120 megapixel resolution from Canon

"Attila Jozsef" wrote in message
...


I met Peter Gowland once at his Santa Monica studio. He showed me his
then pride and joy, the "Gowlandflex" 4" x 5" Twin-Lens Reflex he was
just putting on the market. He said he invented it so he could take
hand-held large format pics of babes on the beach (for those unfamiliar
with his work is consists of about 95% nudes). Now that was a heavy
hand-held camera! He said he was working on an 8x10 version -- I
wouldn't want to wrestle one of those!

I don't know, wrestling nudes might be fun.

Oh! you are talking about a hand held 8x10.


--
Peter

On Fri, 27 Aug 2010 10:49:54 -0400, John A. wrote:

On Fri, 27 Aug 2010 07:46:33 -0400, "Peter"
wrote:

"David J. Littleboy" wrote in message
...

"Peter" wrote:
"Me" wrote:

Also, that type of system can only be effectively used when the camera
is mounted on a tripod.

Exactly.

No - they are very commonly used hand-held with flash systems, just as
tripods are commonly used by landscape shooters using prosumer dslrs in
the field.

I guess most have hands that are fare more steady than mine..

The big* medium format SLR cameras work fine handheld, as long as you can
keep the shutter speed at 1/125 and shorter. I love the brick cameras, and
have shot both 500-series blad and Mamiya 645Pro extensively. But the
context was the pixel shift function, which requires four separate
exposures, so that really does require a tripod.

*: Well, I don't know about the Mamiya RB and RZ 6x7 SLRs. Those are
really big. I've seen people shooting them hand held, but haven't tried it
myself.



Of course many can hand hold MF cameras. I used to hand hold my Bronica.
However, I thought the context of my comment was hand held multiple
exposures,
the context:

"On 08/25/2010 02:49 PM, John A. wrote:
On Wed, 25 Aug 2010 14:34:11 +0200, wrote:


I remember VCRs being marketed with Nx oversampling. And processors
have transitioned to being marketed based on the # of cores as much as
MHz. I figure they could market such a sensor as "30Mp with 4x
oversampling for superior color& detail resolution."

Come to think of it, Doesn't Hassy have a camera that takes four shots
with the sensor shifted a pixel each time to get full color data for
every pixel? This could be similar but with one shot.
The four shots idea is smart, also from the Signal/Noise perspective,
why can't my Canon do that... Still, with the longer effective exposure,
camera motion could spoil everything, unless it is compensated. Vice
versa, if you can combine accurate camera motion tracking with multiple
shots, you could achieve the same."

I simply do not see how that could be accomplished without a tripod.

I agree. You'd have to be dead to hold a camera steady enough for four
shots to register that well.

What I'm wondering is if cameras with in-body shake reduction/image
stabilization could be made to do it with a firmware update/hack.

You're all ****ing morons. You do realize that, don't you? Sub-pixel
resolution is done all the time with hand-held multiple-shots.

On 08/26/2010 02:42 AM, David J. Littleboy wrote:

Rant:
Fine with me!
I don't see the problem with Bayer. I look at 4000 ppi (Nikon 9000) scans of
6x7 film, and it's a tossup whether the same shot with the 5D2 is better or
worse. In real life, Bayer digital is _FOUR TIMES BETTER_ than film* It's
Keep in mind that film is not like an idealized Foveon sensor, but has
quite different resolutions for the different color layers, just like a
Bayer array -- at least for the films I used to use. And as I think you
know, peak lines/mm is one thing, SNR is another.
frigging amazing. Real life detail doesn't have the kinds of red on blue
patterns that Foveon fans worry about. Every digital image of high color
contrast signs I've ever seen has beatifully delineated transitions. The
idea that there's a problem with Bayer is insane.
Given a certain level of technology a Bayer array does indeed quite well
against, say, a Foveon sensor. Perhaps a rule of thumb is that if for
the same generation of sensors a Bayer array provides 2x the pixels of
the Foveon, thanks to clever interpolation, it wins. It's just that each
pixel in a Bayer array won't provide an independent RGB value, which is
where pixel shifting comes in.
End rant.
ok!

*: 6x7 cropped to a 2:3 ratio is 70mm x 47mm, just under four times the area
of the 5D2 sensor.
Not even that much more, especially if you consider the maybe 10x higher
sensitivity of the 5D sensor.

-- Hans

Hanz wrote:

Given a certain level of technology a Bayer array does indeed quite well
against, say, a Foveon sensor. Perhaps a rule of thumb is that if for
the same generation of sensors a Bayer array provides 2x the pixels of
the Foveon, thanks to clever interpolation, it wins.

I'd argue SQRT(2) (distance between 2 green pixels) is
enough more for real-life comparisons.

It's just that each
pixel in a Bayer array won't provide an independent RGB value,

Just as the human eye doesn't provide independent RGB values
at every light sensitive cell.

-Wolfgang

Hi Wolfgang,

Given a certain level of technology a Bayer array does indeed quite well
against, say, a Foveon sensor. Perhaps a rule of thumb is that if for
the same generation of sensors a Bayer array provides 2x the pixels of
the Foveon, thanks to clever interpolation, it wins.

I'd argue SQRT(2) (distance between 2 green pixels) is
enough more for real-life comparisons.
With 2x more pixels in a Bayer array w.r.t. a Foveon stacked array, the
closest (i.e. diagonal) distance between pixels in the green channel is
already equal that of the pixel distance in the Foveon array.
When a Bayer array has the same amount of pixels as the Foveon array,
the resolution from the green Bayer channel is SQRT(2) worse, but then
the technology levels differs and the comparison is not completely fair.

It's just that each
pixel in a Bayer array won't provide an independent RGB value,

Just as the human eye doesn't provide independent RGB values
at every light sensitive cell.
Yup.

-- Hans

Hanz wrote:

Given a certain level of technology a Bayer array does indeed quite well
against, say, a Foveon sensor. Perhaps a rule of thumb is that if for
the same generation of sensors a Bayer array provides 2x the pixels of
the Foveon, thanks to clever interpolation, it wins.

I'd argue SQRT(2) (distance between 2 green pixels) is
enough more for real-life comparisons.

With 2x more pixels in a Bayer array w.r.t. a Foveon stacked array, the
closest (i.e. diagonal) distance between pixels in the green channel is
already equal that of the pixel distance in the Foveon array.

With 1.414...x more pixels, not with 2x more pixels.

When a Bayer array has the same amount of pixels as the Foveon array,
the resolution from the green Bayer channel is SQRT(2) worse, but then
the technology levels differs and the comparison is not completely fair.

It's close enough.

-Wolfgang