Reading mode: Light Dark

Pixel Pitch Universally Defined

Go to last post
I meant to post this at the root level since it ends our discussion.

--

In case you didn't notice. The funny joke above decided this debate by accident. Here is the final word.

Photons hit Foveon silicon. Some wavelengths pentrate farther. Sensors from different channels are placed at different depths to exploit this phenomenon. Therefore sensors cannot underlap but they can share a centroid. This proves silicon surface is conserved.

For Laurence that means that each 7.8 micron of square of silicon gets divided into three horizontally distinct parts. 14M horizontally distinct parts.

Absolute proof that the SD14 has a 4.5 pixel pitch.
 
mgates said:
In case you didn't notice. The funny joke above decided this
debate by accident. Here is the final word.

Photons hit Foveon silicon. Some wavelengths pentrate farther.
Sensors from different channels are placed at different depths to
exploit this phenomenon. Therefore sensors cannot underlap but
they can share a centroid.
Click to expand...
what does 'share a centroid' mean? centroid??
mgates said:
This proves silicon surface is conserved.
Click to expand...
I followed your statements to this point; with "centroid" and henceforth it doesn't make sense IMHO
mgates said:
For Laurence that means that each 7.8 micron of square of silicon
gets divided into three horizontally distinct parts. 14M
horizontally distinct parts.

Absolute proof that the SD14 has a 4.5 pixel pitch.
Click to expand...
 
I shoot good pictures for me and not for you or someone else.
Or should I post it on the Canon forum?
No I don´t think my pictures are superior to any other brand.

I believe the shooter makes the difference, a bit retarded of course but well I am a ****** :D

Despite 2 weeks ago I haven´t posted any picture for a year. Even my pbase gallery just holds mostly old cr@p.

Still I do not get it how the pixel pitch influences the quality of a picture, can I make better compostitions? Or is the message getting better by higher or lower pixel pitch?

This forum is seldom about equipement it is just about rabulism discussions about tech specs
--
http://www.pbase.com/aroid/
http://www.panodrom.de (QTVR site)
 
SandyF said:
mgates said:
In case you didn't notice. The funny joke above decided this
debate by accident. Here is the final word.

Photons hit Foveon silicon. Some wavelengths pentrate farther.
Sensors from different channels are placed at different depths to
exploit this phenomenon. Therefore sensors cannot underlap but
they can share a centroid.
Click to expand...
what does 'share a centroid' mean? centroid??
Click to expand...
A centroid is a geometrically weighted center point.

Two or more shapes can divide a surface but all share the same centroid. Concentric rings are one example. The shapes need not be symetrical or regular.
SandyF said:
SandyF said:
This proves silicon surface is conserved.
Click to expand...
I followed your statements to this point; with "centroid" and
henceforth it doesn't make sense IMHO
Click to expand...
Paraphrasing. Sigma said each pixel in an RGB stack has a full 7.8 micron pitch. Not true. Each only gets some portion of the horizontal silicon surface area within that 7.8 micron square. The centerpoints may still align.

In simple terms. SD14 pixels are smaller than a 7.8 micron square.
 
mgates said:
SandyF said:
mgates said:
In case you didn't notice. The funny joke above decided this
debate by accident. Here is the final word.

Photons hit Foveon silicon. Some wavelengths pentrate farther.
Sensors from different channels are placed at different depths to
exploit this phenomenon. Therefore sensors cannot underlap but
they can share a centroid.
Click to expand...
what does 'share a centroid' mean? centroid??
Click to expand...
A centroid is a geometrically weighted center point.

Two or more shapes can divide a surface but all share the same
centroid. Concentric rings are one example. The shapes need not
be symetrical or regular.
Click to expand...
okay, I see what you mean by centroid, but your further points still do not make sense to me, ie "centroids" don't "prove" anything. Or are you saying the distance from center of photosites vary from layer to layer, eg the difference is x when you compare center to center on one plane and is y when compared one plane to the lower plane?

I agree with Thomas Mottl btw that this thread's argument makes very little sense in practical terms -- a time filler 'til we have available production SD14s
mgates said:
mgates said:
mgates said:
This proves silicon surface is conserved.
Click to expand...
I followed your statements to this point; with "centroid" and
henceforth it doesn't make sense IMHO
Click to expand...
Paraphrasing. Sigma said each pixel in an RGB stack has a full 7.8
micron pitch. Not true.
Click to expand...
Lost you here.
Each only gets some portion of the
mgates said:
horizontal silicon surface area within that 7.8 micron square. The
centerpoints may still align.
Click to expand...
Lost you here.
mgates said:
In simple terms. SD14 pixels are smaller than a 7.8 micron square.
Click to expand...
Lost you here. Are you thinking in cubic or 3D terms. Think of silicon with depth.
 
SandyF said:
mgates said:
SandyF said:
mgates said:
In case you didn't notice. The funny joke above decided this
debate by accident. Here is the final word.

Photons hit Foveon silicon. Some wavelengths pentrate farther.
Sensors from different channels are placed at different depths to
exploit this phenomenon. Therefore sensors cannot underlap but
they can share a centroid.
Click to expand...
what does 'share a centroid' mean? centroid??
Click to expand...
A centroid is a geometrically weighted center point.

Two or more shapes can divide a surface but all share the same
centroid. Concentric rings are one example. The shapes need not
be symetrical or regular.
Click to expand...
Click to expand...
SandyF said:
okay, I see what you mean by centroid, but your further points
still do not make sense to me, ie "centroids" don't "prove"
anything. Or are you saying the distance from center of photosites
vary from layer to layer, eg the difference is x when you compare
center to center on one plane and is y when compared one plane to
the lower plane?
I agree with Thomas Mottl btw that this thread's argument makes
very little sense in practical terms -- a time filler 'til we have
available production SD14s
Click to expand...
The concept is critical to understanding the high noise level of Foveon sensors. I will boil it down.

Pixel pitch is important. It is important because it bounds the maximum silicon available to each pixel. That impacts noise and range. Point and shoot cameras are noisy because they have small pixels with little silicon area.

When we say the 5D has an 8.2 micron pitch that means each sensor has 67 microns^2 of silicon to do its thing. Sigma says they have very big pixels so they have low noise. Aside from their obvious noise the statement remains false. The reason Sigma sensors have high noise is the opposite of what Sigma says is true. Sigma cameras have fairly small pixels. A relatively small amount of silicon for each sensor to do its thing.

The reason is that each square houses three Foveon sensors so total silicon must be divied up into three parts. The parts can be horizontally exclusive and still share the same centroid so all three sensors can still be located in the middle of the square. Foveon's diagram shows that.

So while we don't know the proportions of the three fractions. We do know that the silicon avialable to each sensor is smaller than 7.8 microns square. And we know that the R+G+B total adds up to no more than a 7.8 micron square.

So the SD10 has 10M pixels each with 5.3 microns^2 of dedicated silicon. That is a lot smaller than Sigmas misleading number of 9.1 microns^2. Sigmas number makes no sense. It gives the SD10 much larger pixels than the 5D. The 5D is not noiser than the SD10. The opposite is true. Now we know why. The 5D has a true 8.2 pitch. The SD10 has a true 5.3 pitch. The SD14 will have a true 4.5 pitch.

Pixel size isn't everything. But size matters. It is a physical barrier to overcome. Point and shoot cameras prove that. The SD10 proves that.
 
BarrytheB said:
BarrytheB said:
of some sort through my work? :-)
Ole
Click to expand...
My wife, who is an educator, swears she can hear a whispered
'*****' from fifty feet at the back of a noisy classroom:)
Click to expand...
Barry, I bet it's true, LOL: 'who said that! Nobody leaves the room before he steps foreward!

The ear is a remarkably flexible 'tool'. In my younger days I worked as an industrial psychologist of a sort. We found out that guys who worked in very noisy environments had developed an emphasis on some frequencys in their voices that made them more destinguishable through the noise! And we found out that people who worked as toolmakers at lathes had a greater risk of partial deafness in these environments, because they had to listen for tiny sounds from the lathe during the metalprocessing in order to do the job. The other guys, who didn't do precisionwork, could to some degree turn down the sensitivity of the ears and were a litle better protected against damages and loss or hearing...

Ole
--
http://www.pbase.com/thofte
 
Barry,

I don't know where you've been, but even in the Panhandle Oklahoma is light years away. Anyway, the Republic of Texas doesn't answer to anyone.
BarrytheB said:
BarrytheB said:
'The last resolved iteration of Stusie was from an Oklahoman
institution of higher education, which for many would be as dubious
an academic distinction as any.'
Click to expand...
Yer gittin purty dern close to Texas there padnuh
--
Barry Byrd
http://www.pbase.com/barryb
http://www.pbase.com/sigmadslr
Click to expand...
--
Laurence

My idea of good company is the fellowship of clever, well-informed people, who have a great deal of conversation and liberality of ideas.

Jane Austen

http://www.pbase.com/lmatson/root
http://www.pbase.com/sigmadslr/root
http://www.pbase.com/cameras/sigma/sd10
http://www.pbase.com/cameras/sigma/sd9
http://www.beachbriss.com
 
Thomas Mottl said:
I shoot good pictures for me and not for you or someone else.
Or should I post it on the Canon forum?
No I don´t think my pictures are superior to any other brand.
I believe the shooter makes the difference, a bit retarded of
course but well I am a ****** :D
Despite 2 weeks ago I haven´t posted any picture for a year. Even
my pbase gallery just holds mostly old cr@p.
Click to expand...
I didn't mean the pictures you post, I meant the forum that you participate in. If it were just about the picture, and not the equipment, you'd post in a greater variety of forums rather than just this one.
Thomas Mottl said:
Still I do not get it how the pixel pitch influences the quality of
a picture, can I make better compostitions? Or is the message
getting better by higher or lower pixel pitch?
Click to expand...
Well, I'm not defending the OP's definition of pixel pitch, especially as his definition does not have the same meaning for Bayer as it does for Foveon. A camera is a light capturing device. The more pixels it has to capture light, the more detail it can produce. The bigger these pixels are, the more accurate the color and luminosity information that will be registered for that pixel.

Hence, the quality of an image is a combination of both the number of pixels and the light gathering power of each pixel. What the OP wanted to say was that Foveon pixels, or whatever they're called, gather less light than the pixels on a Bayer sensor, and thus they will have worse noise performance. Well, this is true, but not for the reasons he cited. His notion of "pixel pitch" is not valid since a Foveon pixel captures most of the light that lands on it, whereas a Bayer pixel captures only one specific color. Thus, one cannot directly compare the pixel pitch from Bayer and Foveon as a measure of image performance. Furthermore, you have to consider the efficiency with which each pixel captures light, again, most likely different for Bayer and Foveon, so again, not comparable.

Anyway, the point is that the technical does, of course, directly relate to image quality, and a more in-depth understanding of it will help you decide which sensor will have the optimum quality for the type of photos you take. I sincerely doubt that even that most hardcore Foveon fan would deny that the 5D kills the Sigma DSLRs in high ISO performance. Likewise, you'd find many 5D owners who would probably much prefer a FF Foveon sensor to the current 5D sensor. There is no single solution that is right for every circumstance. Yet.
Thomas Mottl said:
This forum is seldom about equipement it is just about rabulism
discussions about tech specs
Click to expand...
I'm not here enough to comment on that. : )

--
--joe

http://www.josephjamesphotography.com
http://www.pbase.com/joemama/

Please feel free to criticize, make suggestions, and edit my photos. If you wish to use any of my photos for any purpose other than editing in these forums, please ask.
 
mgates said:
Sigma says they have
very big pixels so they have low noise. Aside from their obvious
noise the statement remains false. The reason Sigma sensors have
high noise is the opposite of what Sigma says is true. Sigma
cameras have fairly small pixels. A relatively small amount of
silicon for each sensor to do its thing.

The reason is that each square houses three Foveon sensors so total
silicon must be divied up into three parts. The parts can be
horizontally exclusive and still share the same centroid so all
three sensors can still be located in the middle of the square.
Foveon's diagram shows that.
Click to expand...
But the three photosites are not horizontally exclusive, they occupy the same area (as yoiu look at the sensor face on) but are stacked vertically (as you look at the sensor in cross-section). Light penetrates one layer to get to the next. If your argument was true, light would not penetrate one layer to get to the next because the photosites would be "horizontally exclusive". They are not "horizontally exclusive" and that is the point of the design.

You clearly don't understand enough about the Foveon design, photosensors and light absorption/transmission even at a basic layperson level. Why don't you try looking at this simple explanation here: http://www.dpreview.com/learn/?/Glossary/Camera_System/sensors_01.htm it might help you out.
mgates said:
So while we don't know the proportions of the three fractions. We
do know that the silicon avialable to each sensor is smaller than
7.8 microns square. And we know that the R+G+B total adds up to no
more than a 7.8 micron square.
Click to expand...
erm. duh. Each element is stacked vertically one on top of the other. How can that mean they have less area?
mgates said:
So the SD10 has 10M pixels each with 5.3 microns^2 of dedicated
silicon. That is a lot smaller than Sigmas misleading number of
9.1 microns^2. Sigmas number makes no sense. It gives the SD10
much larger pixels than the 5D. The 5D is not noiser than the
SD10. The opposite is true. Now we know why. The 5D has a true
8.2 pitch. The SD10 has a true 5.3 pitch. The SD14 will have a
true 4.5 pitch.

Pixel size isn't everything. But size matters. It is a physical
barrier to overcome. Point and shoot cameras prove that. The SD10
proves that.
Click to expand...
 
mgates said:
I meant to post this at the root level since it ends our discussion.
Click to expand...
... SuzieQ, sg9 or whatever. I know (we know) that you will take any argument to the extreme and never change your mind. We know that you will answer almost any post.

This time it is pitch it seeams that have cought your attention. We all know that you are wrong there - that there are three layers do not decrease the pitch - as this is a physical measure and not some kind of mean value of individual pitches where the 0 pitches also have to be counted.

Your psuedo physics stuff about photons and silicon area is not totally without merit. It is a part of the explanation for Foveon noise. But ... it is not the main explanation. There are at least two others I can think of that are much more relevant.

But all this is just besides the point. As it is pitch you try to redefine. And there you are just plain wrong. And actually, I think that your technique to start an argument is to very clever choose something you know that others will disagree on. So - in your mind it is probabaly an advantage that it is wrong.

So - we all know that you are wrong - and you will go on and repeat your provocation until no one replies. You are good at keeping it going - so fine lets keep it going then. Its a kind of game after all.

--
Roland
http://klotjohan.mine.nu/~roland/
 
to educate you.

Lets say I have triple glazed windows. Three layers of glass, parallel to each other. Light coming in through the window goes through one layer, then the next and then the next. Each of the three layers has the same area. Each one absorbs some light (but because it is clear glass we are talking about here its not very much).

What you are saying is that the area of each pane of glass in the triple glazing is actually only one third of the area of the window. And what you are saying is clearly nonsense.

Each pane has the same area, but they stack on top of each other. Light that hits the bottom layer has passed through (not passed by) the upper two layers first, and has been altered to some degree by those layers.

Whatever you feel about the strengths and weaknesses of the Foveon sensor in pratice, the argument you are putting forward is just plain rubbish. It doesn't matter how many times you try to recite it, it will remain factually entirely incorrect.
 
mgates said:
The photons do not avoid various filters. All the photons hit
silicon. Longer wavelengths pentrate farther. Therefore sensors
from different channels cannot underly each other. They can share
a centroid. Therefore total silicon must be conserved two
dimensionally.
Click to expand...
Actually - you are wrong. The depth a photon travels in silicon is non deterministic. You can divide the silicon slab in a number of thin layers - and at each layer there is a certain probabaility for capture. The probabability is just higher for shorter wavelengths. Therefore will a longer wavelength photon have a larger expected depth before getting captured. My description - although a joke - is rather accurate.
mgates said:
My definition wins.
Click to expand...
Your defintion is unrealted to photon capture. Your definition is just nonsens you use to start an argument in this forum. It is just your method to have some fun.

--
Roland
http://klotjohan.mine.nu/~roland/
 
Tristan Cope said:
to educate you.
Click to expand...
... that he is wrong. And he is laughing at us all for being so naive to discuss it. Its a game - a kind of joke. He just likes getting a debate starting. I even bet he knew before starting the thread that it is wrong. I bet he have read the debates here what is a pixel - and then twisted that debate to something even more absurd. Knowing that some just have to respond to that kind of stuff. I first fell for it - but now when I know that it is Suzie, Preddy, sg9 I know exactly what it is about. He likes to make fun of us.

--
Roland
http://klotjohan.mine.nu/~roland/
 
mgates said:
spalbird said:
mgates said:
Chunsum J Choi said:
Like Roland said, "Pitch is the physical distance from the center
of one pixel to the center of another."
Click to expand...
Ok. Live by that. There are two more pixels right on top of the
first. They share the same surface area.
Click to expand...
Believe it or not - even a cook understood you are wrong, and I
assure you, you are wrong.
Click to expand...
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.
Click to expand...
The SD14 sensor dimensions are 20.7mm * 13.8mm
The pixel pitch for 4.6 millions effective pixels are calculated as following:
sqrt(20.7mm*13.8mm/4.6e6)=7.88 um

7.8 um is what Sigma and most of us but you believe is true for the SD14.

For every pixel there are three stacked photosites (photodiodes), one for a R, one for a G and one for a B equivalent. This results into 3*4.6 M = 13.8 millions of photo sensors.

Now to the chip area:
The SD14 has an area of 20.7mm*13.8mm = 0.286mm^2 for 4.6 mio effective pixels

The 5D has an area of 24mm*36mm = 0.864mm^2 for 12.7 mio effective pixels/13.3 mio photo detectors

As the SD14 pixels are stacked in three layers we can triple the area occupied by 4.6 mio effective pixels for 13.8 mio photo sensors:

Now the conclusion:
The SD14 has an area of 3*0.7mm*13.8mm = 0.857mm^2 for 13.8 mio of photo sensors
The 5D has an area of 24mm*36mm = 0.864mm^2 for 13.3 mio photo sensors

This means both cameras have about 0.86mm^2 area for about the same number of photo sensors (13/14M). Judging by numbers this is about the same league.

Now, after all this math which I think could be applied for a rough estimation I think I won :)

regards

wolfgang

References:
SD14: http://www.dpreview.com/reviews/specs/Sigma/
5D: http://www.dpreview.com/reviews/specs/Canon/canon_eos5d.asp
 
You did it.

And the collectors items are even better winners with bigger pixels;-)))))))))))))))))))))))))))))))))))))))))))))))))))))))



Aaro
--
!!! Sorry, I don't use blurr filter in my camera, film or digital.



http://www.lumisoft.fi/gallery
 
You must log in or register to reply here.

Keyboard shortcuts

f
Forum
m
My threads
Mobile site
About
Editorial content
Cameras & Lenses
Community
All content, design, and layout are Copyright © 1998–2025 Digital Photography Review All Rights Reserved.
Reproduction in whole or part in any form or medium without specific written permission is prohibited.
When you use DPReview links to buy products, the site may earn a commission.
©GPS Media - Guides, Products, Services.
Back
Top