> >> In any case, with a Bayer system you throw away 2/3 of the light with
> >> the colour filters. It's just a matter of finding a way to use all
> >> incoming light, then there will be no performance degradation.
> >
> >and that 2/3 is regenerated with excellent accuracy.
>
> But not as good as it could be.
with today's technology it is.
Wolfgang Weisselberg
> In any case, with a Bayer system you throw away 2/3 of the light with
> the colour filters.
How do you calculate 2/3rds? Have you counter-checked that with
absorbtion curves from real filters used for that purpose?
And how does that stop and a bit compare to the approximately
two stops you throw away by using a front-lit sensor? And what
about the light loss in the optics --- every glass-glass and
every glass-air surface costs light!
And finally, when you need every photon, how comes that Foveon
suffers so badly in high ISO settings compared to Bayer pattern
sensors?
-Wolfgang
Alfred Molon
>
> > > since bayer cameras have as much as 5 times as many pixels (in the same
> > > 35mm form factor), you can make substantially bigger enlargements than
> > > with foveon. if you step up to medium format backs, it's over 12 times
> > > as many pixels.
> >
> > What are you talking about? 5 times as many pixels compared to what?
>
> all of sigma's current cameras (including the vaporware sd15) have the
> same 4.7 megapixel sensor they introduced a few years ago. meanwhile,
> nikon, canon and sony have sensors as high as 24 megapixels. dividing,
> 24 by 4.7 is 5.11 times as many pixels.
>
> phase one has a 60 megapixel medium format back. 60/4.7 = 12.77 times
> as many pixels.
Alfred Molon
> > >> In any case, with a Bayer system you throw away 2/3 of the light with
> > >> the colour filters. It's just a matter of finding a way to use all
> > >> incoming light, then there will be no performance degradation.
> > >
> > >and that 2/3 is regenerated with excellent accuracy.
> >
> > But not as good as it could be.
>
> with today's technology it is.
bugbear
> Given that SLR makers are already incorporating piezo dust shakers,
> it might not be too far a stretch to include this technology as well.
> It would deal with the Bayer sensor's shortcoming and probably end the
> use of the Foveon sensor.
>
> Hasselblad has announced a Multi-Shot (MS) version of its H3DII-50
> medium format camera. First shown in the H3DII-39 MS in 2008, the
> system captures four shots in a row, moving the sensor by one
> pixel between each shot to record full RGB values at each
> position. THe H3DII-50 MS costs 23,000 euros with less expensive
> trade-in prices and a trade-up route for Hasselblad owners.
> http://www.dpreview.com/news/0911/09111902hasselbladh3d50ms.asp
>
Sounds like what Sinar and Imacon used to do, back in 2001.
I've considered trying to use CHDK and built in IS
to get similar results.
Photoacuate will do this as a retro-process for ANY camera.
BugBear
Wolfgang Weisselberg
> In article <221120091842434933%nos...@nospam.invalid>, nospam says...
>> > But not as good as it could be.
>> with today's technology it is.
> Interpolation is not "clever".
Sensor shifting is not "clever", either.
Foveon is "clever", but doesn't work as well as "stupid"
methods.
Now, what would that tell you about being too clever for your
own good?
-Wolfgang
Ray Fischer
Trolls still comment on what they know not.
It's not dependent on accuracy. Hand-held shots are used in PhotoAcute's
software method. That provides for all the pixel overlap that is required.
All that matter is that a random amount of pixel overlap is available.
There was one CHDK developer that was going to us the IS shift mechanism to
allow for a 4 frame wide-angle mode. Wherein the IS would be shifted to
extremes for each corner of the frame during a high-speed burst sequence,
allowing for stitching later, surpassing the camera's own wide-angle
optical limits. He sadly left the project. But if approached on the
pixel-shifting for greater resolution possibility, I'm sure he would have
experimented on it.
bugbear
Sorry - I wasn't explicit. I wasn't (as you say "couldn't") trying
to do what Sinar do, but I could automate the capture process
for photoacute.
BugBear
Kennedy McEwen
<nos...@nospam.invalid> writes
>
>foveon is anything but elegant. it doesn't actually sense pure red,
>green and blue like in the ads. there's a lot of overlap in the three
>layers,
What makes you think that the Bayer CFA doesn't have a lot of overlap
too?
It gets worse still: both Canon and Nikon have trumpeted their use of
"lower density CFAs" in recent years to get higher ISO results. A lower
density red filter is one that passes more G & B - same with other
colours.
--
Kennedy
Yes, Socrates himself is particularly missed;
A lovely little thinker, but a bugger when he's pissed.
Python Philosophers (replace 'nospam' with 'kennedym' when replying)
Kennedy McEwen
<nos...@nospam.invalid> writes
>In article <he9p7p$ot9$1...@news.eternal-september.org>, Charles
><charles...@comcast.net> wrote:
>
>> I was not clear enough: the "idea" is elegant.
>
>the idea is interesting, but it isn't really solving a problem that
>needs to be solved. the eye can't see the added colour resolution.
>
That myth is pretty much debunked in the Hassy data sheet itself - there
is more detail visible in the second image. I can see it, you can see
it.
How come we can do that if the "eye can't see the added colour
resolution".
Kennedy McEwen
<rfis...@sonic.net> writes
>bugbear <bugbear@trim_papermule.co.uk_trim> wrote:
>>
>>I've considered trying to use CHDK and built in IS
>>to get similar results.
>
>It won't work. You cannot control IS with sufficient accuracy.
>
fractions of a pixel under motion that would otherwise smear over
several pixels - and it is doing that dynamically. It ought to be
possible to get single pixel precision under static conditions.
It doesn't need to be in-camera IS to do this either, any IS system
including optical IS, has enough precision and range to create the 4
single position shifts.
nospam
<r...@nospam.demon.co.uk> wrote:
> >foveon is anything but elegant. it doesn't actually sense pure red,
> >green and blue like in the ads. there's a lot of overlap in the three
> >layers,
>
> What makes you think that the Bayer CFA doesn't have a lot of overlap
> too?
it's nowhere near as much.
nospam
<r...@nospam.demon.co.uk> wrote:
> >the idea is interesting, but it isn't really solving a problem that
> >needs to be solved. the eye can't see the added colour resolution.
>
> That myth is pretty much debunked in the Hassy data sheet itself - there
> is more detail visible in the second image. I can see it, you can see
> it.
i'll wait for an unbiased test.
> How come we can do that if the "eye can't see the added colour
> resolution".
since foveon can change the laws of physics, i suppose hasselblad can
change physiology too.
Ray Fischer
>In article <Spvr4$KwylD...@kennedym.demon.co.uk>, Kennedy McEwen
><r...@nospam.demon.co.uk> wrote:
>
>> >the idea is interesting, but it isn't really solving a problem that
>> >needs to be solved. the eye can't see the added colour resolution.
>>
>> That myth is pretty much debunked in the Hassy data sheet itself - there
>> is more detail visible in the second image. I can see it, you can see
>> it.
>
>i'll wait for an unbiased test.
"Run away!"
>> How come we can do that if the "eye can't see the added colour
>> resolution".
>
>since foveon can change the laws of physics, i suppose hasselblad can
>change physiology too.
Kook.
--
Ray Fischer
rfis...@sonic.net
Ray Fischer
> Ray Fischer
>>bugbear <bugbear@trim_papermule.co.uk_trim> wrote:
>>>I've considered trying to use CHDK and built in IS
>>>to get similar results.
>>
>>It won't work. You cannot control IS with sufficient accuracy.
>>
>IS has adequate accuracy to hold the image stationary on the sensor for
>fractions of a pixel under motion that would otherwise smear over
>several pixels - and it is doing that dynamically. It ought to be
>possible to get single pixel precision under static conditions.
Good luck with that.
>It doesn't need to be in-camera IS to do this either, any IS system
>including optical IS, has enough precision and range to create the 4
>single position shifts.
Nope.
--
Ray Fischer
rfis...@sonic.net