Why doesn't medium format have high ISOs?

[Tyrone Wellhung]

And, as variances are additive, the variance from pooling, say 4 pixels, with read noise A is 4A.

Correct.

Sorry, incorrect. The error is with the identification of read noise A with the variance. If the read noise is A, that is usually taken as the standard deviation, so the variance is A². The variance from pooling 4 pixels with read noise A is 4A², and the standard deviation 2A

Wrong,

Nope.

Yep

Still Poisson (mostly, apart from that bit which is quantisation noise). A is the standard deviation, the noise from pooling four pixels is 2A.

Also correct. Variance is the square of std dev.

Exactly so, and the standard deviation gives the amplitude of the noise.

 

[Tyrone Wellhung]

Well, we get another lecture in sensor theology here.

Simply science and engineering, not theology.

 Spouts the high priest of read noise, shot noise, thermal noise, quantization noise, pixels, quantum efficiency and equivalence. As I wrote earlier . . .

Are not 'read noise, shot noise, thermal noise, quantization noise, pixels, quantum efficiency' all to do with science and engineering?

Such an innocent appearing, very short question. What HorsePix doesn't realize that it's a trap, and when sprung he'll be inundated with reams and reams of esoteric verbiage that goes round and round and never leads anywhere that's even remotely useful.

 . . . and as is apparent, the trap has been sprung. Tyrone Wellhung, aka Shull Bitter, aka . . . is back in his element and yet another thread suffers.

In what sense does it suffer. The issues of 'read noise, shot noise, thermal noise, quantization noise, pixels, quantum efficiency' are all germane to the OP, and it's hardly as if the thread is pushing up against 150.

By that reasoning, with read noise at, say 1/8^=2^-3 electron, we only need . . .

  . . . to check off the IGNORE USER box.

Your privilege. At least if you ignore me, you'll stop clogging up the threads with these incessant attacks.

 

[SNRatio]

And the point is, that high ISO performance IS relevant for many applications, so any changes affecting that performance may be problematic. The "smaller is better" theology is also contradicted by the performance of the D3s sensor.

An outlier, entirely predicted by its exceptional quantum efficiency, and you have no evidence that has anything to do with pixel size.

I really don't understand why you can't just stop. "Outlier", "quantum efficiency": There is of course nothing that stops Nikon (and others, if they don't run into patent issues) from using that same technology on slightly different scales, it's highly improbable that it will not be matched or surpassed very soon. Then we will still have the pixel size dependency, and the only thing to help the higher MP sensors, will be downsampling. Which, so far, to my knowledge has not been demonstrated to give the desired results except for resolution.

I think the basic flaw in your reasoning is that you postulate that read noise is necessarily a function of pixel size. At newer sensors and higher ISOs, there seems to be no strong relationship, and unless the effective read noise (as std dev) can scale with the pixel pitch, sampling has to cost something. Also, statistical effects from quantization may tend to show up other places than in the sensels as very low electron counts are handled.

AFAIK, the only general empirical support so far for "smaller is better", is the traditional domination of read noise from the electronics components around base ISO, but as the D3X shows, this is not something inherent in "large pixel" designs. It's also, so far, of limited practical relevance as we go beyond the base ISO DR of the D3X. If, however, there should be a general demand for 16 stops DR at base ISO, it might , with current technology, be most efficient to implement this as sampling from smaller pixels, the read noise from processing circuitry totally dominating the pixel read noise, making the slightly increased combined pixel read noise a moot point. But because of this being more of an electronics problem, I don't think anybody can be sure it can only be solved by downsampling.

For Nikon, this is a really old idea, I think they even used it in the D1H - have heard about 4x downsampling from a 10.8MP sensor in that model. And if the D4 comes with 16MP, a 4MP downsamling may render 100K+ ISOs workable.

So, the idea of downsampling is IMHO important and far from fully exploited, but please don't keep pretending it is without costs. And IMHO, Heisenberg's uncertainty principle is kind of hint that we better deal with aggregates if we want to make several observations simultaneously when electron counts are in the tens.

And I really hope you stop attacking people that may somewhat lack in insight, like me, but still try to produce sound reasoning, well founded in observations and backed by statistical theory. If you can't do better than mixing up variance and standard dev, I think you should rather stop discussing here.

 

[ejmartin]

--
emil
--



http://theory.uchicago.edu/~ejm/pix/20d/

 

[Billx08]

By that reasoning, with read noise at, say 1/8^=2^-3 electron, we only need . . .

  . . . to check off the IGNORE USER box.

Your privilege. At least if you ignore me, you'll stop clogging up the threads with these incessant attacks.

Accurately describing who you are and how threads evolve once you join them is an attack? Anyone reading your replies in this thread can easily see the turn that they've taken, just as was predicted. But why would I want to check that box and miss the incessant posturing, drivel and ummm, noise o' well hung one? Ah, I see. If those that know what you're about ignore you, it will be all the easier to ensnare and trap others in your sticky threads. I appreciate reading well informed replies that try to actually help people, not the replies that are used more to be a soap box platform for well rehearsed harangues that never seem to change.

At least your offensive sock puppet monikers (I guess that they amuse you , which adds something to judge you by and is useful in spotlighting who you are, the guy that so enjoys Shull Bitting, and who seems so Bitter). There are others in these forums that know as much and more about your favorite topics but make their points in a reasonable manner without belittling and insulting those that they debate/argue with as you so often do, and move on instead of pushing threads to the 150 reply limit as tends to happen to threads you participate in, rarely resolving anything or shedding any new light.

 

[SNRatio]

And, as variances are additive, the variance from pooling, say 4 pixels, with read noise A is 4A.

Correct.

Sorry, incorrect. The error is with the identification of read noise A with the variance. If the read noise is A, that is usually taken as the standard deviation, so the variance is A². The variance from pooling 4 pixels with read noise A is 4A², and the standard deviation 2A

Wrong,

Nope.

Yep

Still Poisson (mostly, apart from that bit which is quantisation noise). A is the standard deviation, the noise from pooling four pixels is 2A.

Also correct. Variance is the square of std dev.

Exactly so, and the standard deviation gives the amplitude of the noise.

It's just me using the word "noise" in my own way, wrong here - of course I meant the variance, as is clear from writing "additive", I was thinking in variances all the time:

And, as variances are additive, the variance from pooling, say 4 pixels, with read noise A is 4A. Unless the read noise of each pixel is less that 1/4 of that of a 4x bigger pixel, the total S/N-ratio will become smaller by pooling this way, resulting in less dynamic range recoverable.'

This is of course variances, otherwise the addition and the whole argument would be meaningless. Taking square roots, we get the "correct" noise, twice the noise of the single pixel. Which leads to the requirement that the read noise must scale linearly with pixel pitch, which is the important thing in the argument.

And, for example, the read noise plot of Roger Clark, http://www.clarkvision.com/articles/digital.sensor.performance.summary/index.html , indicates to me that it may rather be the other way round when we compare same generation sensors. For what I think is a relevant example, Clark lists the 7D as having much higher read noise than 1DIV, in spite of a much smaller pixel pitch.

 

[Tyrone Wellhung]

And the point is, that high ISO performance IS relevant for many applications, so any changes affecting that performance may be problematic. The "smaller is better" theology is also contradicted by the performance of the D3s sensor.

An outlier, entirely predicted by its exceptional quantum efficiency, and you have no evidence that has anything to do with pixel size.

I really don't understand why you can't just stop. "Outlier", "quantum efficiency": There is of course nothing that stops Nikon (and others, if they don't run into patent issues) from using that same technology on slightly different scales, it's highly improbable that it will not be matched or surpassed very soon.

Well actually we don't know that, but it's a reasonable assumption. The QE of the D3S is 53% by Chrisk99's analysis ( http://forums.dpreview.com/forums/read.asp?forum=1019&message=34581894 ) . But yes, it is a reasonable assumption that smaller pixel sensors can achieve similar QE. Like for instance the G11 sensor, which has a higher QE even than the D3s, with much smaller pixels.

Then we will still have the pixel size dependency, and the only thing to help the higher MP sensors, will be downsampling. Which, so far, to my knowledge has not been demonstrated to give the desired results except for resolution.

Fairly frequently. One good place to look is the debate following Phil Askey's 'downsampling doesn'r reduce noise' blog, which included many good examples. Emil has posted several too, as I remember.

I think the basic flaw in your reasoning is that you postulate that read noise is necessarily a function of pixel size.

I haven't said that, what I've said is that read noise reduces with the area of the pixel under strict scaling. Mind you, the trends seem to verify that. As Canon has reduced geometry it has reduced read noise, even though Chipworks asserts that there has been no substantive technology change. Do you have some evidence to the contrary, or can you dispute the line of reasoning that says so? I should add, I'm pretty confident about this, since its been confirmed by Eric Fossum.

At newer sensors and higher ISOs, there seems to be no strong relationship, and unless the effective read noise (as std dev) can scale with the pixel pitch, sampling has to cost something.

Now you've lost me again. Why should sampling a small pixel cost any more than a large one? In fact the reverse is true, a constant sampling voltage noise will be translated to a lower input referred noise as pixel size reduces.

Also, statistical effects from quantization may tend to show up other places than in the sensels as very low electron counts are handled.

Care to suggest where and how.

AFAIK, the only general empirical support so far for "smaller is better", is the traditional domination of read noise from the electronics components around base ISO, but as the D3X shows, this is not something inherent in "large pixel" designs.

Low ISO read noise has generally almost nothing to do with the sensor. It is a product of the off chip read chain. The Sony sensor in the D3x has per-column ADC's which have a much lower read noise contribution than the traditional off-chip methods. It's pixel level read noise is nothing special.

It's also, so far, of limited practical relevance as we go beyond the base ISO DR of the D3X. If, however, there should be a general demand for 16 stops DR at base ISO, it might , with current technology, be most efficient to implement this as sampling from smaller pixels, the read noise from processing circuitry totally dominating the pixel read noise, making the slightly increased combined pixel read noise a moot point. But because of this being more of an electronics problem, I don't think anybody can be sure it can only be solved by downsampling.

I'm not sure why you're into downsampling again.

For Nikon, this is a really old idea, I think they even used it in the D1H - have heard about 4x downsampling from a 10.8MP sensor in that model. And if the D4 comes with 16MP, a 4MP downsamling may render 100K+ ISOs workable.

The D1 models were a particular case. They all shared the same 10.8MP sensor silicon, with the pixels combined differently to give the different resolutions. The D1x was resampled in one direction to provide a non-native resolutionas standard. But essentially the approach wasn't resampling, it was pixel combining, much more akin to binning.

So, the idea of downsampling is IMHO important and far from fully exploited, but please don't keep pretending it is without costs.

What are those costs? In any case, almost every image is resampled before viewing, the only question is the resampling ratio. So there are no additional costs, even if you could identify costs in the first place.

And IMHO, Heisenberg's uncertainty principle is kind of hint that we better deal with aggregates if we want to make several observations simultaneously when electron counts are in the tens.

That is the whole point of the discussion is it not?

And I really hope you stop attacking people that may somewhat lack in insight, like me, but still try to produce sound reasoning, well founded in observations and backed by statistical theory. If you can't do better than mixing up variance and standard dev, I think you should rather stop discussing here.

i) Where have I 'attacked' you? I have simply disputed some of your facts and reasoning - that is a discussion, not an attack. And I have yet to see your observations, and your statistical theory was somewhat flawed, as I showed.

ii) I haven't mixed up variance and standard dev, it was you that did that, by assuming noise amplitude is given by the variance, whereas it is given by the standard deviation.

 

[Tyrone Wellhung]

in your quest to ignore me? The urge to snap at my heels just too big for you?

By that reasoning, with read noise at, say 1/8^=2^-3 electron, we only need . . .

  . . . to check off the IGNORE USER box.

Your privilege. At least if you ignore me, you'll stop clogging up the threads with these incessant attacks.

 Accurately describing who you are and how threads evolve once you join them is an attack? Anyone reading your replies in this thread can easily see the turn that they've taken, just as was predicted. But why would I want to check that box and miss the incessant posturing, drivel and ummm, noise o' well hung one? Ah, I see. If those that know what you're about ignore you, it will be all the easier to ensnare and trap others in your sticky threads. I appreciate reading well informed replies that try to actually help people, not the replies that are used more to be a soap box platform for well rehearsed harangues that never seem to change.

There is a consistency, but that's because the science doesn't change, and people keep on repeating the same misunderstandings and myths. So long as they do, I'll pop up with the truth. Sorry it annoys you, but interestingly the truth does seem to annoy some.

I would note, looking at your profile, I'm not the only one who seems to be on your harass list. You seem to be an eminently irritable person. Perhaps you even enjoy being irritated. Well, being harassed by you bothers me not one little bit, so if it ticks your box, be my guest.

 At least your offensive sock puppet monikers (I guess that they amuse you , which adds something to judge you by and is useful in spotlighting who you are, the guy that so enjoys Shull Bitting, and who seems so Bitter). There are others in these forums that know as much and more about your favorite topics but make their points in a reasonable manner without belittling and insulting those that they debate/argue with as you so often do, and move on instead of pushing threads to the 150 reply limit as tends to happen to threads you participate in, rarely resolving anything or shedding any new light.

Well, if my ID has irritated you, then it's served some purpose, maybe even made you happier.

 

[Tyrone Wellhung]

And, as variances are additive, the variance from pooling, say 4 pixels, with read noise A is 4A.

Correct.

Sorry, incorrect. The error is with the identification of read noise A with the variance. If the read noise is A, that is usually taken as the standard deviation, so the variance is A². The variance from pooling 4 pixels with read noise A is 4A², and the standard deviation 2A

Wrong,

Nope.

Yep

Still Poisson (mostly, apart from that bit which is quantisation noise). A is the standard deviation, the noise from pooling four pixels is 2A.

Also correct. Variance is the square of std dev.

Exactly so, and the standard deviation gives the amplitude of the noise.

It's just me using the word "noise" in my own way, wrong here

using terminology differently from everyone else does cause difficulties in communication.

of course I meant the variance, as is clear from writing "additive", I was thinking in variances all the time:

And what is the significance, in photographic terms, of the variance?

And, as variances are additive, the variance from pooling, say 4 pixels, with read noise A is 4A. Unless the read noise of each pixel is less that 1/4 of that of a 4x bigger pixel, the total S/N-ratio will become smaller by pooling this way, resulting in less dynamic range recoverable.'

This is of course variances, otherwise the addition and the whole argument would be meaningless. Taking square roots, we get the "correct" noise, twice the noise of the single pixel. Which leads to the requirement that the read noise must scale linearly with pixel pitch, which is the important thing in the argument.

It scales better than that, it scales with the area assuming strict scaling. This is because the charge is related to the voltage by the SF gate/floating diffusion capacitance, which is proportional to the area of those components.

And, for example, the read noise plot of Roger Clark, http://www.clarkvision.com/articles/digital.sensor.performance.summary/index.html , indicates to me that it may rather be the other way round when we compare same generation sensors. For what I think is a relevant example, Clark lists the 7D as having much higher read noise than 1DIV, in spite of a much smaller pixel pitch.

That is one comparison, ignoring the weight of the trend which indicates a steady reduction in read noise as pixels have shrunk. Clark's figures for the 1DIV and 7D also do not triangulate well with other sources. Chrisk99's DxO derived figures show a different pattern:



For everything apart form those two cameras these figures do match Clarks. However they show these two cameras to be very similar and much of a muchness with other contemporary Canon cameras, about 2.5 e-, give or take a little. This is consistent with Canon designing their pixels around standard subcomponents, which are updated between generations, rather than simply applying a uniform scale to a standard pixel design. Note though the general trend towards lower read noise as pixlel size shrinks. I found out the correlation between Chrisk99's read noise figures for Canon sensors and pixel pitch, it gives 0.56. Not so strong, but not negative, either

 

[Billx08]

. . .

I would note, looking at your profile, I'm not the only one who seems to be on your harass list. You seem to be an eminently irritable person. Perhaps you even enjoy being irritated. Well, being harassed by you bothers me not one little bit, so if it ticks your box, be my guest.

Thank you for the permission. It's not easy to suffer fools nor the obnoxious gladly. Harassing? No. Certainly not in the recent Nightwings threads, where practically everyone in one thread (another wasted 150 parter) tried to help or correct him until it should have been apparent to everyone that they were responding to a game playing troll. Your reply gives lie to mine not bothering you even "one little bit", but if what you say is true, it's rather odd that you'd feel the need to reply. Thanks for caring, or is it a reply driven by anger at being exposed so quickly? This new sock puppet ID is barely more than a month old. Were all of the others banned? I see (as I also predicted) the condescension and insults in replies to others. Yes, some of us know you very well, even when you resort to hiding behind new DPR IDs, IDs that are all too obvious.

At least your offensive sock puppet monikers (I guess that they amuse you , which adds something to judge you by and is useful in spotlighting who you are, the guy that so enjoys Shull Bitting, and who seems so Bitter). There are others in these forums that know as much and more about your favorite topics but make their points in a reasonable manner without belittling and insulting those that they debate/argue with as you so often do, and move on instead of pushing threads to the 150 reply limit as tends to happen to threads you participate in, rarely resolving anything or shedding any new light.

Well, if my ID has irritated you, then it's served some purpose, maybe even made you happier.

The only purpose that you intended when you created yet another offensive sock puppet ID, the sort of thing that's usually associated with young teenage trolls, fanboys and malcontents. The sorts of names that would rarely be used in real life if they could be easily traced back to the person behind them. The sorts of names that show a lack of respect or contempt for those you interact with.

 

[Tyrone Wellhung]

. . .

I would note, looking at your profile, I'm not the only one who seems to be on your harass list. You seem to be an eminently irritable person. Perhaps you even enjoy being irritated. Well, being harassed by you bothers me not one little bit, so if it ticks your box, be my guest.

 Thank you for the permission. It's not easy to suffer fools nor the obnoxious gladly.

I don't know, it just takes a little patience. See how well I'm suffering you.

Harassing? No. Certainly not in the recent Nightwings threads, where practically everyone in one thread (another wasted 150 parter) tried to help or correct him until it should have been apparent to everyone that they were responding to a game playing troll. Your reply gives lie to mine not bothering you even "one little bit", but if what you say is true, it's rather odd that you'd feel the need to reply.

Just a little compassion, keeping you company. That seems to be what you want.

Thanks for caring, or is it a reply driven by anger at being exposed so quickly?

I'm used to being exposed. Comes with the territory. Call it a professional risk.

This new sock puppet ID is barely more than a month old. Were all of the others banned? I see (as I also predicted) the condescension and insults in replies to others. Yes, some of us know you very well, even when you resort to hiding behind new DPR IDs, IDs that are all too obvious.

I'm not hiding. Being obscured is not the same thing as hiding at all.

At least your offensive sock puppet monikers (I guess that they amuse you , which adds something to judge you by and is useful in spotlighting who you are, the guy that so enjoys Shull Bitting, and who seems so Bitter). There are others in these forums that know as much and more about your favorite topics but make their points in a reasonable manner without belittling and insulting those that they debate/argue with as you so often do, and move on instead of pushing threads to the 150 reply limit as tends to happen to threads you participate in, rarely resolving anything or shedding any new light.

Well, if my ID has irritated you, then it's served some purpose, maybe even made you happier.

 The only purpose that you intended when you created yet another offensive sock puppet ID, the sort of thing that's usually associated with young teenage trolls, fanboys and malcontents. The sorts of names that would rarely be used in real life if they could be easily traced back to the person behind them. The sorts of names that show a lack of respect or contempt for those you interact with.

Yes, you are entirely right. The satisfaction in seeing the self-righteous indignation of the fools and hypocrites who are upset by such things is very puerile indeed. But still satisfying.

 

[Joseph S Wisniewski]

The only purpose that you intended when you created yet another offensive sock puppet ID,

Technically, a "sock puppet" refers to a second ID someone uses to either support their own position, or to engage in fake attacks and arguments, so that the main account has an excuse to argue.

I've never known Bob to do this. He creates accounts serially, and doesn't support his own arguments with multiple voices.

If anything, he's the "victim" of socks. There are plenty of people who create multiple accounts purely for the pleasure of being able to "stuff the ballot box" in dpReview challenges or hit the complaint button so many times on people that they don't like that they get people banned.

Actually, the only way he would be exhibiting sock puppet behavior in this thread is if he were also you.

the sort of thing that's usually associated with young teenage trolls, fanboys and malcontents. The sorts of names that would rarely be used in real life if they could be easily traced back to the person behind them.

Actually, a lot of people know who he really is, and we don't really mind.

The sorts of names that show a lack of respect or contempt for those you interact with.

Or a sense of humor. Sometimes technical types have those. Sometimes, a bit strange.

--
Rahon Klavanian 1912-2008.

Armenian genocide survivor, amazing cook, scrabble master, and loving grandmother. You will be missed.

Ciao! Joseph

http://www.swissarmyfork.com

 

[Billx08]

The only purpose that you intended when you created yet another offensive sock puppet ID,

Technically, a "sock puppet" refers to a second ID someone uses to either support their own position, or to engage in fake attacks and arguments, so that the main account has an excuse to argue.

Your definition is valid but a bit limited. It's also described this way :

A sockpuppet is an online identity used for purposes of deception within an online community. In its earliest usage, a sockpuppet was a false identity through which a member of an Internet community speaks with or about himself or herself, pretending to be a different person,[1] like a ventriloquist manipulating a hand puppet.

http://en.wikipedia.org/wiki/Sock_puppet_%28internet%29

and

sock puppet (noun): a name or identity used online to deceive others and that is often used to direct praise or attention to oneself

http://www.britannica.com/blogs/2008/01/masstige-nanite-etc-the-open-dictionary/

I've never known Bob to do this. He creates accounts serially, and doesn't support his own arguments with multiple voices.

Who is Bob? Has Tyrone Wellhung linked himself to "Bob" or are you basing that on some other information gleaned from your troll studies? The only Bob that I recall ever getting involved with any of these related sock puppets is Bob Elkind, who along with Eliah Borg also commented on the juvenile names and Bob recognized what these puppies (whether sock or not) are all about, making the same points that I have :

I repeat, you seem to be trying to make points about... making points. Arguing for the sake of arguing. You're new to dpreview forums, and yet you've written 335 posts in the last 3 weeks.

You must have a lot to say. Your login name is classy (not!).

http://forums.dpreview.com/forums/read.asp?forum=1021&message=34723970


[Iliah Borg replying to Bob Elkind]

Warning: only 7 more posts to hit the 150 per thread limit.

Then we'll be looking forward to a new thread?

Out of curiosity, have you tried switching "B" and "Sh" in the nickname of this "anonymous" graduate?

http://forums.dpreview.com/forums/read.asp?forum=1021&message=34751336


Getting back to your S.P. definition of a "second ID someone uses to either support their own position", that's what it looked like when Crocodile Gena and Shull Bitter did their tag-team thing, arguing about their pet topics (including "equivalence"), sounding quite a bit like Tyrone Wellhung and displaying nearly identical personalities. Croc was the first of these puppies to appear, and this occurred not long after Joe Mama (Joseph James) was permanently banned, whose website URLs were often shown at the bottom of their replies. Curious coincidence, that Croc and Bitter (and now Wellhung) share Joe Mama's passion and argue it just as doggedly.

If anything, he's the "victim" of socks.

Who is this "he"? Wellhung? The unidentified Bob? And who would the socks be. Links could be useful.

There are plenty of people who create multiple accounts purely for the pleasure of being able to "stuff the ballot box" in dpReview challenges or hit the complaint button so many times on people that they don't like that they get people banned.

None of which is occurring here that I'm aware of, and I haven't hit the "Complain" button.

Actually, the only way he would be exhibiting sock puppet behavior in this thread is if he were also you.

That might be one way but surely not the only way and I recognize the behavior even if you don't see it . . . yet.

the sort of thing that's usually associated with young teenage trolls, fanboys and malcontents. The sorts of names that would rarely be used in real life if they could be easily traced back to the person behind them.

Actually, a lot of people know who he really is, and we don't really mind.

Why be so inscrutable? Are you referring to the person behind Mr. Wellhung? If so, might it be the same one that was behind Joe Mama? If so, you posted in the thread that discussed Joe Mama's banning and seem to have felt differently at the time :

He violated some rules and got banned. No big deal. Anyone can come
back with a new neame.

Would you like it if it happened to you, and you could never be known here as "Glen Barrington" again?

That's the great unfairness of how this shakes out. Trolls like "natureman" don't use their real names (maybe they're embarrassed by who they really are). They know they have been banned before and will be banned again (and again and again) and they don't care. They come back under new names all the time, because people hated the last person they posed as (and the one before that, etc) so why not let the last ID die...

http://forums.dpreview.com/forums/read.asp?forum=1018&message=28597773

The sorts of names that show a lack of respect or contempt for those you interact with.

Or a sense of humor. Sometimes technical types have those. Sometimes, a bit strange.

Sometimes. I have a pretty good sense of humor as do you, but it's never been in evidence with these puppies, except the forced, false humor frequently attempted by those that use and abuse LOL and ROFL.

 

[Joseph S Wisniewski]

And the point is, that high ISO performance IS relevant for many applications, so any changes affecting that performance may be problematic. The "smaller is better" theology is also contradicted by the performance of the D3s sensor.

An outlier, entirely predicted by its exceptional quantum efficiency, and you have no evidence that has anything to do with pixel size.

I really don't understand why you can't just stop. "Outlier", "quantum efficiency": There is of course nothing that stops Nikon (and others, if they don't run into patent issues) from using that same technology on slightly different scales, it's highly improbable that it will not be matched or surpassed very soon.

Well actually we don't know that, but it's a reasonable assumption. The QE of the D3S is 53% by Chrisk99's analysis ( http://forums.dpreview.com/forums/read.asp?forum=1019&message=34581894 ) . But yes, it is a reasonable assumption that smaller pixel sensors can achieve similar QE. Like for instance the G11 sensor, which has a higher QE even than the D3s, with much smaller pixels.

Then we will still have the pixel size dependency, and the only thing to help the higher MP sensors, will be downsampling. Which, so far, to my knowledge has not been demonstrated to give the desired results except for resolution.

Fairly frequently. One good place to look is the debate following Phil Askey's 'downsampling doesn'r reduce noise' blog, which included many good examples. Emil has posted several too, as I remember.

Personally, I've always been of the view that downsampling is great, when the deliverable is low res (movies, web, digital picture frame), but for higher res use, it's slightly harmful. As long as you have the storage, processing, and printer bandwidth to handle it, the best image is native resolution upsampled to printer resolution after all the processing is done.

Of course, this requires a lot of side-by-side comparisons with the examples printed by a really good print-maker.

But in the end, it's near a wash.

For Nikon, this is a really old idea, I think they even used it in the D1H - have heard about 4x downsampling from a 10.8MP sensor in that model. And if the D4 comes with 16MP, a 4MP downsamling may render 100K+ ISOs workable.

The D1 models were a particular case. They all shared the same 10.8MP sensor silicon, with the pixels combined differently to give the different resolutions. The D1x was resampled in one direction to provide a non-native resolutionas standard. But essentially the approach wasn't resampling, it was pixel combining, much more akin to binning.

Exactly. Hardware binning, with a filter pattern screened onto the sensor as appropriate for the binning pattern. Same underlying 10.7mp sensor in both D1H and D1X. Look at it under the scope, and D1H actually looked like

R R G G R R G G
R R G G R R G G
G G B B G G B B
G G B B G G B B
R R G G R R G G
R R G G R R G G
G G B B G G B B
G G B B G G B B

And D1X like this

R G R G R G R G
R G R G R G R G
G B G B G B G B
G B G B G B G B
R G R G R G R G
R G R G R G R G
G B G B G B G B
G B G B G B G B

As far as anyone can tell, it was done purely to allow one sensor, with just changes to the interconnections (not sure if that was on-chip at the pixel pair or quad level, or off-chip. You can bin a CCD in one direction by getting clever with the clocking, the other direction usually needs at least a pair of amplifiers) and a change in the organic part of the fab (the screening of the color filters), to serve two different cameras.

Personally, I've always been of the opinion that Nikon wanted D1X to be a 10.6mp camera, but couldn't cram enough processing power into the box to deal with it, so they applied half the D1H "solution" and ended up with 5.3mp in a mathematically disastrous 2:1 binning.

Now, the 2:2 binning of the D1H did have some advantages. It cut the effective gutter between pixels about in half, which made the sensor output a lot more like a "boxcar" filter, instead of a comb filter with a low width.

Downsampling requires that the Bayer pattern be properly demultiplexed first, and wasn't happening in realtime in 2001.

--
Rahon Klavanian 1912-2008.

Armenian genocide survivor, amazing cook, scrabble master, and loving grandmother. You will be missed.

Ciao! Joseph

http://www.swissarmyfork.com

 

[Tyrone Wellhung]

And the point is, that high ISO performance IS relevant for many applications, so any changes affecting that performance may be problematic. The "smaller is better" theology is also contradicted by the performance of the D3s sensor.

An outlier, entirely predicted by its exceptional quantum efficiency, and you have no evidence that has anything to do with pixel size.

I really don't understand why you can't just stop. "Outlier", "quantum efficiency": There is of course nothing that stops Nikon (and others, if they don't run into patent issues) from using that same technology on slightly different scales, it's highly improbable that it will not be matched or surpassed very soon.

Well actually we don't know that, but it's a reasonable assumption. The QE of the D3S is 53% by Chrisk99's analysis ( http://forums.dpreview.com/forums/read.asp?forum=1019&message=34581894 ) . But yes, it is a reasonable assumption that smaller pixel sensors can achieve similar QE. Like for instance the G11 sensor, which has a higher QE even than the D3s, with much smaller pixels.

Then we will still have the pixel size dependency, and the only thing to help the higher MP sensors, will be downsampling. Which, so far, to my knowledge has not been demonstrated to give the desired results except for resolution.

Fairly frequently. One good place to look is the debate following Phil Askey's 'downsampling doesn'r reduce noise' blog, which included many good examples. Emil has posted several too, as I remember.

Personally, I've always been of the view that downsampling is great, when the deliverable is low res (movies, web, digital picture frame), but for higher res use, it's slightly harmful. As long as you have the storage, processing, and printer bandwidth to handle it, the best image is native resolution upsampled to printer resolution after all the processing is done.

Agreed. I can't see why you'd ever downsample if the output resolution was greater than the camera resolution. Interpolation algorithms tend to produce artifacts which look remarkably like the high spatial frequency noise you'd get from the native sensor resolution anyway, so all you'd be doing is removing the high frequency noise (and the detail) and the generating something that looks very much like it (without the detail). Very rarely are you outputting at native resolution, so if you're not upsampling you're downsampling - and then it pays to keep it under your control. Printer and screen drivers often have absurdly poor downsampling methods, for screen drivers simply pixel decimation. I think this is partially responsible for the noise reputation of high pixel density sensors - displayed using such a crass screen driver, they often do look noisy.

Of course, this requires a lot of side-by-side comparisons with the examples printed by a really good print-maker.

But in the end, it's near a wash.

For Nikon, this is a really old idea, I think they even used it in the D1H - have heard about 4x downsampling from a 10.8MP sensor in that model. And if the D4 comes with 16MP, a 4MP downsamling may render 100K+ ISOs workable.

The D1 models were a particular case. They all shared the same 10.8MP sensor silicon, with the pixels combined differently to give the different resolutions. The D1x was resampled in one direction to provide a non-native resolutionas standard. But essentially the approach wasn't resampling, it was pixel combining, much more akin to binning.

Exactly. Hardware binning, with a filter pattern screened onto the sensor as appropriate for the binning pattern. Same underlying 10.7mp sensor in both D1H and D1X. Look at it under the scope, and D1H actually looked like

R R G G R R G G
R R G G R R G G
G G B B G G B B
G G B B G G B B
R R G G R R G G
R R G G R R G G
G G B B G G B B
G G B B G G B B

And D1X like this

R G R G R G R G
R G R G R G R G
G B G B G B G B
G B G B G B G B
R G R G R G R G
R G R G R G R G
G B G B G B G B
G B G B G B G B

As far as anyone can tell, it was done purely to allow one sensor, with just changes to the interconnections (not sure if that was on-chip at the pixel pair or quad level, or off-chip. You can bin a CCD in one direction by getting clever with the clocking, the other direction usually needs at least a pair of amplifiers) and a change in the organic part of the fab (the screening of the color filters), to serve two different cameras.

Personally, I've always been of the opinion that Nikon wanted D1X to be a 10.6mp camera, but couldn't cram enough processing power into the box to deal with it, so they applied half the D1H "solution" and ended up with 5.3mp in a mathematically disastrous 2:1 binning.

Now, the 2:2 binning of the D1H did have some advantages. It cut the effective gutter between pixels about in half, which made the sensor output a lot more like a "boxcar" filter, instead of a comb filter with a low width.

Downsampling requires that the Bayer pattern be properly demultiplexed first, and wasn't happening in realtime in 2001.

Thanks for that - but I'll cut the conversation short now, lest you be identified as another of my sock puppets

 

[Billx08]

Downsampling requires that the Bayer pattern be properly demultiplexed first, and wasn't happening in realtime in 2001.

Thanks for that - but I'll cut the conversation short now, lest you be identified as another of my sock puppets

Hardly. Joseph writes clearly and to the point, without the need to continue writing the same thing over and over until the thread maxes out. He has also been here many years and I don't think that anyone would deny that there's no one else like him! You on the other hand have been here only a couple of moons and if your DPR ID was changed to Croc's or Shull's, nobody would find anything amiss, or as Eric Burdon once sang/emoted "It's all meat from the same bone", Mr. Wellhung.

 

[Joseph S Wisniewski]

The only purpose that you intended when you created yet another offensive sock puppet ID,

Technically, a "sock puppet" refers to a second ID someone uses to either support their own position, or to engage in fake attacks and arguments, so that the main account has an excuse to argue.

Your definition is valid but a bit limited. It's also described this way :

A sockpuppet is an online identity used for purposes of deception within an online community. In its earliest usage, a sockpuppet was a false identity through which a member of an Internet community speaks with or about himself or herself, pretending to be a different person,[1] like a ventriloquist manipulating a hand puppet.

I don't see that as conflicting with my definition, at all. The salient point of the sock puppet is "speaks with or about himself".

I've never seen Bob do that, and I know at least 17 of his previous accounts. Always one after the other. Never interacting with himself, supporting himself. And never, after the first account, talking "about himself".

In fact, his first account had exactly the same form as yours,

[first name][initial][number]

http://en.wikipedia.org/wiki/Sock_puppet_%28internet%29

and

sock puppet (noun): a name or identity used online to deceive others and that is often used to direct praise or attention to oneself

http://www.britannica.com/blogs/2008/01/masstige-nanite-etc-the-open-dictionary/

He doesn't "direct praise or attention to oneself". He never mentions his previous accounts, especially his first one, the one that had all the identifying information.

I've never known Bob to do this. He creates accounts serially, and doesn't support his own arguments with multiple voices.

Who is Bob? Has Tyrone Wellhung linked himself to "Bob" or are you basing that on some other information gleaned from your troll studies?

Both. He's open enough to admit it, when asked directly. Like I said, I don't really consider him a "troll". He's more a "troll victim", ambushed by packs of trolls and banned because dpReview places a high emphasis on people who get a lot of complaints, and real trolls do run fleets of sock puppets (one of the dpReview mods once commented that the "sock puppet army" that was "stuffing the ballot box" on the challenges needed to be cleaned up several hundred at a time. That's one troll, with a script, running hundreds of socks).

The only Bob that I recall ever getting involved with any of these related sock puppets is Bob Elkind,

Nope, entirely different Bob. But he is a real engineer, and not hard to find.

His first account was actually more open and transparent than yours, he had his real last name in the email address he used.

who along with Eliah Borg also commented on the juvenile names

Iliah Borg. And one can comment on the names without disrespecting the arguments or POV.

I agree 100% with Joseph Jame's arguments, but ended up creating some enmity between us by criticizing his presentation of those arguments.

and Bob recognized what these puppies (whether sock or not) are all about, making the same points that I have :

I repeat, you seem to be trying to make points about... making points. Arguing for the sake of arguing. You're new to dpreview forums, and yet you've written 335 posts in the last 3 weeks.

You must have a lot to say. Your login name is classy (not!).

http://forums.dpreview.com/forums/read.asp?forum=1021&message=34723970

[Iliah Borg replying to Bob Elkind]

Warning: only 7 more posts to hit the 150 per thread limit.

Then we'll be looking forward to a new thread?

Out of curiosity, have you tried switching "B" and "Sh" in the nickname of this "anonymous" graduate?

http://forums.dpreview.com/forums/read.asp?forum=1021&message=34751336

In other words, he "got the joke".

(to be continued)

--
Rahon Klavanian 1912-2008.

Armenian genocide survivor, amazing cook, scrabble master, and loving grandmother. You will be missed.

Ciao! Joseph

http://www.swissarmyfork.com

 

[Joseph S Wisniewski]

Due to the 6,000 character limit, this post, like a sock, is part of a pair...

Billx08 wrote:

Getting back to your S.P. definition of a "second ID someone uses to either support their own position", that's what it looked like when Crocodile Gena and Shull Bitter did their tag-team thing,

They are definitely two different people.

arguing about their pet topics (including "equivalence"), sounding quite a bit like Tyrone Wellhung and displaying nearly identical personalities.

Interesting. It was the glaring personality differences that make them each so easy to identify.

Croc was the first of these puppies

Once you add a new, insulting nickname, I'm quite unlikely to make an effort to follow your arguments.

to appear, and this occurred not long after Joe Mama (Joseph James) was permanently banned, whose website URLs were often shown at the bottom of their replies.

Joseph James also maintains a list of his dpReview aliases on his website.

Curious coincidence, that Croc and Bitter (and now Wellhung) share Joe Mama's passion and argue it just as doggedly.

Not at all.

This is going to come as a complete shock to you, but Joseph James is correct in nearly everything he says. I was teaching what he calls "equivalence" 15 years ago, with examples from 35mm, 6x7, and 4x5. Including the use of different film speeds on the different formats to demonstrate the same concept as we have with different sensor ISO ratings.

If anything, he's the "victim" of socks.

Who is this "he"? Wellhung?

Yes.

The unidentified Bob?

Again, yes.

As far as I'm concerned, he's much more "identified" than you are, Mr. X.

And who would the socks be. Links could be useful.

Actually, there have been a couple of threads that linked them.

There are plenty of people who create multiple accounts purely for the pleasure of being able to "stuff the ballot box" in dpReview challenges or hit the complaint button so many times on people that they don't like that they get people banned.

None of which is occurring here that I'm aware of, and I haven't hit the "Complain" button.

And if you had, would you admit it?

Actually, the only way he would be exhibiting sock puppet behavior in this thread is if he were also you.

That might be one way but surely not the only way and I recognize the behavior even if you don't see it . . . yet.

the sort of thing that's usually associated with young teenage trolls, fanboys and malcontents. The sorts of names that would rarely be used in real life if they could be easily traced back to the person behind them.

Actually, a lot of people know who he really is, and we don't really mind.

Why be so inscrutable? Are you referring to the person behind Mr. Wellhung?

The context made that perfectly obvious. You have repeated the question so many times that it's gone past "rhetorical" into simply "annoying".

I'm not sharing his "real life" identity with an "X".

If so, might it be the same one that was behind Joe Mama? If so, you posted in the thread that discussed Joe Mama's banning and seem to have felt differently at the time :

He violated some rules and got banned. No big deal. Anyone can come
back with a new neame.

Would you like it if it happened to you, and you could never be known here as "Glen Barrington" again?

That's the great unfairness of how this shakes out. Trolls like "natureman" don't use their real names (maybe they're embarrassed by who they really are). They know they have been banned before and will be banned again (and again and again) and they don't care. They come back under new names all the time, because people hated the last person they posed as (and the one before that, etc) so why not let the last ID die...

http://forums.dpreview.com/forums/read.asp?forum=1018&message=28597773

In other words, I've presented an argument as to why someone who was cut off from being able to use their real name, or at least a well known name (a "brand", as it were), might get upset.

Joe Mama and Bob both originally used at least their real first names in their user IDs, and they provided their real last names (Joe in his website address on, as you pointed out, most every post, Bob in his email address in the profile from his first account).

As far as I know, Mr. X, your first name isn't "Bill".

Sorry, but you're the unknown, here.

The sorts of names that show a lack of respect or contempt for those you interact with.

Or a sense of humor. Sometimes technical types have those. Sometimes, a bit strange.

Sometimes. I have a pretty good sense of humor as do you, but it's never been in evidence with these puppies, except the forced, false humor frequently attempted by those that use and abuse LOL and ROFL.

Have you ever taken anything other than an adversarial attitude with them?

--
Rahon Klavanian 1912-2008.

Armenian genocide survivor, amazing cook, scrabble master, and loving grandmother. You will be missed.

Ciao! Joseph

http://www.swissarmyfork.com

 

[Tyrone Wellhung]

I think everyone has got the point that you're not my greatest fan, but I'm cool with that.

BTW, is that picture on your gravatar your natural colour? I'd seek medical help if its a realistic likeness.

 Hardly. Joseph writes clearly and to the point, without the need to continue writing the same thing over and over until the thread maxes out. He has also been here many years and I don't think that anyone would deny that there's no one else like him!  You on the other hand have been here only a couple of moons and if your DPR ID was changed to Croc's or Shull's, nobody would find anything amiss, or as Eric Burdon once sang/emoted "It's all meat from the same bone", Mr. Wellhung.
Â

 

[Billx08]

Your definition is valid but a bit limited. It's also described this way :

A sockpuppet is an online identity used for purposes of deception within an online community. In its earliest usage, a sockpuppet was a false identity through which a member of an Internet community speaks with or about himself or herself, pretending to be a different person,[1] like a ventriloquist manipulating a hand puppet.

I don't see that as conflicting with my definition, at all. The salient point of the sock puppet is "speaks with or about himself".

You well know that I didn't say that your definition was in error. It was just incomplete, and the definition I used was "an online identity used for purposes of deception". You're overly proud to show that you weren't deceived but deception was absolutely part of the intent even if it wasn't intended to be absolutely or entirely deceptive.

I've never seen Bob do that, and I know at least 17 of his previous accounts. Always one after the other. Never interacting with himself, supporting himself. And never, after the first account, talking "about himself".

Totally irrelevant and as it's clear that you have no intent to identify this "Bob" that you so frequently mention, useless other than as a debating tactic.

In fact, his first account had exactly the same form as yours,

[first name][initial][number]

Proving exactly what? That you're a devotee of numerology? There are probably hundreds of DPR IDs that have that form, many of which are are used by very nice, helpful contributors. This debating tactic is based on baseless innuendo. It's clear at this point that you have an agenda, and are on the attack, something you are so well known for. When that happens, like the sock puppets, you'll go on and on and on, never giving in even when you're wrong.

http://en.wikipedia.org/wiki/Sock_puppet_%28internet%29

and

sock puppet (noun): a name or identity used online to deceive others and that is often used to direct praise or attention to oneself

http://www.britannica.com/blogs/2008/01/masstige-nanite-etc-the-open-dictionary/

He doesn't "direct praise or attention to oneself". He never mentions his previous accounts, especially his first one, the one that had all the identifying information.

Try to focus on the word "deceive". I tried to be fair and include the quote with more than sufficient context, but it doesn't mean that I thought that Wellhung was seeking praise (at least directly). As far as calling "attention to oneself", anyone as outspoken as you, Wellhung and even myself does that to a degree, but I never mentioned either that or directing "praise". It was all about the deception of hiding behind a new ID after being banned.

Who is Bob? Has Tyrone Wellhung linked himself to "Bob" or are you basing that on some other information gleaned from your troll studies?

Both. He's open enough to admit it, when asked directly. Like I said, I don't really consider him a "troll". He's more a "troll victim", ambushed by packs of trolls and banned because dpReview places a high emphasis on people who get a lot of complaints, and real trolls do run fleets of sock puppets (one of the dpReview mods once commented that the "sock puppet army" that was "stuffing the ballot box" on the challenges needed to be cleaned up several hundred at a time. That's one troll, with a script, running hundreds of socks).

Neither Wellhung or any of the others have admitted it, but I'll grant that what you're saying is true because when I've mentioned it he/they took it in stride, making no attempts to deny it. I've never seen any of the sock puppets being attacked by packs of trolls or by any other groups, although from what I read in the "Joe Mama Banned" thread it certainly appeared that he had many supporters as well as detractors. Unless you have some convincing evidence to the contrary, I'll continue to assume that those detractors were simply that, people that found Joe Mama to be a forum irritant, but who weren't trolls.

The only Bob that I recall ever getting involved with any of these related sock puppets is Bob Elkind,

Nope, entirely different Bob. But he is a real engineer, and not hard to find.

Yes, and a valuable contributor who unfortunately is not nearly as active as he used to be. Don't see much of Marianne or Thom anymore, and there are reasons for that that don't need to be gone into.

His first account was actually more open and transparent than yours, he had his real last name in the email address he used.

who along with Eliah Borg also commented on the juvenile names

Iliah Borg. And one can comment on the names without disrespecting the arguments or POV.

It has nothing to do with the arguements or POV. It has always been about two things. Endless, circular debate about what could have been discussions of interesting topics, but always devolved into obfuscatory minutiae, and the frequent resorting to snide comments and insults. I've also been critical of the tone of some of Iliah's replies, but he's a good contributor that deserves a lot of respect, and who has practically nothing in common with the sock puppets other than a greater than usual knowledge of complex photographic principles.

I agree 100% with Joseph Jame's arguments, but ended up creating some enmity between us by criticizing his presentation of those arguments.

Wow. I don't recall ever replying to any of his posts, but I also don't disagree with the sock puppet's arguments, mainly with their presentation. We both know that one of your many talents is that you so easily create avoidable enmity so I'm not surprised.

(continued)