Pixel Pitch Universally Defined

Professionals who offer to help Sigma and Foveon engineer non-photographers figure out why the camera is failing is not a bad thing. Some professionals are open to a Foveon sensor. Most are not based on existing problems. There are few things stopping Sigma and Foveon from working.

1. Most Sigma users don't understand other cameras. This is important. You seem to have convinced yourselves that Sigma is doing well. It is clear from these noise discussions that most of you have no clue what other cameras are doing. It sounds like a put down but it isn't intendend that way. You have no choice. Sigma uses an SA mount. That is one reason why their mount is a bad decision. Most Sigma users don't use the better cameras.

Users who don't understand other systems give Sigma wrong feedback. Sigma tries to listen. They continue to release dSLRs farther and farther off the mark. The SD14 is the result.

2. No C/N mount. A few good lenses but no great lenses. The very small real pixel pitch of the SD10 (5.3 microns) and SD14 (4.5 microns) equals high ISO 100 resolution. That means great lenses are required. The small pitch also means relatively poor noise characteristics. Fast lenses that remain sharp are crtical but not available. The SD14 cannot reach its potential in a Sigma mount.

Rick. f/2.8 and faster is critical. You think it is useless because Sigma dSLRs are useless in low light. Profitable photographers live a f/2.8 and wider. f/2.8 is fairly stopped down. Frankly. Ls and some non-Ls are very sharp at f/2.8 and need to get faster.

Professionals cannot afford to buy outside of their mount. Even if some might they would not add SA.

3. Very high noise. SD10 high ISO images are unusable from a professional persepective. Ergo the camera is unusable. The ISO 1600 house samples posted by Laurence Matson are were taken in broad daylight. IQ is a mess. They show extreme shadow noise. Unruly hue shifts. Poor sharpness. Dead color. Even ISO 200 is noisy. Professionals buying in volume need a camera that can take a great picture in very low light. Sigma cameras cannot. Specifications and samples suggest the smaller pixel pitch of the SD14 will result in the same or worse noise.

What? You say? No one says the SD10 is extremely noisy. Most tell us it is "ok". Well. We are being nice.

4. Lighting is more important than lenses. Consumers and amatures don't understand why. So it only makes it to reason #4. Sigma SA is incompatible with professional lighting systems. Until Sigma is compatible with C or N lighting it is a dead system.

5. Sigma. I think people were open to them succeeding. They missed their opportunity. The SD14 is less competitive than was the SD9. With that Sigma's opportunity to be a smart camera Brand is gone.

So what can be done? The answer is nothing until Sigma seeks competent advice. Even if they admit they are failing and change course it will take a long time. Sigma's course of action is this.

1. Forget about sensor IQ. No one who matters cares. A few amateurs think film is the most important part of a body. IQ has to be good but it still doesn't matter because the market is flooded with awsome quality. Better than you have at the moment. Forget your users misguided notion that film can build your Brand. Look where that got Kodak and Fuji. Build a body that would sell if everyone still used film.

For Foveon. Since you are stuck in the film business. Not a place I'd want to be.

1. Stop!

2. Seek a competent professional photographer's advice before proceding. Find ten wedding photographers who charge in the $10,000/event range and are booked for a minimum of three years. Above that price range is not a group open to you. Ask them what they need and want from film and sensors. Not all professionals shoot weddings but they are a good test for a film company like you. The direction you are headed is all wrong.

3. Here is the main thing you need to learn about digital IQ. Efficiency and effectiveness are not the same. IQ efficiency is meaningless to photographers. IQ effectiveness is everything. You need to produce a sensor that is effective even if at the direct expense of efficiency. That is why Bayer sensors are killing you.

The next two are not sensor design related.

4. Your greatest strength is faster workflow per quality. Not basic IQ. All digital professionals understand that processing time is the most expensive part of their business. It is by far the largest challenge of any digital system. You hold the key to great success but you don't see it.

5. Get out from under Sigma glass.
 
With all this noise comparison now, where did you factor in
in-camera noise-reduction processing. I have yet to see those
roots, squares, and sines.
Noise reduction is not low noise. Foveon has less silicon than the 5D. Not more silicon as you claim with 14MP * 7.8^2 pitch. That means high resolution per film area and high noise. Noise-reduction doesn't make Sigma's wrong characterization of pixel pitch right.
 
Do you agree the SD10 has more noise than the 5D at ISO 1600?
Foveon claims more silicon surface on a 4/3 die than the 5D has on
a full frame die. 14.1MP * 7.8 microns^2. Than is 858mm^2 of
silicon surface for the SD14. 838 mm^2 of silicon surface for the
5D.

So why is SD10 noise higher? Its 10MP pixel pitch is 5.3
microns/pixel after silicon surface sharing gets accounted for. 5D
8.2 pitch/SD10 5.3 pitch. Or. 5D 16.4 pitch/SD10 9.1 pitch. Not
choose from each. What our eyes see makes sense.
Thanks mgates for this interesting read. It is always interesting
to see opposing views argued so well, before conceding that the
opposition is actually correct :)

Anyway, how can you get away with comparing technologies that are
years apart on timescale? We all know the rate at which the digital
world revolves these days so it hardly makes for a good comparison.
It doesn't matter since the SD14 remains behind. But it does provide insight as to why. That is the first step to finding a path forward.
Then there is the factor of "in-camera" noise reduction techniques.
Canon's, and now Nikon's, appear to have it pretty well under
control, but what else would you expect from companies with their
R&D funds? Also, I may be wrong but is it not the case that Sigma
do the vast majority of their noise reduction via their Photo Pro
software? Just something I think I read here lately, but as I said
it may be wrong :)
They have better noise performance. That isn't fundatmentally why they are winning. They are winning because their physics is better. The 5D has a 16.4 micron pitch by Foveon's full color pixel method. The SD14 can't compete with that at "7.8 pitch" as they want to claim.

Foveon is a good concept but a 4/3 footprint is way too small for clean 14MPs. Clean high ISO is expected today. Foveon needs to be at least 2 stops cleaner to be purchased instead. The SD10 is 3 stops noisier. The SD14 will probably about be the same given the smaller pixels and bad samples. Meanswhile other cameras add stops.
Lastly, of course, we eagerly await the real thing from the SD14.
Previews will be out soon and then we can all judge for ourselves
whether or not it will pay to "miss a generation" if results are
disappointing, or perhaps jump ship entirely.
Too bad they don't want to share IQ.

But it doesn't matter. 4.5 micron real pitch is too small. All the great minds at Foveon cannot defy gravity.
 
Like Roland said, "Pitch is the physical distance from the center
of one pixel to the center of another."
Ok. Live by that. There are two more pixels right on top of the
first. They share the same surface area.
Believe it or not - even a cook understood you are wrong, and I
assure you, you are wrong.
Show me a Sigma dSLR with more silicon than the 5D and you win. That is what Sigma says when they claim 14MP * 7.8^2 pitch.
 
Professionals who offer to help Sigma and Foveon engineer non-photographers figure out why the camera is failing is not a bad thing. Some professionals are open to a Foveon sensor. Most are not based on existing problems. There are few things stopping Sigma and Foveon from working.

1. Most Sigma users don't understand other cameras. This is important. You seem to have convinced yourselves that Sigma is doing well. It is clear from these noise discussions that most of you have no clue what other cameras are doing. It sounds like a put down but it isn't intendend that way. You have no choice. Sigma uses an SA mount. That is one reason why their mount is a bad decision. Most Sigma users don't use the better cameras.

Users who don't understand other systems give Sigma wrong feedback. Sigma tries to listen. They continue to release dSLRs farther and farther off the mark. The SD14 is the result.

2. No C/N mount. A few good lenses but no great lenses. The very small real pixel pitch of the SD10 (5.3 microns) and SD14 (4.5 microns) equals high ISO 100 resolution. That means great lenses are required. The small pitch also means relatively poor noise characteristics. Fast lenses that remain sharp are crtical but not available. The SD14 cannot reach its potential in a Sigma mount.

Rick. f/2.8 and faster is critical. You think it is useless because Sigma dSLRs are useless in low light. Profitable photographers live a f/2.8 and wider. f/2.8 is fairly stopped down. Frankly. Ls and some non-Ls are very sharp at f/2.8 and need to get faster.

Professionals cannot afford to buy outside of their mount. Even if some might they would not add SA.

3. Very high noise. SD10 high ISO images are unusable from a professional persepective. Ergo the camera is unusable. The ISO 1600 house samples posted by Laurence Matson are were taken in broad daylight. IQ is a mess. They show extreme shadow noise. Unruly hue shifts. Poor sharpness. Dead color. Even ISO 200 is noisy. Professionals buying in volume need a camera that can take a great picture in very low light. Sigma cameras cannot. Specifications and samples suggest the smaller pixel pitch of the SD14 will result in the same or worse noise.

What? You say? No one says the SD10 is extremely noisy. Most tell us it is "ok". Well. We are being nice.

4. Lighting is more important than lenses. Consumers and amatures don't understand why. So it only makes it to reason #4. Sigma SA is incompatible with professional lighting systems. Until Sigma is compatible with C or N lighting it is a dead system.

5. Sigma. I think people were open to them succeeding. They missed their opportunity. The SD14 is less competitive than was the SD9. With that Sigma's opportunity to be a smart camera Brand is gone.

So what can be done? The answer is nothing until Sigma seeks competent advice. Even if they admit they are failing and change course it will take a long time. Sigma's course of action is this.

1. Forget about sensor IQ. No one who matters cares. A few amateurs think film is the most important part of a body. IQ has to be good but it still doesn't matter because the market is flooded with awsome quality. Better than you have at the moment. Forget your users misguided notion that film can build your Brand. Look where that got Kodak and Fuji. Build a body that would sell if everyone still used film.

For Foveon. Since you are stuck in the film business. Not a place I'd want to be.

1. Stop!

2. Seek a competent professional photographer's advice before proceding. Find ten wedding photographers who charge in the $10,000/event range and are booked for a minimum of three years. Above that price range is not a group open to you. Ask them what they need and want from film and sensors. Not all professionals shoot weddings but they are a good test for a film company like you. The direction you are headed is all wrong.

3. Here is the main thing you need to learn about digital IQ. Efficiency and effectiveness are not the same. IQ efficiency is meaningless to photographers. IQ effectiveness is everything. You need to produce a sensor that is effective even if at the direct expense of efficiency. That is why Bayer sensors are killing you.

The next two are not sensor design related.

4. Your greatest strength is faster workflow per quality. Not basic IQ. All digital professionals understand that processing time is the most expensive part of their business. It is by far the largest challenge of any digital system. You hold the key to great success but you don't see it.

5. Get out from under Sigma glass.
 
Joe Wisniewski calls him
"Suzie" but I don't use that term, since he's a he....years ago he
posted his own photo...
I call him George Preddy - because thats the identity he used while
pestering news:rec.photo.digital. At that time he was extremely pro
Foveon and wrote hundreds (thosands?) of posts praising the only
real digital camera (SD9) on the planet.
That makes sense.
 
mgates,

Pitch is the physical distance from one pixel to another,
This can be misunderstood. It is the physical stride or pitch of
the pixels, i.e. the distance from center to center of the pixels.
That is zero for a Foveon sensor. Do you have a denfintion that works?

How do you calulate 3CCD camera pitch? Your oversimplified definition does not work.
 
In your noble effort, you are not coming to the defence of a fellow
technocrat but rather an idiot. And that is putting it in polite
terms.
Laurence - maybe he is right. Maybe it is just a huge bluff. Maybe
Foveon uses just some kind of CFA solution and tells us that they
are stacking. Maybe - to hide the fact they use strange filters
instead of R, G and B. And the fantastic X3 sharpness only is a
more clever CFA design with very clever software.
Your post was meant to be dorky but there is no sharpness advantage. Size RGGB down to a pixel and Fovoen is no better at ISO 100 and fuzzier above ISO 100. Plus the other high ISO problems.
 
Both use layers. One reads out layers. Same circuitry in either
case.
You either don't understand the Foveon sensor design, or you are
being deliberately obtuse. I suspect the latter.
You can't explain why?
However, to give you the benefit of the doubt, I suggest you look
at the Foveon design again. The blue, green and red sensors are
stacked vertically on top of each other. Photosite area for each
colour is the same, and on the SD14 that area is approx 7.8
microns.
7.8 microns of surface shared by three sensors. So not 7.8 microns true pitch.

How do you ignore the problem of a similar 3CCD design that really has a 7.8 pitch? With the silicon to back up their pitch^2*MP claim. You can't igore real silicon. It matters.
This is the physical design of the sensor. There is no
point in trying to convince people that the "pixel pitch" is 1/3 of
that because of the number of photosites - that's just rubbish.
It isn't 1/3. It is SQRT(7.8^2/3).
What might be significant is the fact that there is light
absorption in each layer which may affect the signal reaching the
layer below. On the other hand light reaching a Bayer photosite has
been filtered to remove most of the spectrum, which is a similar
effect.
No. Small pitch is bad due to a lack of silicon surface area per pixel which means smaller storage capacity.
 
There's nothing in the definition of pitch that has to do with color.

A Foveon sensor with 9-micron pixel pitch will have three times the pixel density (pixels per unit area) of a Bayer sensor with 9-micron pixel pitch, due to the stacking of three layers. Pitch is still pitch: the distance between centers.

Pitch is a simple concept. Don't overload it with a new idea of yours about "control logic" whatever that means.

j
http://www.xenics.com/Products/Glossary.php
...the distance between the center of two adjacent pixels in an array.
Ok. Red to red. Green to green. Blue to blue. Again Foveon is
worse off. The math always works the same. You never get more
silicon by adding control logic.
 
microlenses which increase the effective collection areas above
that of the active pixel area alone. I'm not sure how it plays
regarding Foveon and Bayer but is does impact the issue.
They both have them. So it doesn't impact Sigmas incorrect pitch calculations.
 
I think your request is a 5D/D2X/1DsMkII. Why spend time on beating a dead horse? Sigma SD-XX is not those $$ gaining commercial product. Believe it or not, Foveon will be one of the new success species sensor and ONLY Sigma (, Hanvison and Polariod) willing to step in and make it success. But, it just don't fit your taste.

I hate to compare which is better and which is worse. To me, at ISO 100 and good lighting, my SD9 and Sigma lenses provide me very good pictures. They are not expensive equipment (and some are not perfect), but the result is so good that on other equipment can counter.

SD10 is even better and i believe SD14 will continue to improve as well.

I don't care what is the the "standard" entry-level resolution of DSLR is, I just care if my gear can do the right job.

You may think SD14 is heading a "wrong way", but the improvement is exactly what i need.

--
'May Foveon be with you'

Thomas the C.Wolf 8^)
Gallery:
http://www.pbase.com/sigmasd9/c_wolf
http://www.pbase.com/nethomas
 
I'm missing your point, obviously.

The SD10 has 9.1 micron pitch and 10.2 M pixels. It does this by using three layers of pixels. What's the mystery here?

j
Why redefine pixel pitch? It already has a perfectly standard
definition. The SD10 is 9.1 microns and the SD14 is 7.8 microns.
Ok 3.4MP and 4.7MP. Not 10MP and 14MP. Can't have it both ways.
Both ways the 5D has 12.7MP and RGGB center-to-center pitch of
16.4. Foveon is relatively worse off.
 
Both use layers. One reads out layers. Same circuitry in either
case.
You either don't understand the Foveon sensor design, or you are
being deliberately obtuse. I suspect the latter.
You are right that is it's one or the other. I'll stop there.

...
What might be significant is the fact that there is light
absorption in each layer which may affect the signal reaching the
layer below.
Of course it affects the layers below! That's how the filtering is done. The upper layer filters out mostly blue, leaving mostly just red and green for the lower layers; etc.

j
 
There's nothing in the definition of pitch that has to do with color.
Oh well. Then we count the sensors piled on top of one another. Not skip to the next sensor of the same color like Sigma does.
A Foveon sensor with 9-micron pixel pitch will have three times the
pixel density (pixels per unit area) of a Bayer sensor with
9-micron pixel pitch, due to the stacking of three layers. Pitch
is still pitch: the distance between centers.
But not for a Foveon stack. Then you switch to whatever defintion suits your fancy.
 
Both use layers. One reads out layers. Same circuitry in either
case.
You either don't understand the Foveon sensor design, or you are
being deliberately obtuse. I suspect the latter.
You are right that is it's one or the other. I'll stop there.
You have no choice. You know that three sensors sharing the same silicon surface is more than one.

Pitch is SQRT(Area/MPs). No other defintiion works with all single and many sensor designs.
 
I'm missing your point, obviously.

The SD10 has 9.1 micron pitch and 10.2 M pixels. It does this by
using three layers of pixels. What's the mystery here?
Layers are not more silicon. (Pixel pitch)^2*MPs must always add up to silicon.

The "mystery" is how anyone can cite MORE layers and MORE densely packed pixels as a reason for lower noise. Sigma recently tried. That is wrong.

The SD10 is all the evidence we need.
 
Foveon is claiming virtual photosites.
You may want to read this:
http://www.dpreview.com/news/0203/02030602foveonx3notation.asp

The X3 chip in the SD14 is a 14mp sensor that has 14mp photosites
that generates a 4.7mp output image pixels. It can be called either
14mp or 4.7mp.
Exactly. As it explains, it's 14.1 M effective pixels, 4.7 M
recorded pixels.
Call it 14MP with 4.5 pitch. Or call it 4.7MP with 7.8 pitch. Both add up. Sigma wants to call it 14MP with 7.8 pitch. That adds up to more silicon than the 5D. That is false advertising.
 

Keyboard shortcuts

Back
Top