Advantages of Foveon sensor Locked

Started Jun 23, 2013 | Discussions
This thread is locked.
Roland Karlsson
Forum ProPosts: 20,671Gear list
Advantages of Foveon sensor
Jun 23, 2013

In the thread named "Will Thom's prediction come true" an interesting sub-thread started - and then the entire thread did hit 150 - pang - thread closed. A rather irritation rule to say the least.

The sub-thread was about what advantages Foveon sensors have.

I said: the main advantage is that all colors are sampled in the same place and did give the reference to http://www.dpreview.com/reviews/nikon-coolpix-a/14 , where you can look at the coin above Mickey Mouse. IN particular, look at the Fuji sample, and please look at ISO 200 for the Sigma sample.

Bobn2 said: its the only advantage.

Lin and Laurence said: there are more advantages, and Laurence also listed:

  1. Edge roll-off
  2. Acuity
  3. no moiré
  4. color accuracy (an arguable point for many but absolutely achievable)
  5. greater "presence" or a 3D effect (nebulous but mentioned in so many publications over 10 years)

Personally I would categorise all Laurence's advantages, except the one about color accuracy (4), as a result of the fact that all colors are sampled at the same position. Laurence's list is a list seen from the user perspective, which of course also is a valid way of seeing it.

NOTE that no moire (3) should be no colored moiré. Moiré overall is also less than with Bayer without AA filters, but its absolutely not "no moiré".

So - whats the advantages?

NOTE - we can all make a list of disadvantages if we want to, but can we make a list of advantages?

And - color accuracy? Is that an advantage of Foveon, as Laurence claims?

I can add three more potential advantages, which (just as Laurence's color accuracy advantage) can be discussed.

  1. The color filters in Bayer sensors, do they age?
  2. It is theoretically possible to add a fourth IR layer for Foveon sensors
  3. potentially a sensor like Foveon could detect all photons, Bayer do filter away 2/3.
 Roland Karlsson's gear list:Roland Karlsson's gear list
Canon PowerShot G10 Pentax K-5 Pentax smc DA 15mm F4 ED AL Limited Pentax smc FA 50mm F1.4 Pentax smc DA 21mm F3.2 AL Limited +6 more
SigmaTog
Contributing MemberPosts: 641
Re: Advantages of Foveon sensor
In reply to Roland Karlsson, Jun 23, 2013

An advantage with Foveon is no CFA.

The spin-off is that, with a Foveon full spectrum camera,  UV & IR photography is possible with no false colour introduced by the CFA.

Another advantage is no CFA interpolation or color reconstruction, no CFA mush.

ΣigmaTog

bobn2
Forum ProPosts: 29,360
Re: Advantages of Foveon sensor
In reply to Roland Karlsson, Jun 23, 2013

Hi Roland, thanks for re-opening - it is indeed an interesting discussion.

Roland Karlsson wrote:

In the thread named "Will Thom's prediction come true" an interesting sub-thread started - and then the entire thread did hit 150 - pang - thread closed. A rather irritation rule to say the least.

The sub-thread was about what advantages Foveon sensors have.

I said: the main advantage is that all colors are sampled in the same place and did give the reference to http://www.dpreview.com/reviews/nikon-coolpix-a/14 , where you can look at the coin above Mickey Mouse. IN particular, look at the Fuji sample, and please look at ISO 200 for the Sigma sample.

Bobn2 said: its the only advantage.

Lin and Laurence said: there are more advantages, and Laurence also listed:

  1. Edge roll-off

Not sure what that means, unless its acuity.

  1. Acuity

Well, that is an 'advantage' of missing out the AA filter, not of the Foveon sensor per se. It's also achievable by missing out the AA filter on Bayer cameras. Myself, I'd prefer to nt have the aliasing artifacts, and an AA filter, even with a Foveon sensor.

  1. no moiré

Certainly there is 'moiré' - simply not colour moiré, which is the advantage of sampling all three channels at the same point, so that is the same advantage restated.

  1. color accuracy (an arguable point for many but absolutely achievable)

Indeed it is arguable, so arguable as to be false. The Foveon sensor cannot begin to compete with a Bayer sensor for colour accuracy. The filter characteristics of the silicon depth filter is way off the optimum to match human vision and there is no simple way of correcting this.

  1. greater "presence" or a 3D effect (nebulous but mentioned in so many publications over 10 years)

Is this a real effect? As soon as people start using words like 'presence' I think we're in the area of extreme subjectivism.

Personally I would categorise all Laurence's advantages, except the one about color accuracy (4), as a result of the fact that all colors are sampled at the same position. Laurence's list is a list seen from the user perspective, which of course also is a valid way of seeing it.

NOTE that no moire (3) should be no colored moiré. Moiré overall is also less than with Bayer without AA filters, but its absolutely not "no moiré".

I would concur, and add that the 'greater colour accuracy' is absolutely false.

So - whats the advantages?

NOTE - we can all make a list of disadvantages if we want to, but can we make a list of advantages?

And - color accuracy? Is that an advantage of Foveon, as Laurence claims?

I can add three more potential advantages, which (just as Laurence's color accuracy advantage) can be discussed.

  1. The color filters in Bayer sensors, do they age?

That is a very interesting question.

  1. It is theoretically possible to add a fourth IR layer for Foveon sensors

Yes, but one big problem with Foveon sensors is efficiency loss due to two effects, one of which is the lost regions between the layers. Adding a layer makes it even worse.

  1. potentially a sensor like Foveon could detect all photons, Bayer do filter away 2/3.

That's potentially the case, but in actuality Foveon sensors lose more photons than do Bayer. Bayer filters don't in fact lose 2/3 - that is true only for the very simplistic view of how the sensors work. The Bayer design is much more subtle than that. Typically the luminance channel, 'green', which covers half the sensor area captures much more than 1/3 of the available photons in a typical scene, the luminance channels ('red' and 'blue') capture a smaller proportion, but an average scene has a smaller proportion of those photons in any case.

-- hide signature --

Bob

bobn2
Forum ProPosts: 29,360
Re: Advantages of Foveon sensor
In reply to SigmaTog, Jun 23, 2013

SigmaTog wrote:

An advantage with Foveon is no CFA.

That is the advantage we are all agreed on.

The spin-off is that, with a Foveon full spectrum camera, UV & IR photography is possible with no false colour introduced by the CFA.

UV and IR photography naturally have 'false colour', since UV and IR don't take part in the tristimulus response of the human eye.

Another advantage is no CFA interpolation or color reconstruction, no CFA mush.

Restating the same advantage again.

-- hide signature --

Bob

Laurence Matson
Forum ProPosts: 11,173Gear list
Re: Advantages of Foveon sensor
In reply to bobn2, Jun 23, 2013

Bob (if that is your name),

Before we start up this old squeezebox, because you have decided to be all over this topic, and because you yourself once wrote:

"Better guide is what you yourself think about the cameras, try them see which you feel more comfortable with."

What is you real experience with a Sigma (or Polaroid branded but really Foveon) camera?

Do you have any samples of what you got?

My list below with your annotations reveals a considerable lack of familiarity with terms commonly used in this forum for over ten years. Whether they are correct or incorrect is immaterial unless you understand the jargon. We can redefine everything for you again, and then we can get into arguments about the definitions, and ultimately recycle some very old discussions. Or, you could use the search function and do some digging yourself. Or you could just continue to poke with your skepticism and ultimately leave everyone frustrated.

In short, do some homework please. The dpreview function now works quite well.

For instance:

http://www.dpreview.com/forums/post/8717374
http://www.dpreview.com/search/forums?query=%22edge+roll+off%22&page=3&forum=1027

There is, admittedly, some confusion about what this means and it is related to acuity, but they are not pure synonyms.

Bobn2 wrote:

Hi Roland, thanks for re-opening - it is indeed an interesting discussion.

Roland Karlsson wrote:

In the thread named "Will Thom's prediction come true" an interesting sub-thread started - and then the entire thread did hit 150 - pang - thread closed. A rather irritation rule to say the least.

The sub-thread was about what advantages Foveon sensors have.

I said: the main advantage is that all colors are sampled in the same place and did give the reference to http://www.dpreview.com/reviews/nikon-coolpix-a/14 , where you can look at the coin above Mickey Mouse. IN particular, look at the Fuji sample, and please look at ISO 200 for the Sigma sample.

Bobn2 said: its the only advantage.

Lin and Laurence said: there are more advantages, and Laurence also listed:

  1. Edge roll-off

Not sure what that means, unless its acuity.

  1. Acuity

Well, that is an 'advantage' of missing out the AA filter, not of the Foveon sensor per se. It's also achievable by missing out the AA filter on Bayer cameras. Myself, I'd prefer to nt have the aliasing artifacts, and an AA filter, even with a Foveon sensor.

  1. no moiré

Certainly there is 'moiré' - simply not colour moiré, which is the advantage of sampling all three channels at the same point, so that is the same advantage restated.

  1. color accuracy (an arguable point for many but absolutely achievable)

Indeed it is arguable, so arguable as to be false. The Foveon sensor cannot begin to compete with a Bayer sensor for colour accuracy. The filter characteristics of the silicon depth filter is way off the optimum to match human vision and there is no simple way of correcting this.

  1. greater "presence" or a 3D effect (nebulous but mentioned in so many publications over 10 years)

Is this a real effect? As soon as people start using words like 'presence' I think we're in the area of extreme subjectivism.

Personally I would categorise all Laurence's advantages, except the one about color accuracy (4), as a result of the fact that all colors are sampled at the same position. Laurence's list is a list seen from the user perspective, which of course also is a valid way of seeing it.

NOTE that no moire (3) should be no colored moiré. Moiré overall is also less than with Bayer without AA filters, but its absolutely not "no moiré".

I would concur, and add that the 'greater colour accuracy' is absolutely false.

So - whats the advantages?

NOTE - we can all make a list of disadvantages if we want to, but can we make a list of advantages?

And - color accuracy? Is that an advantage of Foveon, as Laurence claims?

I can add three more potential advantages, which (just as Laurence's color accuracy advantage) can be discussed.

  1. The color filters in Bayer sensors, do they age?

That is a very interesting question.

  1. It is theoretically possible to add a fourth IR layer for Foveon sensors

Yes, but one big problem with Foveon sensors is efficiency loss due to two effects, one of which is the lost regions between the layers. Adding a layer makes it even worse.

  1. potentially a sensor like Foveon could detect all photons, Bayer do filter away 2/3.

That's potentially the case, but in actuality Foveon sensors lose more photons than do Bayer. Bayer filters don't in fact lose 2/3 - that is true only for the very simplistic view of how the sensors work. The Bayer design is much more subtle than that. Typically the luminance channel, 'green', which covers half the sensor area captures much more than 1/3 of the available photons in a typical scene, the luminance channels ('red' and 'blue') capture a smaller proportion, but an average scene has a smaller proportion of those photons in any case.

-- hide signature --

Bob

-- hide signature --

Laurence
laurence at appledore-farm dot com
"I thought: I read something in a book, I dream, I imagine, and it comes true. And it is exactly like this in life.
"You can dream, and it comes true, as long as you can get out of the certitudes. As long as you can get a pioneering spirit, as long as you can explore, as long as you can think off the grid. So much time we spend in our education, in our lives is spent learning certitudes, learning habits, trying to fight against the unknown, to avoid the doubts and question marks. As soon as you start to love the unknown, to love the doubts, to love the question marks, life becomes an absolutely fabulous adventure."
Bertrand Piccard, a Swiss person
http://www.pbase.com/lmatson
http://www.pbase.com/sigmadslr
http://www.howardmyerslaw.com

Kendall Helmstetter Gelner
Forum ProPosts: 18,353Gear list
Relative color accuracy
In reply to Roland Karlsson, Jun 23, 2013

Roland Karlsson wrote:

<...>

And - color accuracy? Is that an advantage of Foveon, as Laurence claims?

I have done a number of model shoots now, where a variety of people are shooting the same models in the same location (often at the same time) in a variety of poses.

Some things I have learned from this:

1) No camera is "accurate", at least not all the time.

2) The Sigma cameras seem to come closest on average.

3) One aspect of accuracy I think the Foveon sensors do have, is relative color accuracy.  If you've not blown out any color channels the relations of colors to each other come through really well, and adjustments you make can bring a whole image in line instead of having to tweak individual colors much.

 Kendall Helmstetter Gelner's gear list:Kendall Helmstetter Gelner's gear list
Sigma 70-200mm F2.8 EX DG OS HSM Sigma 8-16mm F4.5-5.6 DC HSM Sigma 24-70mm F2.8 EX DG HSM Sigma 50-500mm F4.5-6.3 DG OS HSM Sigma 85mm F1.4 EX DG HSM +4 more
Kendall Helmstetter Gelner
Forum ProPosts: 18,353Gear list
There is at least one absolute, and you refuse it.
In reply to bobn2, Jun 23, 2013

Bobn2 wrote:

<...>

That's potentially the case, but in actuality Foveon sensors lose more photons than do Bayer.

In actuality they lose almost none, can you truly disagree that most of the photons hitting the sensor end up in the sensor at one depth or another?.  It's the processing that matters.

Bayer filters don't in fact lose 2/3

In discussions in photography there are a lot of things that are more subjective than not, because there are so many moving parts - things like color accuracy for example, when people could use any number of options for processing RAW, all of which have their own take on what color means.  All of that fed into monitors that may more may not be calibrated, may or may not be calibrated well. Then viewed by eyes that may will interpret color slightly differently one person to the next...

However in this world of subjective reflected dream capturing that we all take part in, there is at least one physical absolute that remains true now and forever.  By definition, a bayer filter does in fact block 2/3 of all photons from ever reaching the sensor.  That's just how filters work.  When talking about photons it does not matter that humans happen to "see" luminance more clearly, it's a matter of the most basic physics, a colored filter letting through some wavelengths and not others.

If you can't admit that simple and obvious absolute, what are we to make of the rest of your positions?

 Kendall Helmstetter Gelner's gear list:Kendall Helmstetter Gelner's gear list
Sigma 70-200mm F2.8 EX DG OS HSM Sigma 8-16mm F4.5-5.6 DC HSM Sigma 24-70mm F2.8 EX DG HSM Sigma 50-500mm F4.5-6.3 DG OS HSM Sigma 85mm F1.4 EX DG HSM +4 more
mike earussi
Veteran MemberPosts: 4,732
Re: Advantages of Foveon sensor
In reply to bobn2, Jun 23, 2013

I would never claim "greater color accuracy" as one of the Foveon advantages, as color accuracy seems to be more a result of the software processing the raw files than any inherent property of the chip (as such, many Bayer cameras have greater color accuracy than Sigma's Foveon bodies). I would, though, claim a greater possibility for more tonal variations within a specific color range due to the high overlap of colors in the chip compared to more narrow color cutoff of Bayer filtration.

DMillier
Forum ProPosts: 17,521
Re: There is at least one absolute, and you refuse it.
In reply to Kendall Helmstetter Gelner, Jun 23, 2013

Filter loss would only 2/3 if the filters were brick wall surely? Modern sensors tend to have weaker filters too, I think.

-- hide signature --
DMillier
Forum ProPosts: 17,521
Re: Relative color accuracy
In reply to Kendall Helmstetter Gelner, Jun 23, 2013

I would say every sigma model seems to have different colour. The newer ones look better to me' more bluish' less yellow green. Five on method is prone to getting wildly wrong sometimes. Desaturates colours in shadows too. But overall, colours look good these days.

-- hide signature --
xpatUSA
Senior MemberPosts: 3,007Gear list
Re: There is at least one absolute, and you refuse it.
In reply to Kendall Helmstetter Gelner, Jun 23, 2013

Kendall Helmstetter Gelner wrote:

Bobn2 wrote:

<...>

That's potentially the case, but in actuality Foveon sensors lose more photons than do Bayer.

In actuality they lose almost none, can you truly disagree that most of the photons hitting the sensor end up in the sensor at one depth or another?. It's the processing that matters.

We seem to be forgetting Quantum Efficiency (QE) in this thread. QE is the measure of how many photons get converted to electrons and are thereby detected in the sensitive parts (diffusion zones) of the sensor. Most of the photons are not detected and many pass all the way through the silicon and exit unimpeded from the back of the slab.

Early Foveon F7 sensors had quite a low QE - thirty something per cent max (which Foveon hid by taking the spectral average of each layer and adding them together - shades of 10MP versus 3.4MP).

The low QE was exacerbated by the broad spectral response of each layer which causes quite a bit of subtraction during interpolation. Small wonder that the SD9 was limited to ISO 400. For this very reason, Foveon's first step downward was taken in the F7-C9110 sensor by increasing the fill factor through the use of a micro-lens layer added to the sensor face.

By comparison, Bayer sensors - with their more separated and peaky CFA spectral responses - generally have higher QEs than later Foveons (is my guess). Does anyone have comparative figures for Merrills vs. Bayers?

-- hide signature --

Regards,
Ted http://kronometric.org
SD9, SD10, EF-500, GH1.

 xpatUSA's gear list:xpatUSA's gear list
Sigma SD9 Sigma SD10 Sigma SD14 Panasonic Lumix DMC-GH1 +2 more
SandyF
Forum ProPosts: 14,875Gear list
Re: Advantages of Foveon sensor
In reply to mike earussi, Jun 23, 2013

mike earussi wrote:

I would never claim "greater color accuracy" as one of the Foveon advantages, as color accuracy seems to be more a result of the software processing the raw files than any inherent property of the chip (as such, many Bayer cameras have greater color accuracy than Sigma's Foveon bodies). I would, though, claim a greater possibility for more tonal variations within a specific color range due to the high overlap of colors in the chip compared to more narrow color cutoff of Bayer filtration.

I agree, greater tonal variation. For examples with multiple greens, look at all the green variations iwithin one photo in my new Hawaiian vegetation shots on flickr. I find the Sigma cameras capture these variations better than my Canon 5DII. Of course since I didn't take the Canon to Hawaiian, I cannot prove this to DMillier's satisfaction LOL

Best regards, Sandy
http://www.pbase.com/sandyfleischman (archival)
http://www.flickr.com/photos/sandyfleischmann (current, new Hawaii photos)

 SandyF's gear list:SandyF's gear list
Sigma DP2 Sigma DP1 Sigma DP2 Merrill Canon EOS 5D Mark II Sigma SD9 +5 more
Lin Evans
Forum ProPosts: 15,642Gear list
Re: Advantages of Foveon sensor
In reply to mike earussi, Jun 23, 2013

Hi Mike,

The issue with color "accuracy," as you say, is primarily one of post capture processing, however, no amount of post processing can retrieve subtle variations of hue which were not captured in the first place.

Greater tonal variations within a color range allow closer approximations of reality in terms of color reproduction. As I have said previously, I've noticed this primarily in greens myself, but Mike Chaney, developer of Qimage Ultimate software, did a very nice analysis using his Sigma camera and his Canon 5D and demonstrated that with red hues, the same issues exists with CFA based sensors. In fact, it's quite easy to look at color resolution charts and quickly see that color resolution from the Foveon sensor is uniform regardless of hue. On the other hand, CFA sensors lose a great deal of resolution primarily in the reds so that when photographing flowers, as Mike did for his samples, the differences in color "accuracy" were quite easy to see. The Foveon captured the full red spectrum. The CFA (5D) could not render the reds as his eyes actually saw them in the flower.

So in the case with reds, and in my own experiments with greens, the Foveon can reproduce very real color subtleties which exist in reality which the CFA simply doesn't "see." Therefore I think it's fair to say that an advantage of the Foveon sensor is better color accuracy. It's more difficult to achieve over the full color spectrum and requires careful lighting and lower ISO, but as Laurence correctly stated, it "is" achievable in my experience. Most of my professional work involved very precise matching of colors for gallery fine art. I found it quite possible to please my most discriminating clients using Sigma cameras.

Best regards,

Lin

mike earussi wrote:

I would never claim "greater color accuracy" as one of the Foveon advantages, as color accuracy seems to be more a result of the software processing the raw files than any inherent property of the chip (as such, many Bayer cameras have greater color accuracy than Sigma's Foveon bodies). I would, though, claim a greater possibility for more tonal variations within a specific color range due to the high overlap of colors in the chip compared to more narrow color cutoff of Bayer filtration.

Laurence Matson
Forum ProPosts: 11,173Gear list
Really fine greens
In reply to SandyF, Jun 23, 2013

http://www.dpreview.com/forums/post/51661758

SandyF wrote:

mike earussi wrote:

I would never claim "greater color accuracy" as one of the Foveon advantages, as color accuracy seems to be more a result of the software processing the raw files than any inherent property of the chip (as such, many Bayer cameras have greater color accuracy than Sigma's Foveon bodies). I would, though, claim a greater possibility for more tonal variations within a specific color range due to the high overlap of colors in the chip compared to more narrow color cutoff of Bayer filtration.

I agree, greater tonal variation. For examples with multiple greens, look at all the green variations iwithin one photo in my new Hawaiian vegetation shots on flickr. I find the Sigma cameras capture these variations better than my Canon 5DII. Of course since I didn't take the Canon to Hawaiian, I cannot prove this to DMillier's satisfaction LOL

-- hide signature --

Laurence
laurence at appledore-farm dot com
"I thought: I read something in a book, I dream, I imagine, and it comes true. And it is exactly like this in life.
"You can dream, and it comes true, as long as you can get out of the certitudes. As long as you can get a pioneering spirit, as long as you can explore, as long as you can think off the grid. So much time we spend in our education, in our lives is spent learning certitudes, learning habits, trying to fight against the unknown, to avoid the doubts and question marks. As soon as you start to love the unknown, to love the doubts, to love the question marks, life becomes an absolutely fabulous adventure."
Bertrand Piccard, a Swiss person
http://www.pbase.com/lmatson
http://www.pbase.com/sigmadslr
http://www.howardmyerslaw.com

dmaclau
Senior MemberPosts: 1,395Gear list
Color Accuracy as Reality
In reply to Kendall Helmstetter Gelner, Jun 23, 2013

I'm starting to believe that a parasite of sorts infects our brain at birth and subtly alters it's chemistry throughout our lives.  How else to understand all the different simultaneous "realities."  Colors are reality when seen, but only then.  They're then processed by a lens, camera, computer, screen and perhaps a printer.  The color guide through each of these steps is our own infallible memory of what we thought we saw.  Hmmm.

Color perception like art is not scientifically definable.  Not yet at least.  Within it's range a Foveon sensored camera is capable of quite unique and stunning imagery.  This is certainly true for colors and transfers quite well into B&W.

What I'm reading here are various attempts to parse that into individual elements which may then be "scientifically" dis-proved.  OK.  If that's what floats your boat.  I have to ask what is to be gained though.

Although most images are not good, I seem to generate more little masterpieces with the Sigma than with any other camera I've used.  The color signature is quite different and pleasing and they are much better for B&W.

Go ahead and scientifically prove that a photon can't double fribble a pixel.  Or demonstrate why upsizing one image and combining with a downsized image from another camera and then sideways sizing the lot while rubbing a cut potato on a wart...what on earth are you getting on about?  Are you creating images you're pleased with?  No?  Move on.  Yes?  That's wonderful!  Not shooting at all?  Pull up your socks and get out of the house.

 dmaclau's gear list:dmaclau's gear list
Fujifilm FinePix X100 Sigma DP3 Merrill Panasonic Lumix DMC-G2 Nikon 1 V1 Fujifilm X-M1 +1 more
bobn2
Forum ProPosts: 29,360
Re: Advantages of Foveon sensor
In reply to Laurence Matson, Jun 23, 2013

Laurence Matson wrote:

Bob (if that is your name),

Hi Laurence (if you're not posting under a pseudonym)

Before we start up this old squeezebox, because you have decided to be all over this topic, and because you yourself once wrote:

"Better guide is what you yourself think about the cameras, try them see which you feel more comfortable with."

That is a something that I say often, and that I think is very important. I have no argument with anyone's personal choice of camera, or any reason that they choose it. These things are personal, and none of us has any right to question another's preferences. So, if someone prefers Sigma colours, or Sigma rendition, or anything else about their camera, that is their right and privilege, and you will not find me ever saying anything else. That, however is not what this thread is about. This thread is about what are the advantages of Foveon sensors. That is an objective, not subjective question, one of engineering fact, not personal preference.

What is you real experience with a Sigma (or Polaroid branded but really Foveon) camera?

Do you have any samples of what you got?

All of course irrelevant, since we aren't talking about cameras, just sensors.

My list below with your annotations reveals a considerable lack of familiarity with terms commonly used in this forum for over ten years. Whether they are correct or incorrect is immaterial unless you understand the jargon.

The question would more be whether you understand the jargon.

We can redefine everything for you again, and then we can get into arguments about the definitions, and ultimately recycle some very old discussions. Or, you could use the search function and do some digging yourself. Or you could just continue to poke with your skepticism and ultimately leave everyone frustrated.

In short, do some homework please. The dpreview function now works quite well.

For instance:

http://www.dpreview.com/forums/post/8717374

Uses the term 'roll-off at edges' but is quite unrevealing about what it means, and has not much precise or meaningful.

http://www.dpreview.com/search/forums?query=%22edge+roll+off%22&page=3&forum=1027

The search is similarly unrevealing about what is meant by 'edge roll-off'.

There is, admittedly, some confusion about what this means and it is related to acuity, but they are not pure synonyms.

So, then, what does it mean? What is it other than acuity?

-- hide signature --

Bob

Lin Evans
Forum ProPosts: 15,642Gear list
Re: Color Accuracy as Reality
In reply to dmaclau, Jun 23, 2013

Although I agree in substance, we can indeed often do more than use our "own infallible memories" when it comes to verifying colors. Obviously, when shooting most landscapes, wildlife and other ephemeral subjects, we necessarily depend on our memories to guide us in recreating our subject accurately. But when we are photographing art or still life, taking portraits or other subject matter which we can, post processing, view and compare with the results of our manipulations, know pretty accurately whether we have been successful.

In the end, however, as you say it's really about creating images which either you are pleased with.

Best regards,

Lin

dmaclau wrote:

I'm starting to believe that a parasite of sorts infects our brain at birth and subtly alters it's chemistry throughout our lives. How else to understand all the different simultaneous "realities." Colors are reality when seen, but only then. They're then processed by a lens, camera, computer, screen and perhaps a printer. The color guide through each of these steps is our own infallible memory of what we thought we saw. Hmmm.

Color perception like art is not scientifically definable. Not yet at least. Within it's range a Foveon sensored camera is capable of quite unique and stunning imagery. This is certainly true for colors and transfers quite well into B&W.

What I'm reading here are various attempts to parse that into individual elements which may then be "scientifically" dis-proved. OK. If that's what floats your boat. I have to ask what is to be gained though.

Although most images are not good, I seem to generate more little masterpieces with the Sigma than with any other camera I've used. The color signature is quite different and pleasing and they are much better for B&W.

Go ahead and scientifically prove that a photon can't double fribble a pixel. Or demonstrate why upsizing one image and combining with a downsized image from another camera and then sideways sizing the lot while rubbing a cut potato on a wart...what on earth are you getting on about? Are you creating images you're pleased with? No? Move on. Yes? That's wonderful! Not shooting at all? Pull up your socks and get out of the house.

bobn2
Forum ProPosts: 29,360
Re: There is at least one absolute, and you refuse it.
In reply to Kendall Helmstetter Gelner, Jun 23, 2013

Kendall Helmstetter Gelner wrote:

Bobn2 wrote:

<...>

That's potentially the case, but in actuality Foveon sensors lose more photons than do Bayer.

In actuality they lose almost none, can you truly disagree that most of the photons hitting the sensor end up in the sensor at one depth or another?. It's the processing that matters.

Sorry, you are incorrect. The quantum efficiency of Foveon sensors is low, meaning that they lose a lot of photons - as in fail to count them. There are two reasons for this. One is that there are dead bands between the layers, where the collected photoelectrons fail to make it to the gate of any read transistor. If those bands were not dead the colour selectivity would be even worse than it is. Secondly tehre is incomplete charge transfer from the lower layers. Incomplete charge transfer means, simply that photons don't get counted and the quantum efficiency is poort. poor QE is why Foveon sensor don't do so well at high ISO's - if they were truly capturing more of the photons than do Bayer sensors they would be better in low light - but they are not.

Bayer filters don't in fact lose 2/3

In discussions in photography there are a lot of things that are more subjective than not, because there are so many moving parts - things like color accuracy for example, when people could use any number of options for processing RAW, all of which have their own take on what color means. All of that fed into monitors that may more may not be calibrated, may or may not be calibrated well. Then viewed by eyes that may will interpret color slightly differently one person to the next...

However in this world of subjective reflected dream capturing that we all take part in, there is at least one physical absolute that remains true now and forever. By definition, a bayer filter does in fact block 2/3 of all photons from ever reaching the sensor. That's just how filters work. When talking about photons it does not matter that humans happen to "see" luminance more clearly, it's a matter of the most basic physics, a colored filter letting through some wavelengths and not others.

If you can't admit that simple and obvious absolute, what are we to make of the rest of your positions?

That 'simple and obvious absolute' is wrong, and derive from a misunderstanding of the principles of colour vision, the nature of filters and the how colour photography works. Indeed, there is no definition that says a Bayer filter blocks 2/3 of all photons ever reaching the sensor. Consider the following:

- The number of photons blocked in any case depends on the wavelength distribution of the light reaching the sensor. That distribution depends both on the colour temperature of the light source and the colour reflectivity of the subject.

- Filter passbands can and do overlap.

Now can you see that your 'simple and obvious absolute' is not at all absolute.

-- hide signature --

Bob

Lin Evans
Forum ProPosts: 15,642Gear list
Re: Advantages of Foveon sensor
In reply to bobn2, Jun 23, 2013

Bobn2 wrote:

Laurence Matson wrote:

Bob (if that is your name),

Hi Laurence (if you're not posting under a pseudonym)

Before we start up this old squeezebox, because you have decided to be all over this topic, and because you yourself once wrote:

"Better guide is what you yourself think about the cameras, try them see which you feel more comfortable with."

That is a something that I say often, and that I think is very important. I have no argument with anyone's personal choice of camera, or any reason that they choose it. These things are personal, and none of us has any right to question another's preferences. So, if someone prefers Sigma colours, or Sigma rendition, or anything else about their camera, that is their right and privilege, and you will not find me ever saying anything else. That, however is not what this thread is about. This thread is about what are the advantages of Foveon sensors. That is an objective, not subjective question, one of engineering fact, not personal preference.

What is you real experience with a Sigma (or Polaroid branded but really Foveon) camera?

Do you have any samples of what you got?

All of course irrelevant, since we aren't talking about cameras, just sensors.

Of course it's "very" relevant since, in the relevant sense, only Sigma cameras and one Polaroid model use Foveon sensors.

My list below with your annotations reveals a considerable lack of familiarity with terms commonly used in this forum for over ten years. Whether they are correct or incorrect is immaterial unless you understand the jargon.

The question would more be whether you understand the jargon.

I think Laurence understands the "jargon" very well Bob. But obviously you are having issues.

We can redefine everything for you again, and then we can get into arguments about the definitions, and ultimately recycle some very old discussions. Or, you could use the search function and do some digging yourself. Or you could just continue to poke with your skepticism and ultimately leave everyone frustrated.

In short, do some homework please. The dpreview function now works quite well.

For instance:

http://www.dpreview.com/forums/post/8717374

Uses the term 'roll-off at edges' but is quite unrevealing about what it means, and has not much precise or meaningful.

It' is "very" precise Bob, but because you don't know and understand the term, your knowledge of how to relate to it is faulty.

Normally, I wouldn't bother to repeat what is quite commonly known here on the Sigma forum, but for your elucidation: Edge roll-off is a very simple concept. How many pixels are there which separate adjacent detail objects? If you photograph a very plain and featureless pure black object with sharp edges on a pure white and featureless backdrop, at the edge boundaries there is not a single pixel of black adjacent to a single pixel of white. There are a number of pixels, from the black perspective of varying shades of gray until finally a pixel of 255, 255, 255 white. If you make the identical photograph with a CFA based camera using identical focal length, identical pixel file dimensions, identical ISO, identical lens and other relevant settings, there will be fewer pixels separating the white and black with the Foveon sensor. This is irrespective of the AA filter which adds to the number of these pixels. The CFA process itself creates additional transiting pixels. The Foveon image is therefore sharper than the CFA image. It's been tested, measured and posted on this forum and discussed numerous times.

Lin

http://www.dpreview.com/search/forums?query=%22edge+roll+off%22&page=3&forum=1027

The search is similarly unrevealing about what is meant by 'edge roll-off'.

Then it's your search skills or the amount of effort you wish to put into finding it.

There is, admittedly, some confusion about what this means and it is related to acuity, but they are not pure synonyms.

So, then, what does it mean? What is it other than acuity?

-- hide signature --

Bob

bobn2
Forum ProPosts: 29,360
Re: There is at least one absolute, and you refuse it.
In reply to xpatUSA, Jun 23, 2013

xpatUSA wrote:

Kendall Helmstetter Gelner wrote:

Bobn2 wrote:

<...>

That's potentially the case, but in actuality Foveon sensors lose more photons than do Bayer.

In actuality they lose almost none, can you truly disagree that most of the photons hitting the sensor end up in the sensor at one depth or another?. It's the processing that matters.

We seem to be forgetting Quantum Efficiency (QE) in this thread. QE is the measure of how many photons get converted to electrons and are thereby detected in the sensitive parts (diffusion zones) of the sensor. Most of the photons are not detected and many pass all the way through the silicon and exit unimpeded from the back of the slab.

That would make the silicon transparent, which it is not. The conversion rate for photons to free electrons is in fact pretty good in most silicon sensors. The problem is how much of that electron charge makes its way to the gates of the pixel source follower transistors. That depends on the right electric field patterns to transport those electrons to the gates. The ones that don't find their way to the gates in the end find their way to a 'hole', releasing the original photon energy as heat.

One of the reason for the Foveon's low QE is the difficulty in engineering the electric fields to ensure the collection of all the electrons.

Early Foveon F7 sensors had quite a low QE - thirty something per cent max (which Foveon hid by taking the spectral average of each layer and adding them together - shades of 10MP versus 3.4MP).

The low QE was exacerbated by the broad spectral response of each layer which causes quite a bit of subtraction during interpolation. Small wonder that the SD9 was limited to ISO 400. For this very reason, Foveon's first step downward was taken in the F7-C9110 sensor by increasing the fill factor through the use of a micro-lens layer added to the sensor face.

By comparison, Bayer sensors - with their more separated and peaky CFA spectral responses - generally have higher QEs than later Foveons (is my guess). Does anyone have comparative figures for Merrills vs. Bayers?

-- hide signature --

Regards,
Ted http://kronometric.org
SD9, SD10, EF-500, GH1.

-- hide signature --

Bob

Keyboard shortcuts:
FForum MMy threads