D30 Replacement

Al Goldis said:

This comes up on a fairly regular basis, but there are a couple
problems with a 35mm square format that always seem to get glossed
over.

Many lenses produce a circular image. However, many other lenses
for the 35mm system we use are designed for a rectangular image.
Ignoring the flower petal hoods which sometimes can easily be
removed (or sometimes not), many lenses still have internal
rectangular baffles.

I suppose you could still use those rectangular lenses with a
smaller image circle to get a square image, but why force those of
us who want a 3:2 image into a square when all you have to do is
crop?

The 35mm systems have always been 3:2 and I don't see that
changing. If you want square, you can crop or buy a Hasselblad
with a fantastic digital back.

BTW, the other reason I don't see square format coming to 35mm-type
systems... virtually all of the consumer cameras are rectangular
too (although more 4:3 than 3:2). If people wanted square images,
we would see it in consumer cameras.

Click to expand...

Al,

Ah, yes the internal baffles. I figured there was probably something I was missing, and that probably is the Achilles' heal. And certainly, a square image can be cropped from a rectangular one as easy as vice versa. The thought I had was taking advantage of the as much of the sweet spots of the lens as possible, but as you pointed out, this may not be possible anyway.

Thanks for responding (and enlightening).

-Tom

Tom Colling said:

Ah, yes the internal baffles...

Click to expand...

There you have it... no square formats in 35mm are possible (at least, not at maximum size).

But the idea of cropping to different aspect ratios in-camera from the native 24x36mm format remains. You could still use the viewfinder mask to crop your 3:2 image to 4:3, 5:4, or even 1:1 by the same method as previously described. If it's engineered correctly, it should be a useful feature for people who print routinely in a non-3:2 format, since they would then have the opportunity for more accurate framing.

JCDoss

Member said:

Since you are an engineer, I'll ask you this question...
Right now, the rule of thumb is that the larger the pixel, the
better the S/N ratio and the better the capabilities of the sensor
at high sensitivities (ISO).

Member said:

How do you think technology will solve the problem of incorporating

Click to expand...

higher S/N ratios and high ISO capability in a large sensor with
pixels, say, 5um square (like ones present in consumer digicams)?
First, do you think this will happen? If so, is on-chip processing
the way to go?

FYI, if a the D30's chip were expanded to full frame, it's
resolution would be over 8MP... however, if we reduce the pixel
size from 10um to 5um, that will quadruple the resolution to 32MP,
which many believe exceeds 35mm film resolution (estimated at 24MP).

Member said:

There will be no "stitching" of the two or more sensors. Although
fesable, this requires a lot of processing power which takes
precious time and yields unsatisfactory results.

Click to expand...

Really? The articles I've read recently seem to suggest that it's
more economical than attempting to produce larger chips without
stitching.

JCDoss

Click to expand...

JC,

I believe that CMOS as a production technology for use in photographic applications is only in its infancy. In time, pixel size will shrink, signal to noise ratio, speed and light sensitivity will all improve. This will take a few years to level out, but it will be a continous improvement driven by competition. CMOS has a huge advantage in power consumption and cost of production. CMOS has been around for close to two decades in other applications and it is only now becoming a viable solution as a high resolution photosensitive sensor/transducer.

Stitching is a means of mending two pieces after the fact whether you are talking about fabric, footballs or photos. It is not a seamless process and one would hope to record a scene with a camera in a seamless way.

I arrived at 8.3 mega pixels by multiplying the height times the width of a 35 mm frame (23.3 X 35 = 815.5 square milimeters or 815,500,000 square micrometers) to detemine the area. One pixel that is 9.9 um square is 98.01 square um in area. Therefore the pxel count is determined by dividing 815,500,000 by 98.01 to equal 8,320,579.5 pixels.

On chip processing will absolutely be the way to go as things evolve. This improves transit time and thus speeds frame rates. Additionally, on chip processing simplifies the circuitry interface from the data acquition source (the sensor) to the processor. The caveat/sticking point is that on chip processing is technically complex. I believe that Canon is on the right path in discovering the secrets of CMOS as a core technology and that they will be the leader in this field for the next several years, probably selling their sensors to other OEMs.

I neglected to mention in my first post that it is my belief that the auto-focus shortcomings will largely be solved with the introduction of the next generation prosumer camera.

Digital camera technology is now where cell phones and ink jet printers were in the early to mid 1990's. The point and shoot market will continue to evolve at at a blinding speed for a couple of more years. Digital SLR's (both Professional and Prosumer models) will evolve rapidly for probably the next four years. During this time, prices will continue to shrink and then level out. By that time, I believe that things will be about as stable as they were for film SLR's in the '70s and '80s.

I also believe that in the next five years the film camera market will shrink to a reminant of its present size. Inexpensive single use cameras, infrared cameras, and some other special-use models will continue to be produced. I also believe that film will be widely available for about twenty more years. However, the choice of speeds and other choices will shrink to about one third of what is presently available. --Steve Bryant

Even with improvements in CMOS technology, I'd still want my pixels at around 10um. I'd love to see 800/1600 ISO as clean as 200/400 on the D-30. With 3200/6400 as good as 800/1600 on the D-30.

For low-light, non-flash work, this would be incredible. I'd take that over 4 times as many pixels any day.

Granted, others might prefer more pixels at today's S/N ratio.

JCDoss said:

The D30 performs excellently at ISO
100-400, pretty good at 800, and so-so at 1600, however the pixel
size of 10um is somewhat limiting for enlarging and cropping.

How do you think technology will solve the problem of incorporating
higher S/N ratios and high ISO capability in a large sensor with
pixels, say, 5um square (like ones present in consumer digicams)?

Click to expand...

--The Unofficial Photographer of The Wilkinsons http://thewilkinsons.crosswinds.net

Member said:

Al,

Ah, yes the internal baffles. I figured there was probably
something I was missing, and that probably is the Achilles' heal.
And certainly, a square image can be cropped from a rectangular one
as easy as vice versa. The thought I had was taking advantage of
the as much of the sweet spots of the lens as possible, but as you
pointed out, this may not be possible anyway.

Thanks for responding (and enlightening).

-Tom

Click to expand...

Did I actually say "Achilles' heal"?

Ewe no, sum thymes eye Ken bee sew stupid

DavidP said:

Even with improvements in CMOS technology, I'd still want my pixels
at around 10um. I'd love to see 800/1600 ISO as clean as 200/400
on the D-30. With 3200/6400 as good as 800/1600 on the D-30.

For low-light, non-flash work, this would be incredible. I'd take
that over 4 times as many pixels any day.

Granted, others might prefer more pixels at today's S/N ratio.

Click to expand...

The beauty should be that downsizing a 32MP image should provide the same signal/noise ratio as an 8MP image taken with larger pixels. That's by today's standards, though.

This stuff is really interesting...
JCDoss

Steve,

Steve Bryant said:

I believe that CMOS as a production technology for use in
photographic applications is only in its infancy. In time, pixel
size will shrink, signal to noise ratio, speed and light
sensitivity will all improve...

Click to expand...

I hope this is the case. Although the resolution of the D30 is good, there's room for improvement (even if it were full frame).

Steve Bryant said:

Stitching is a means of mending two pieces after the fact whether
you are talking about fabric, footballs or photos. It is not a
seamless process and one would hope to record a scene with a camera
in a seamless way.

Click to expand...

Actually, stitching is pretty close to a pre-fab process (at least "intra-fab") that occurs before the sensors are cut out of the wafer. Some companies have developed a way to overlap the mask on a wafer and create a much larger and seamless chip.

Steve Bryant said:

I neglected to mention in my first post that it is my belief that
the auto-focus shortcomings will largely be solved with the
introduction of the next generation prosumer camera.

Click to expand...

Care to elaborate?

JCDoss

I'm not so sure it works that way in practice.

My guess it that the S/N would be improved over using 32 Mpixels, but wouldn't be as good as using 8 Mpixels of larger size.

And I'm not sure that just a "downsizing" in photoshop would give that S/N improvement. It might have to be done in-camera, or with some other external program other than Photoshop.

Hopefully some of the people here more familiar with this can give some insight.

If what you're saying is right, then Yippeeee!

JCDoss said:

The beauty should be that downsizing a 32MP image should provide
the same signal/noise ratio as an 8MP image taken with larger
pixels. That's by today's standards, though.

Click to expand...

--The Unofficial Photographer of The Wilkinsons http://thewilkinsons.crosswinds.net

I believe the sides of the square would be 30.6 mm and the area would be 936 rather than the rectangle's 864.....

James

JCDoss said:

JCDoss said:

The most important factor is the diameter of the circle which must
equal (or be slightly greater than) the diagonal of the frame. If
we assume this is true for a 24x36 frame, the diagonal of which is
43mm, then the diameter of the circle is 43mm, and the diagonal of
the square must not exceed 43mm. The resulting square would be 6mm
taller and 6mm 'skinnier.' Area within both the rectangle and the
square would be equal.

Click to expand...

The area within the square will NOT equal the area within the
rectangle.
A(square) = 30x30 = 900sqmm.
A(rectangle) = 24x36 = 864sqmm.

Therefore, a square format will use more of the lens.

JCDoss

Click to expand...

--James

James Knowles said:

James

JCDoss said:

JCDoss said:

The most important factor is the diameter of the circle which must
equal (or be slightly greater than) the diagonal of the frame. If
we assume this is true for a 24x36 frame, the diagonal of which is
43mm, then the diameter of the circle is 43mm, and the diagonal of
the square must not exceed 43mm. The resulting square would be 6mm
taller and 6mm 'skinnier.' Area within both the rectangle and the
square would be equal.

Click to expand...

The area within the square will NOT equal the area within the
rectangle.
A(square) = 30x30 = 900sqmm.
A(rectangle) = 24x36 = 864sqmm.

Therefore, a square format will use more of the lens.

JCDoss

Click to expand...

--
James

Click to expand...

--James

Member said:

Member said:

I neglected to mention in my first post that it is my belief that
the auto-focus shortcomings will largely be solved with the
introduction of the next generation prosumer camera.

Click to expand...

Care to elaborate?

JCDoss

Click to expand...

JC,

The reason that I believe that Canon can fix the auto focus issue is that they have to have had a lot of knowledge in this arena. They have probably analized the root cause(s) and evaluated several approaches to a fix. However, it is likely that a firmware approach will not suffice, or it would have already been rectified. Probably a hardware modification is needed and that will probably be coming in the next model. It could be further improved on a subsequent model. I don't think that there is an intentional strategy to slowly fix the problem so as to sell more camera bodies of each version. Competion and fundamental integrity should rule that out. However, if it might take an additional six months and a number of additional prototype iterations to fully fix the issue and if they presently have a more viable solution than with the D30 autofocus and do not have other glaring product issues, they will introduce the next prosumer model. --Steve Bryant

JCDoss said:

Steve Bryant said:

...Eventually the 1D derivitive will evolve with a smaller
pixel measurement, perhaps approaching a 15 mega pixel array which
will remain the size of existing 35 mm film frames.

Click to expand...

Since you are an engineer, I'll ask you this question...
Right now, the rule of thumb is that the larger the pixel, the
better the S/N ratio and the better the capabilities of the sensor
at high sensitivities (ISO). The D30 performs excellently at ISO
100-400, pretty good at 800, and so-so at 1600, however the pixel
size of 10um is somewhat limiting for enlarging and cropping.

How do you think technology will solve the problem of incorporating
higher S/N ratios and high ISO capability in a large sensor with
pixels, say, 5um square (like ones present in consumer digicams)?
First, do you think this will happen? If so, is on-chip processing
the way to go?

Click to expand...

JCDoss, Steve,

Although the current ccd sizes being less than film are a problem as far as multiplier factors(greater than 1) are concerned for standard 35mm lenses, they will not pose the same problem once sensors are larger than 35mm film. The reason is simple physical optics, its possible to distribute or disperse the incoming light from the external lens by use of a fixed internal lens that adjusts for the fractional multiplier ( it would be less than 1 for sensors larger than film when processing light passing through standard 35mm lenses) Using a technique such as this would ensure improvements of low signal to noise ratio's, high resolution and larger pixel area for sensors larger than 35mm film dimensions. I think the digital cameras of 10 years from now will display resolution, noise and color fidelity far beyond the capabilities of any traditional film formats.

Member said:

FYI, if a the D30's chip were expanded to full frame, it's
resolution would be over 8MP... however, if we reduce the pixel
size from 10um to 5um, that will quadruple the resolution to 32MP,
which many believe exceeds 35mm film resolution (estimated at 24MP).

Click to expand...

I've actually read of estimates of much lower resolutions in digital being able to eclipse film (like 6mp) In fact many have done subjective tests that consistently rate the 3mp D30 prints as having greater apparent resolution than a 35mm film camera print up to 11 x 14 size. In my opinion a 24mp digital image would provide resolution comparable or better than medium format film sizes leaving 35mm long behind. Also remember that by increasing the number of pixels while maintaining the same ccd dimension we reduce signal to noise per photosite, this makes image processing more difficult and less likely to provide as noise free an image as would be possible if the photosite dimensions are not reduced.

Dsl

Member said:

Member said:

There will be no "stitching" of the two or more sensors. Although
fesable, this requires a lot of processing power which takes
precious time and yields unsatisfactory results.

Click to expand...

Really? The articles I've read recently seem to suggest that it's
more economical than attempting to produce larger chips without
stitching.

Click to expand...

Member said:

JCDoss

Click to expand...

--DSL

Tom Colling said:

Ewe no, sum thymes eye Ken bee sew stupid

Click to expand...

Didn't you mean "stew pit"? (...or is it stoop-Ed"?)

Monica says, "Close, ...but no cigar!"

Larry

Steve / JC,

Entertaining, enlightening exchange.

Thanks!, ...you have my permission to continue. ;-)

Larry

Well, while that isn't too far from the truth JC.....It had better be at least a "significant" improvement or I may well take a pass. The one that is in R&D for late 2002 announcement is the more likely unit I'll go for. We will just have to see. It will take that long for me to financially prepare (and get my wife's mind off of the $$$ I have already spent. LOL). It's a darn good thing I dont have rug rats or I'd never pull this off.

JCDoss said:

The beauty should be that downsizing a 32MP image should provide
the same signal/noise ratio as an 8MP image taken with larger
pixels. That's by today's standards, though.

Click to expand...

though not an expert, the type of digital noise is totally random so i very much doubt it can be "averaged" out. afterall the noise replaces signal. probably better to have fewer, larger pixels.--.ICQ Canon EOS activelist 136431230---BreezeBrowser RAW conversion times http://www.davidbirkin.com

Lee Rothman said:

It's a darn good thing I
dont have rug rats or I'd never pull this off.

Click to expand...

Reminds me of the comedienne who said she and her husband were considering a pet, vs. a child. They couldn't decide whether they wanted to ruin their apartment, ...or their LIFE!

I did the (2) kid-thing, ...now I have a cat. She's RIGHT!

Larry (kidding, ..sort-of, they're both great adults now!)

Steve Bryant said:

I am in the market for my first digital camera and plan to await a
successor to the D30. Whether it is the D60 or another model
designation, I believe that it will be worth waiting. I have been
shooting 35 mm for over thirty years and have been scanning and
printing for the last six years.

I've been very interested in the speculation on the D60 and I too
will be following any announcements from the PMA convention in
Orlando in February. I have worked as an Avionics Engineer
(aircraft electronics and instrumentation systems) for the past
twenty-five years. I want to offer the following
observations/guesses about the eventual successor to the D30:

The basic architecture of the D30 will be preserved (form
factor/ergonomics, electronics including the CMOS sensor
technology, etc.)

The CMOS sensor will eventually be enlarged to the full 35 mm size
of 23.3 X 35 mm using the existing 9.9 micro meter pixel size
yielding approximately 8.3 Mega Pixels. Once the details of this
CMOS technology are refined in the D30 prosumer derivitive, it will
be applied to a EOS-1D derivitive. The 9.9 micro meter/8.3 mega
pixel array will probably stay with the prosumer camera for a
number of years and features will be refined in the coming
versions. Eventually the 1D derivitive will evolve with a smaller
pixel measurement, perhaps approaching a 15 mega pixel array which
will remain the size of existing 35 mm film frames.

Click to expand...

Consider that Leica started the current (double frame) technology over 70 years ago by doubling the then standard single movie film frame... there are many precedents to show that shorter lenses are equal if not sharper than longer lenses. The small CCDS today are comparable to the minox or 16mm cameras and are producing consistently excellent prints at 11x14, 16x20 and larger. Smaller lenses are lighter faster and perhaps easier to make...ie the 28 to 200 on the Minolta D7.

The only problem with the newer cameras is that they are tiny... I find it difficult to get my fingers into the little buttons and switches... and of course anyone over 40 has a hard time reading any of the printing on the camera or menu screens.
Stuart

Stuart said:

Consider that Leica started the current (double frame) technology
over 70 years ago by doubling the then standard single movie film
frame... there are many precedents to show that shorter lenses are
equal if not sharper than longer lenses. The small CCDS today are
comparable to the minox or 16mm cameras and are producing
consistently excellent prints at 11x14, 16x20 and larger. Smaller
lenses are lighter faster and perhaps easier to make...ie the 28 to
200 on the Minolta D7.

Click to expand...

Interesting arguement. However almost everyone agrees that medium format is still superior to 35mm even 70 years later. APS and all other attempts at a smaller than 25mm film frame have been miserable failures
I tink it is clear that it cameras at least, size matters.

David Birkin said:

though not an expert, the type of digital noise is totally random
so i very much doubt it can be "averaged" out. afterall the noise
replaces signal. probably better to have fewer, larger pixels.

Click to expand...

You very well could be right, David.

The process I'm talking about is called "binning" (for those who don't know), and involves the combination of data from more than one pixel to a single data point. I'm not sure whether pixel binning would equal the performance offered by larger pixels, though.

If one had to choose between two cameras: Camera A capable of only low-res, noise-free images and Camera B capable of switching between extremely high-res, somewhat noisy images and low-res, almost noise-free images, I suspect most would go for Camera B.

I think large-scale sensors used for astronomy that cost more than my house use pixel binning to increase S/N ratio where high resolution is also important. Again, not being one of those astronomers, I could be wrong.

JCDoss