39 megapixels vs. 4x5

Mike wrote:
. . . . . . . . . . . .

After you invest the $5 billion in the fab. Of course the niche players
will contract out the fab. I wonder who is manufacturing the PhaseOne
silicon?



Kodak. Same chip being used by Imacon (Hasselblad), but with different
hardware and software attached.



Even if you find a fab, creating those masks is not cheap and existing
manufacturing processes are geared towards economies of scale (which you
correctly pointed out will not exist for MF/LF sensors).


Probably never. I don't think that means these devices are frozen in
high prices, since some other technology could be developed that becomes
more economical to produce . . . maybe.



Do you think a $3000 digital capture solution will be available in 5-8
years that can deliver 4x5 quality? I don't think so...but would value
your thoughts.



Scanning back?

Ciao!

Gordon Moat
A G Studio
http://www.allgstudio.com

On Fri, 20 Jan 2006 21:35:58 -0500, rafe b rafebATspeakeasy.net
wrote:

The mass market is already well served with
image sensors of 5x7mm. For a grand, you
get 15x22mm, which easily beats 35mm.

For three grand, you get 24x36mm, which
begins to encroach on 645 film territory

Examples ? I haven't seen any digital cameras that can eaqual a well
exposed TMX film yet.

John

On Sat, 21 Jan 2006 04:47:26 -0600, John
wrote:

On Fri, 20 Jan 2006 21:35:58 -0500, rafe b rafebATspeakeasy.net
wrote:

The mass market is already well served with
image sensors of 5x7mm. For a grand, you
get 15x22mm, which easily beats 35mm.

For three grand, you get 24x36mm, which
begins to encroach on 645 film territory

Examples ? I haven't seen any digital cameras that can eaqual a well
exposed TMX film yet.


Google is your friend, John. There are lots of
Canon 5D captures available for download off
the web. Several at Canon's website, for starters.

I'm not familiar with TMX. Given a choice, I grab
the slowest C41 film I can find and shoot with that.
In MF, I use Reala. In LF, Portra 160.

645 scanned at 4000 dpi yields around 55-60
million pixels, so this seems like an absurd
comparison, right?

But if you upsample to 5D image to match the
scanned-645 dimensions, it holds its own. That's
only 2:1 linear upsampling, 4x in terms of pixel count.

This isn't only my conclusion, of course.
It's one of the reasons high-end digital is
already taking a toll on MF as well.


rafe b
www.terrapinphoto.com

Gordon Moat wrote:


Scanning back?

Scanning backs can be done very inexpensively but they are very limited
in what you can do with them. There are however some really large
linear CCD with a lot of pixels so it would not be hard to put one
together. I would say a scanning back would be pretty much limited to
studio work.

You could also image a hybrid scanning back, one that used an area CCD
but with a very large aspect ratio. Imagine a CCD what is 8000 pixels
by 1000, this is only an 8 MP sensor. But if you used this to scan
instead of a linear CCD you could scan at close to 1000 times faster.
By using 10 exposures you could build up an image that was 8000 x 10000
pixels. To do the same with a linear CCD would take 10000 exposures,
and a lot of time.

It might be possible to use TFT to somehow make a large sensor on a
glass plate. The problem is that the electronics industry was willing
pour billions of dollars into TFT research for displays because they
know the pay off would be huge. It would be hard to get anyone to pour
billions into research on making very large photo-sensor arrays.

Scott

On Sat, 21 Jan 2006 04:47:26 -0600, John
wrote:

On Fri, 20 Jan 2006 21:35:58 -0500, rafe b rafebATspeakeasy.net
wrote:

The mass market is already well served with
image sensors of 5x7mm. For a grand, you
get 15x22mm, which easily beats 35mm.

For three grand, you get 24x36mm, which
begins to encroach on 645 film territory

Examples ? I haven't seen any digital cameras that can eaqual a well
exposed TMX film yet.


I probably missed your point in my first response.

If you're dedicated to shooting BW, then a Bayer-
based digicam is probably not for you.

As dedicated BW shooter, you're well outside of
the mainstream and not really significant to the
bean counters at Kodak, Fuji, Nikon, or Canon.

That's not a value judgment, just a statement
of fact.

It's a bit ironic, of course, since there's no
technological reason the same sensors
couldn't work for you -- you'd just need the
sensor without the Bayer filter, and of course
it would need very different (actually, much
simpler) post-processing.

There are some high-end backs by Leaf
that work that way, but again, mega $$$.


rafe b

"rafe b" rafebATspeakeasy.net wrote

[...]
If you're dedicated to shooting BW, then a Bayer-
based digicam is probably not for you.

[...] you'd just need the
sensor without the Bayer filter, and of course
it would need very different (actually, much
simpler) post-processing.

At the moment I have to wonder if that's true. What is the naked spectral
sensitivity of a sensor without a Bayer filter? Is it not dead straight,
and extending much farther into UV and IR than film?

If it is, then I think B&W photographers would be very unhappy with a
nonfiltered sensor. B&W is, truly, color without hue; that is, B&W
photographers always have to deal with the modulation of colors by the film.

On Sat, 21 Jan 2006 11:44:25 -0600, "AH2"
wrote:

"rafe b" rafebATspeakeasy.net wrote

[...]
If you're dedicated to shooting BW, then a Bayer-
based digicam is probably not for you.

[...] you'd just need the
sensor without the Bayer filter, and of course
it would need very different (actually, much
simpler) post-processing.

At the moment I have to wonder if that's true. What is the naked spectral
sensitivity of a sensor without a Bayer filter? Is it not dead straight,
and extending much farther into UV and IR than film?

Not sure about UV response, but yes, most CCDs
respond readily to IR, and in fact require IR filtration
for "normal" applications.

IIRC, glass attenuates UV (unless it's quartz) so
UV isn't going to get through the taking lens in
any case.


If it is, then I think B&W photographers would be very unhappy with a
nonfiltered sensor. B&W is, truly, color without hue; that is, B&W
photographers always have to deal with the modulation of colors by the film.


The point is that the sensels for a BW imaging
chip would not want the standard Bayer filtration,
but a something that's the uniform over all sensels.

Such sensors and filters certainly exist, but have
no role in the consumer/prosumer market.

This cuts both ways. It's not just the manufacturers
unwilling to serve the BW market. It's that the BW
market (such as it is) is unreceptive to digital capture
in the first place. You can't serve high-tech to Luddites.

There's a comparable issue with digital printing.
Those desiring to print in monochrome with inkjets
are pretty much ignored by the mass market, and
left to fend for themselves. The market niche there
has been picked up by expensive RIPs and share-
ware products like Roy Harrington's QuadTone RIP.


rafe b
www.terrapinphoto.com

AH2 wrote:
"rafe b" rafebATspeakeasy.net wrote

[...]
If you're dedicated to shooting BW, then a Bayer-
based digicam is probably not for you.

[...] you'd just need the
sensor without the Bayer filter, and of course
it would need very different (actually, much
simpler) post-processing.

At the moment I have to wonder if that's true. What is the naked spectral
sensitivity of a sensor without a Bayer filter? Is it not dead straight,
and extending much farther into UV and IR than film?

The sensors extend way out in the IR but not very far into the UV at
all. To get a BW
CCD camera to work well you need to use a filter somewhere, this can
either be in front of the sensor or in front of the lens. These filters
are fairly simple and don't stop much of the visible light. To get the
effect you want the use of color filter on the lens would work about
the same as with a panchromatic film.

The only thing stopping digital cameras from going to BW is there is
little market for BW.

Scott

"rafe b" rafebATspeakeasy.net wrote in message
...

The point is that the sensels for a BW imaging
chip would not want the standard Bayer filtration,
but a something that's the uniform over all sensels.

Sorry I am having such a hard time wrapping my little mind around the idea
that it is desirable to produce a flat luminal response to all colors; I am
so accustomed to conventional film's relatively similar curves, and
modulating outcomes using filters.

This cuts both ways. It's not just the manufacturers
unwilling to serve the BW market. It's that the BW
market (such as it is) is unreceptive to digital capture
in the first place. You can't serve high-tech to Luddites.

Are you saying that B&W photographers are Luddites? I find them more
inclinded to some complex metrics - in particular, translating colors as
they wish to. B&W is color photography.

There's a comparable issue with digital printing.
Those desiring to print in monochrome with inkjets
are pretty much ignored by the mass market, and
left to fend for themselves.

Well, so-called B&W printing is far better on most consumer ink-jet printers
than it was only four years ago, but Yes, to get very good grey-tones one
has to resort so some spendy media and the public isn't interested.

rafe b wrote:
On Sat, 21 Jan 2006 04:47:26 -0600, John
wrote:


On Fri, 20 Jan 2006 21:35:58 -0500, rafe b rafebATspeakeasy.net
wrote:


The mass market is already well served with
image sensors of 5x7mm. For a grand, you
get 15x22mm, which easily beats 35mm.

For three grand, you get 24x36mm, which
begins to encroach on 645 film territory

Examples ? I haven't seen any digital cameras that can eaqual a well
exposed TMX film yet.



Google is your friend, John. There are lots of
Canon 5D captures available for download off
the web. Several at Canon's website, for starters.

I'm not familiar with TMX. Given a choice, I grab
the slowest C41 film I can find and shoot with that.
In MF, I use Reala. In LF, Portra 160.



TMX commonly refers to Kodak TMax 100, which is a B/W film. Kodak
Technical Document F-32 has information on all TMax films, including
developing options. Probably one of the best choices for getting the
maximum resolution from large format lenses.

645 scanned at 4000 dpi yields around 55-60
million pixels, so this seems like an absurd
comparison, right?


Actually, if we just use the 39 MP Kodak CCD, and look past PhaseOne to
Imacon, we can find that the Imacon has a micro step mode that moves the
entire CCD a small amount. While some would claim the resolution
increases, what actually happens is that edge definition becomes more
accurate. The maximum file would then be 8000 by 10000 pixels, which is
more like an 80 MP digital back. The slightly smaller than 645 area
remains the same.

To match the file size with 4x5 in a scan would only need a 2000 dpi
scan. With a very good high end scanner, even a perfect 2000 dpi scan
only means about 40 lp/mm maximum. Some large format lenses are capable
of closer to 60 lp/mm, though to get that would mean a 3000 dpi or
greater scan, meaning a much larger file size.

In reality, the PhaseOne and Imacon backs should be expected to be close
to 40 to 45 lp/mm, based upon similar technologies (cell site sizes on
other CCDs). A Canon 5D is actually higher resolution in real tests, a
Canon 1Ds Mark II is slightly better, and a Nikon D2X has even greater
real resolution capability. Kodak TMax 100 is actually beyond the
(resolution) capability of most large format lenses at common working
apertures, so the lens is the greatest limit in large format film usage.

When the film area increases, a greater scan is needed to capture all
the information. This implies overscanning beyond needs to capture as
much as possible from the film. Then when it comes to printing, as in
commercial printing or wide format inkjet systems, the only advantage of
such a large scan, and the only advantage of that much resolution over
that area is the ability (potential) to make a really large print.

Those considering optical prints might be better thinking in terms of
contact printing. If 4x5 film is contact printed, the resolution is
extremely high. In fact, it is beyond the ability of the average person
to even see the limits of that resolution. If we take 5 to 6 lp/mm as
the common limits of eyesight, and a best capability of 60 lp/mm as a
common capability of large format, then we would need more than 10x
optical enlargement before we could see the limits of resolution from
4x5 film . . . that would be a print near 40" by 50". Obviously, some
really good optics on the camera and enlarger would be necessary, and
these would be very expensive items needing a great deal of skills to
operate. So far, the colour prints of Edward Burtynsky are the closest I
have seen to that.

But if you upsample to 5D image to match the
scanned-645 dimensions, it holds its own. That's
only 2:1 linear upsampling, 4x in terms of pixel count.

This isn't only my conclusion, of course.
It's one of the reasons high-end digital is
already taking a toll on MF as well.



I think it is more a matter of the skills and time needed to get the
most out of medium format or large format. Many of us have seen some
really nice prints from medium format and large format. When the need
for scanning to create that print comes up, then the scanner
requirements are very high. When the need to get the best optical
enlargements from those films arises, then the very top quality of
enlarger, lenses, and techniques is needed. To put this in plain
language, some photographers are finding it substantially easier to use
direct digital and a computer, and many of those people are quite happy
with the results.

Ciao!

Gordon Moat
A G Studio
http://www.allgstudio.com