Pages: 1 ... 5 6 [7] 8 9 10   Go Down

Author Topic: larger sensors  (Read 191431 times)

bjanes

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3387
larger sensors
« Reply #120 on: January 06, 2007, 01:44:52 pm »

Quote
Smaller pixels have greater gain, AOTBE, for the same ISO. 
[{POST_SNAPBACK}][/a]

No, John, you have it backwards. For CCD and CMOS imaging chips, [a href=\"http://www.photomet.com/library_enc_gain.shtml]gain[/url] is reported in terms of electrons/ADU.  A gain of 8 means that the camera system digitizes the CCD signal so that each ADU corresponds to 8 photoelectrons. This is the inverse of amplifier gain, and can be confusing.

A pixel with half the linear size of the above example might have a gain of 2 electrons/ADU. If both chips had a read noise of 4 electrons, the noise would be 0.5 ADU with the larger chip and 2 ADU for the smaller pixel.

I did have some typos in my previous message, which I have corrected, but the thrust of my assertion that a given read noise in electrons has a greater effect with a small pixel is still true.

Bill
Logged

bjanes

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3387
larger sensors
« Reply #121 on: January 06, 2007, 06:29:45 pm »

Quote
I think 1 electron read noise would be a great improvement.  0.1, even greater.

OK, I now see you had a typo in the previous sentence.
Read noise is the issue, and it is not proportional to absolute gain.  Read noise is *NOT* the amplification of existing noise in the sensor wells; it is noise *GENERATED* in the reading of the sensor.  That's why it is different as enumerated in electrons, at different ISOs in Roger's experiments, and in mine.

From some of the statements you have made, you seem to think that read noise is a quantum event, like the captured electrons.  When someone says"the read noise at ISO 100 is 30.1 electrons", this doesn't mean each pixel is off by some integer number of electrons, the standard deviation of which is 30.1 electrons for the entire image.  It means that the read noise was measured in ADUs, and then on assumed information about the relationship between ADUs and electrons for that ISO, the ADUs are translated into units of "electrons".  This figure has nothing at all to really do with sensor electrons and one can calculate the gain in electrons per ADU.

[{POST_SNAPBACK}][/a]

Read noise, like other forms of noise, can be expressed in terms of ADUs or electrons and it is expressed in terms per pixel, not for the entire picture as you seem to imply. Interested readers can refer to this [a href=\"http://www.qsimaging.com/ccd_noise.html]reference[/url], a portion of which is quoted below:

"CCD manufacturers measure and report CCD noise as a number of electrons RMS (Root Mean Square).  You’ll typically see it presented like this, 15eˉ RMS, meaning that with this CCD, you should expect to see about 15 electrons of noise per pixel.  More precisely, 15eˉ RMS is the standard deviation around the mean pixel value."

Nowhere did I state that read noise was some type of constant or offset and not related to ISO. The 3-4 electron value to which I referred is for the ISO at unity gain. At base ISO under normal photographic conditions, noise is shot limited, not read limited. The read noise of 30 electrons you quoted at ISO 100 might be related to a full well value of 80,000 electrons, and is not significant except in the deepest shadows. With a full well of 80,000 electrons, the shadows 10 f/stops below clipping would still have 78 electrons.

Read noise is normally determined by subtracting two bias frames as described in the above reference or in Roger's essay. The bias frames could include the whole picture, but one normally measures a representative cropped area from the center, perhaps an area of 256 by 256 pixels. The results can be shown as a histogram or expressed as a SD. If the sensor is completely uniform, one could measure a single pixel 65,536 times. One can easily convert from ADUs to electrons, since the number of electrons is the square of the signal to noise ratio as Roger explains.

The main source of Read Noise is from the on chip pre-amplifier.

Contrary to what you imply, the number of electrons collected is not constant for a given luminance, but follows a Poisson distribution.

Bill
« Last Edit: January 06, 2007, 10:21:34 pm by bjanes »
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
larger sensors
« Reply #122 on: January 06, 2007, 08:19:45 pm »

Bill,
A lot of readers are probably not going to understand the significance of these differences of opinion between you and John, but probably do appreciate the fact that read noise, and noise in general, as a proportion of the total signal, will be greater for the smaller pixel, all else being equal.

However, all else is rarely equal. Developments in one area can compensate for deficiencies in another area. I see no reason to suppose there is no further scope for improvement with regard to (1) actual pixel size, as opposed to pixel pitch, (2) read noise as a proportion of the signal.

For example, is there any ultimate technological impossibility of having all the photon-collecting receptors on one side of the sensor, and all the signal processors for each photodiode on the other side of the sensor, directly behind, just a few microns away? Maybe such a chip would be too expensive to manufacture. There might be lots of reasons why such a solution would not be practical. I wouldn't know.

It would be interesting to see a direct comparison between a 1Ds and a 400D, cropping the 1Ds image to 10mp and taking both shots from an appropriately different distance to keep the FoV identical and/or using lenses with equal MTF50 responses at resolutions corresponding to the resolution limits of the 2 sensors. (That is, the lens used with the 1Ds should have a MTF50 response at, say 50 lp/mm and the lens used with the 400D should have 50%MTF at around 70 lp/mm .)

We could then get a clearer idea as to how a modern 'small' pixel compares with a slightly older 'bigger' pixel.  
Logged

John Sheehy

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 838
larger sensors
« Reply #123 on: January 07, 2007, 08:31:26 am »

Quote
No, John, you have it backwards. For CCD and CMOS imaging chips, gain is reported in terms of electrons/ADU.  A gain of 8 means that the camera system digitizes the CCD signal so that each ADU corresponds to 8 photoelectrons. This is the inverse of amplifier gain, and can be confusing.
If that terminology is used like that somewhere, then it *is* confusing, and a person with an interest in meaningful terminology should take no part in propagating such terminology.  Gain at a loss and with no "gain" - what a concept.

Quote
A pixel with half the linear size of the above example might have a gain of 2 electrons/ADU. If both chips had a read noise of 4 electrons, the noise would be 0.5 ADU with the larger chip and 2 ADU for the smaller pixel.
What are the actual results?; that's what matters.  Does squeezing 9x as many pixels into the same area more than triple read noise levels?  Squeezing 4x more than doubles it?  Your examples are all hypothetical, and don't answer the question.

Quote
I did have some typos in my previous message, which I have corrected, but the thrust of my assertion that a given read noise in electrons has a greater effect with a small pixel is still true.[a href=\"index.php?act=findpost&pid=94163\"][{POST_SNAPBACK}][/a]
The read noise *IS* the effect.  It exists relative to absolute signal, and it exists relative to the DR of the digitization.  What are its measurements?
Logged

bjanes

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3387
larger sensors
« Reply #124 on: January 07, 2007, 09:29:01 am »

Quote
If that terminology is used like that somewhere, then it *is* confusing, and a person with an interest in meaningful terminology should take no part in propagating such terminology.  Gain at a loss and with no "gain" - what a concept.
[{POST_SNAPBACK}][/a]
All I can say here is that I did not originate the terminology. However, communication is enhanced when you use accepted terminology rather than making up your own definitions. When Rome do as the Romans do.
Quote
What are the actual results?; that's what matters.  Does squeezing 9x as many pixels into the same area more than triple read noise levels?  Squeezing 4x more than doubles it?  Your examples are all hypothetical, and don't answer the question.
[a href=\"index.php?act=findpost&pid=94292\"][{POST_SNAPBACK}][/a]
Look at Rogers [a href=\"http://www.clarkvision.com/imagedetail/digital.sensor.performance.summary/]Figure 3[/url]. You will see that read noise is not correlated well with pixel size and in current Canon cameras is about 3-4 electrons with small and large pixels. If you squeeze more pixels into a given area, read noise remains more or less constant per pixel, but the effect of this noise is worse with the resultant small pixels because of the effect of gain when converting from electrons to ADU. One way to circumvent this problem is via binning where a group of pixels is readout as one superpixel, and where the read noise for the superpixel is the same as for the individual pixel. This ability is usually present only on scientific systems. For interested readers, more information is given here.
 
Some readers may find the signal/noise calculator on the Nikon site to be helpful. Integration time is exposure and one can adapt the concept to demonstrate statistics for the zones in the image. The question is answered to my satisfaction by the above references.

Bill
Logged

John Sheehy

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 838
larger sensors
« Reply #125 on: January 08, 2007, 01:21:14 am »

Quote
All I can say here is that I did not originate the terminology. However, communication is enhanced when you use accepted terminology rather than making up your own definitions. When Rome do as the Romans do.

Burn it?

I have never heard anyone but you use terminology like that.  YOU are the propagator, AFAIC.

Quote
Look at Rogers Figure 3. You will see that read noise is not correlated well with pixel size and in current Canon cameras is about 3-4 electrons with small and large pixels.

I don't see any charts or ways to derive this information on that page.  The real information of value, for my interests, is not given.

That chart doesn't tell what ISO this is determined for.  I'd be more interested in ADUs than electrons, anyway, but whatever the unit, give the range.  Blackframe read noise at every ISO, at the very least, for all cameras concerned, and how many effective RAW levels from black to saturation, how many electrons for the same, etc, etc.  I can't work with Roger's data.  From what I see, he ignores a lot of important things.  He seems to believe (and bases calculations on  the assumption) that all cameras use 4096 RAW levels, for instance, which is not true, and whitepoint is different, too.

Quote
If you squeeze more pixels into a given area, read noise remains more or less constant per pixel, but the effect of this noise is worse with the resultant small pixels because of the effect of gain when converting from electrons to ADU.

Get me the read noise in a useable form.  Roger's "electrons" are without a useful context, IMO.

Quote
One way to circumvent this problem is via binning where a group of pixels is readout as one superpixel, and where the read noise for the superpixel is the same as for the individual pixel. This ability is usually present only on scientific systems.[a href=\"index.php?act=findpost&pid=94302\"][{POST_SNAPBACK}][/a]

Your eyes do similar, automatically, when you view an ultra-hires image.  There is no need to bin.  You only see increased noise from smaller pixels when you zoom in to the same *PIXEL* resolution.  When you let your eyes do it, you keep the detail.
Logged

eronald

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6642
    • My gallery on Instagram
larger sensors
« Reply #126 on: January 08, 2007, 07:02:38 am »

I don't know why, but Roger's article looks a bit out of date. Cherry pick the terminology, ignore the content ...

Edmund
Logged
If you appreciate my blog posts help me by following on https://instagram.com/edmundronald

bjanes

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3387
larger sensors
« Reply #127 on: January 08, 2007, 07:55:15 am »

Quote
I have never heard anyone but you use terminology like that.  YOU are the propagator, AFAIC.
[{POST_SNAPBACK}][/a]

Apparently you don't read the references that are given to you.

[a href=\"http://www.clarkvision.com/imagedetail/evaluation-1d2/index.html]Clark[/url]. Refer to Table 1a and the accompanying text. (gain is in electrons/ADU)

University Paper gain = #electrons per pixel/ # counts per pixel (ADU)

Roper Scientific paper gain = electrons / ADU

Quote
That chart doesn't tell what ISO this is determined for.  I'd be more interested in ADUs than electrons, anyway, but whatever the unit, give the range.  Blackframe read noise at every ISO, at the very least, for all cameras concerned, and how many effective RAW levels from black to saturation, how many electrons for the same, etc, etc.  I can't work with Roger's data.  From what I see, he ignores a lot of important things.  He seems to believe (and bases calculations on  the assumption) that all cameras use 4096 RAW levels, for instance, which is not true, and whitepoint is different, too.
[{POST_SNAPBACK}][/a]

If you look at any spec sheet for a sensor you will see a single read noise in electrons RMS. Since read noise is dependent on readout rate, that value is also given, but for most digital cameras used in routine photography the rate is fixed. The read noise is for unity gain.

[a href=\"http://www.kodak.com/ezpres/business/ccd/global/plugins/acrobat/en/productsummary/FullFrame/KAF-39000ProductSummaryRev.2.0.pdf]Kodak KAF 3900[/url]
Quote
Your eyes do similar, automatically, when you view an ultra-hires image.  There is no need to bin.  You only see increased noise from smaller pixels when you zoom in to the same *PIXEL* resolution.  When you let your eyes do it, you keep the detail.
[{POST_SNAPBACK}][/a]

You have completely ignored the principle behind binning. The pixels have to be binned into one super pixel prior to readout. In that case, the read noise for the super pixel is the same as the individual pixel. Your eyes do not do this automatically.

[a href=\"http://www.photomet.com/library_enc_binning.shtml]Roper Scientific[/url] explains these principles.

I formerly thought that you were quite knowledgeable, but you are beginning to lose credibility with my due to your outrageous statements not backed up by any references or data.
« Last Edit: January 08, 2007, 08:04:14 am by bjanes »
Logged

bjanes

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3387
larger sensors
« Reply #128 on: January 08, 2007, 08:07:20 am »

Quote
I don't know why, but Roger's article looks a bit out of date. Cherry pick the terminology, ignore the content ...

Edmund
[a href=\"index.php?act=findpost&pid=94488\"][{POST_SNAPBACK}][/a]

Edmund,

Who is doing the cherry picking and ignoring content here. Your post is not clear. BTW,
Roger has just updated his sensor analysis as of December 2006, and like an academic, he supplies references.

Bill
Logged

John Sheehy

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 838
larger sensors
« Reply #129 on: January 08, 2007, 09:35:56 am »

Quote
Apparently you don't read the references that are given to you.
Sometimes I do and sometimes I don't, but my experience has been that most of your references do not address what I thought we were talking about, and if you think they do, then I don't think you understand them or what we were talking about (and on top of that, they may just be wrong).  A URL only shows what someone else thinks.

What we were talking about here is the fact that statistical noise is not *VISIBLE* noise.  I, and others said that more and smaller pixels in the same size sensor does not result in a noisier image, necessarily.  You seemed to disagreed with it, but never did anything to discredit the idea that had any relevance.

Quote
You have completely ignored the principle behind binning.  The pixels have to be binned into one super pixel prior to readout. In that case, the read noise for the super pixel is the same as the individual pixel. Your eyes do not do this automatically.
Not exactly, but I purposely used the word "similar", in reference to the reduction of visible noise.

Frankly,  I would think that any such system will work better in theory than practice.  Do you have an example of a sensor that does this and has truly lower read noise relative to signal than would be had with software binning?  Downsampling and software binning reduce read noise quite a bit, too.  I think it is generally better to leave the image in its higher-res state.  Besides having more visible detail, this allows future overlapped binning at the original resolution minus n-1 pixels in each dimension.  The binnable neighbors for a CFA camera will be very far from each other, as well.

Quote
I formerly thought that you were quite knowledgeable, but you are beginning to lose credibility with my due to your outrageous statements not backed up by any references or data.
[a href=\"index.php?act=findpost&pid=94492\"][{POST_SNAPBACK}][/a]
What outrageous statements?  I remember saying that noise is relevant to image viewing not just in statistical intensity, but in frequency content relative to the image.  Then, you start throwing URLs at me that completely ignore the spatial aspects of noise, like Bugs Bunny throwing banana peels behind him while being pursued, or someone tossing smokebombs.  Now, you've gone on tangents like some poorly-thought-out use of the word "gain", as if that had any bearing on anything being discussed.  Now, hardware binning.  Let's get back to what we were talking about; the idea that microscopic noise is virtually lower-noise.  Can you find some way to disprove that?
« Last Edit: January 08, 2007, 09:45:29 am by John Sheehy »
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
larger sensors
« Reply #130 on: January 08, 2007, 09:55:37 am »

Quote
The pixels have to be binned into one super pixel prior to readout. In that case, the read noise for the super pixel is the same as the individual pixel. Your eyes do not do this automatically.

Roper Scientific explains these principles.
[a href=\"index.php?act=findpost&pid=94492\"][{POST_SNAPBACK}][/a]

Bill,
I read Roper Scientific's articles years ago when they were one of the few sources on the net explaining the basic principles of imaging devices. I think you have to read between the lines sometimes. This is what they actually wrote on the issue of binning.

Quote
However, in binning mode, read noise is added to each superpixel, which has the combined signal from multiple pixels. In the ideal case, this produces SNR improvement equal to the binning factors (4x in the above example).

I recall BJL commented on this a while back. In practice, the read-noise of the superpixel is somewhat greater than the read-noise of a single small pixel, but of course not as great as the sum of the read-noise of all the individual pixels before they were binned.
Logged

John Sheehy

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 838
larger sensors
« Reply #131 on: January 08, 2007, 10:25:43 am »

Quote
I recall BJL commented on this a while back. In practice, the read-noise of the superpixel is somewhat greater than the read-noise of a single small pixel, but of course not as great as the sum of the read-noise of all the individual pixels before they were binned.
[a href=\"index.php?act=findpost&pid=94511\"][{POST_SNAPBACK}][/a]
...and you don't even get the full sum in a software binning Or even a downsample).  For a software binning, noise is reduced to 1/n, where n is the linear binning factor (2, where a 2x2 tile is binned into one).
Logged

jwoolf

  • Newbie
  • *
  • Offline Offline
  • Posts: 12
larger sensors
« Reply #132 on: January 08, 2007, 11:02:57 am »

Rainer,

There is a new, exciting technology from Seitz and Dalsa which is just coming to market.  The D3 scanning back uses a 60mm linear array that is 7,500 pixels high.  It is 100 times faster and more sensative than the standard scan back technology.  It is full frame medium format.  Approximately 6cm x 7cm.  About 70 megapixels!!!!  

Here is link to info:

http://www.roundshot.ch/xml_1/internet/de/...8/d925/f931.cfm

John Woolf
Digital Systems Manager
Museum of Fine Arts
Boston/USA
Logged

bjanes

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3387
larger sensors
« Reply #133 on: January 08, 2007, 11:10:11 am »

Quote
Bill,
I read Roper Scientific's articles years ago when they were one of the few sources on the net explaining the basic principles of imaging devices. I think you have to read between the lines sometimes. This is what they actually wrote on the issue of binning.
I recall BJL commented on this a while back. In practice, the read-noise of the superpixel is somewhat greater than the read-noise of a single small pixel, but of course not as great as the sum of the read-noise of all the individual pixels before they were binned.
[{POST_SNAPBACK}][/a]

Ray,

If BJL has some additional information then I would like to see it and his references. As far as the Roper Scientific article goes, I don't think that the physics of CCD has changed since then, and as you frequently point out technology is improving.  At any rate, binning can not be done with full advantage with respect to read noise after the fact so far as I know, since the read has already occurred. Of course, the full well is effectively increased.

[a href=\"http://www.microscopyu.com/tutorials/java/digitalimaging/signaltonoise/index.html]Nikon Java Calculator[/url] also gives an interactive calculator where the parameters under discussion can be shown in real time.

Bill
Logged

bjanes

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3387
larger sensors
« Reply #134 on: January 08, 2007, 11:15:27 am »

Quote
Sometimes I do and sometimes I don't, but my experience has been that most of your references do not address what I thought we were talking about, and if you think they do, then I don't think you understand them or what we were talking about (and on top of that, they may just be wrong).  A URL only shows what someone else thinks.

What we were talking about here is the fact that statistical noise is not *VISIBLE* noise.  I, and others said that more and smaller pixels in the same size sensor does not result in a noisier image, necessarily.  You seemed to disagreed with it, but never did anything to discredit the idea that had any relevance.

Not exactly, but I purposely used the word "similar", in reference to the reduction of visible noise.

Frankly,  I would think that any such system will work better in theory than practice.  Do you have an example of a sensor that does this and has truly lower read noise relative to signal than would be had with software binning?  Downsampling and software binning reduce read noise quite a bit, too.  I think it is generally better to leave the image in its higher-res state.  Besides having more visible detail, this allows future overlapped binning at the original resolution minus n-1 pixels in each dimension.  The binnable neighbors for a CFA camera will be very far from each other, as well.

What outrageous statements?  I remember saying that noise is relevant to image viewing not just in statistical intensity, but in frequency content relative to the image.  Then, you start throwing URLs at me that completely ignore the spatial aspects of noise, like Bugs Bunny throwing banana peels behind him while being pursued, or someone tossing smokebombs.  Now, you've gone on tangents like some poorly-thought-out use of the word "gain", as if that had any bearing on anything being discussed.  Now, hardware binning.  Let's get back to what we were talking about; the idea that microscopic noise is virtually lower-noise.  Can you find some way to disprove that?
[a href=\"index.php?act=findpost&pid=94506\"][{POST_SNAPBACK}][/a]
 
As to what we were talking about, only you seem to know.

So far, you are all bluster but no facts, only quotations from yourself. I don't find this discussion useful at this point.

Bill
Logged

eronald

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6642
    • My gallery on Instagram
larger sensors
« Reply #135 on: January 08, 2007, 11:43:20 am »

Quote
Edmund,

Who is doing the cherry picking and ignoring content here. Your post is not clear. BTW,
Roger has just updated his sensor analysis as of December 2006, and like an academic, he supplies references.

Bill
[a href=\"index.php?act=findpost&pid=94493\"][{POST_SNAPBACK}][/a]

I meant one should use Roger's paper to set the terminology, but that I am not quite so sanguine about the usefulness of the content to the present audience viz. its applicability to CMOS sensors such as those used by Canon and possibly soon to be used by Dalsa and Kodak.

Edmund
Logged
If you appreciate my blog posts help me by following on https://instagram.com/edmundronald

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
larger sensors
« Reply #136 on: January 08, 2007, 06:04:26 pm »

Quote
I meant one should use Roger's paper to set the terminology, but that I am not quite so sanguine about the usefulness of the content to the present audience viz. its applicability to CMOS sensors such as those used by Canon and possibly soon to be used by Dalsa and Kodak.

Edmund
[a href=\"index.php?act=findpost&pid=94536\"][{POST_SNAPBACK}][/a]

That's a good point. So many explanations on the net, on these matters, refer to CCD sensors. We make an erroneous assumption if we think everything applies equally to CMOS sensors. Just what principles are common to both types of sensors is not clear to me, but I wouldn't be surprised if the advantages of binning are greater with a CCD design than a CMOS design where each photodiode has its own, personal, analog preamplifier.
Logged

John Sheehy

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 838
larger sensors
« Reply #137 on: January 08, 2007, 06:11:07 pm »

Quote

As to what we were talking about, only you seem to know.

I said quite clearly what the issue was - that smaller pixels, with higher statistical noise, does not necessarily mean that the image itself is noisier, if there are more of them (they fill the same sensor space).  To this you objected, with links to Roger's page, where he fails to discuss the significance of a pixel in an entire image.  I am very interested in pixel statistics, but I also realize that the quality of pixels does not necessarily affect the image the same way.  Your link to Roger's page had nothing of value to contribute to the argument; *EVERYONE* here involved in this discussion is aware that smaller pixels have more shot noise, and usually more read noise, relative to signal, in current cameras.  That is not the issue.  The issue is what implications the greater per-pixel noise has for the entire image, which has smaller, but more numerous pixels.

Quote
So far, you are all bluster but no facts, only quotations from yourself. I don't find this discussion useful at this point.
[a href=\"index.php?act=findpost&pid=94530\"][{POST_SNAPBACK}][/a]

Be specific.  What did I say, and why do you find it unlikely.  IN YOUR OWN WORDS, not in Roger's.

To this point, you have still totally eluded discussion of the pixel's role in the image as a whole, concerning noise.
Logged

John Sheehy

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 838
larger sensors
« Reply #138 on: January 08, 2007, 06:14:56 pm »

Quote
That's a good point. So many explanations on the net, on these matters, refer to CCD sensors. We make an erroneous assumption if we think everything applies equally to CMOS sensors. Just what principles are common to both types of sensors is not clear to me, but I wouldn't be surprised if the advantages of binning are greater with a CCD design than a CMOS design where each photodiode has its own, personal, analog preamplifier.
[a href=\"index.php?act=findpost&pid=94638\"][{POST_SNAPBACK}][/a]

Not only that, but the camera adds more complications to the sensor.  We don't use sensors; we use cameras, which use sensors.  Any camera will have more read noise than occurs on-chip, unless the image is digitized right on the sensor chip; then total read noise could be part of the sensor spec.
Logged

bjanes

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 3387
larger sensors
« Reply #139 on: January 08, 2007, 06:39:32 pm »

Quote
I meant one should use Roger's paper to set the terminology, but that I am not quite so sanguine about the usefulness of the content to the present audience viz. its applicability to CMOS sensors such as those used by Canon and possibly soon to be used by Dalsa and Kodak.

Edmund
[{POST_SNAPBACK}][/a]

Kodak and Dalsa have been making CMOS for quite some time. In fact, DALSA founder and CEO Dr. Savvas Chamberlain was a pioneer in developing both technologies. The Dalsa web site has a good comparison of the two technologies and a long list of refereces.

[a href=\"http://vfm.dalsa.com/products/CCD_vs_CMOS.asp]CCD vs CMOS[/url]

You are correct that CMOS and CCD are not interchangeable. Roger uses Canon and most of his tests are on Canon CMOS sensors, so his references are applicable there. Roger also participates in NASA imaging projects and probably knows more about the subject than anyone participating in this thread, even though John criticizes his work and has apparently not published his own research. Roger makes no mention of binning, which is usually used with high end scientific CCD sensors and I do not even know if binning is available with CMOS

I am admittedly an amateur enthusiast and would appreciate any authoritative information on this subject that anyone can contribute, but I will listen most attentively to recognized authorities in the area.

Bill
Logged
Pages: 1 ... 5 6 [7] 8 9 10   Go Up