Velocity Reviews - Computer Hardware Reviews

Velocity Reviews > Newsgroups > Computing > Digital Photography > Re: The death of the Bayer filter? Maybe not.

Reply
Thread Tools

Re: The death of the Bayer filter? Maybe not.

 
 
TheRealSteve
Guest
Posts: n/a
 
      05-13-2012

On Sat, 12 May 2012 11:17:42 -0400, nospam <(E-Mail Removed)>
wrote:

>In article <(E-Mail Removed)>, TheRealSteve
><(E-Mail Removed)> wrote:
>
>> >> You've proven that you are several times. The latest is by your
>> >> statement above that says you get aliasing in the sampled waveform at
>> >> less than the nyquist frequency. That's totally factually wrong and
>> >> proves beyond a shadow of a doubt that you don't know what you're
>> >> spewing about.
>> >
>> >just to be clear, are you saying that for a given sampling frequency n,
>> >*nothing* at all below n it will alias? that the *only* aliasing is
>> >above n? and what happens *at* n?

>>
>> Even your question shows you don't know what you're talking about.
>>
>> >if that's what you're saying, you're *very* wrong.

>>
>> If you actually mean n/2 and not n, then your question makes some
>> sense. But you are still confused. You can look up online what the
>> answer to your question is. But as long as you realize that it has no
>> bearing on either what the video was showing or our disussion on the
>> bayer sensor and is merely a specious attempt by you to prove that you
>> know at least something about sampling theory, and that attempt
>> failed, then we're good.

>
>i will take your avoidance of answering the question to mean you don't
>actually know the answer and don't want to embarrass yourself any
>further by guessing.


Yet another thinh you're wrong about. And the fact that you're even
asking what happens at n/2 shows you don't know what happens at below
n/2, which is what we're talking about. You want to try and prove
there's aliasing below n/2 by asking about n/2. Well, the answer at
n/2 is "it depends". It's a limit function where some wordings of
sample theory define the limit and others don't. There may be aliasing
if you consider DC an alias and there may not be if you don't.

But while some define what happens at n/2 as aliasing and others
don't, no one who knows anything at all about sample theory defines
what happens at anything less than n/2 as aliasing because any
frequency content at anything less than n/2 can always be converted
back from digital to analog. But in order to do so, you must use a
reconstruction filter. The idiot in the video with the O-Scope was not
using a reconstruction filter.
 
Reply With Quote
 
 
 
 
TheRealSteve
Guest
Posts: n/a
 
      05-13-2012

On Sat, 12 May 2012 11:17:45 -0400, nospam <(E-Mail Removed)>
wrote:

>In article <(E-Mail Removed)>, TheRealSteve
><(E-Mail Removed)> wrote:
>
>> >> >> But if you want to use their marketing hype to compare, then you have
>> >> >> to use it fully. You can't cherry pick. And using their marketing hype
>> >> >> fully, you would have to compare the 45MP Foveon to a 15MP Bayer since
>> >> >> both are counting sensel locations. Now you see why their marketing
>> >> >> hype is somewhat deceptive. But it does give a simple way of showing
>> >> >> why their resolution is better so maybe it's not hype after all.
>> >> >
>> >> >except it's not better.
>> >> >
>> >> >the photos may *look* better to some people because alias artifacts are
>> >> >mistaken for real detail, and sigma adds a lot of sharpening and boosts
>> >> >the contrast too. it's all smoke and mirrors.
>> >> >
>> >> >meanwhile, the colours are off and there's all sorts of ugly blotching
>> >> >in the shadows and banding in the highlights.
>> >>
>> >> The resolution is better.
>> >
>> >it's not, as any test chart shows.

>>
>> It is, as test charts do show.

>
>test charts do *not* show foveon has higher resolution.


But they do.

>> Bayer sensors (without a strong AA
>> filter that defeats the purpose of a resolution test of the sensor)
>> will show alias artifacts at below nyquist. Foveon or 3 sensor systems
>> will not. It's pretty simple, but I guess not simple enough for you to
>> understand.

>
>*any* test chart will show that alias artifacts can occur below nyquist.


Ok, I'll play your game. But bayer sensors have alias artifacts at
much further below the nyquist spatial resolution than Foveon

>if you can't see that, you're either blind or don't understand what
>you're looking at (maybe both).


I know you don't know waht you're looking at.

>> >> But there's plenty of things that can affect
>> >> an image quality other than resolution. They're probably faltering at
>> >> those other things.
>> >
>> >and they do. foveon has a lot of problems and it doesn't solve anything
>> >that needs to be solved either. it's one of the biggest scams.

>>
>> There's yet another area you're confused about. Either that or you
>> just don't care enough about the problems with bayer cfa to worry
>> about them. Floyd's paper described very well some of the problems of
>> the bayer sensor that Foveon solves. You just choose to discount it,
>> which explains a lot.

>
>first of all it wasn't floyd's paper, it was his *link* to someone
>elses paper. you're so confused you can't even get *that* right.


Well, since you were the first one to call it "floyds paper", I went
with that so you would know what I was reffering to. I admit I was
wrong for not correcting you on yet something else you were wrong
about.

>second, foveon doesn't solve a problem that needs to be solved. humans
>can't see the additional chroma resolution that foveon has. it's a
>complete waste. bayer is a *much* better solution for human viewing.


Yet another thing you are confused about. You seem to think that human
vision has anything at all to do with aliasing artifacts of different
sensor technology. The only place where human vision is involved is in
seeing the artifacts, which has nothing to do with chroma resolution
of the human eye and has everything to do with spatial resolution of
the sensor.

We already know what the overall resolution of the human eye is, and
not just the chroma resolution. Apple even trademarked a term for it,
a retina display. So according to your specious argument, a photo
sensor doesn't need to have any more resolution than a retina display.
And that's total nonsense.

>foveon might sound good on paper, but the reality is, humans can't see
>any difference and it's full of shortcomings they *can* see, such as
>noise, blotching, inaccurate colour and being limited to low iso. the
>cameras are slow, buggy and don't have anywhere near the features of
>competing cameras.


I agree with the other limitations of Foveon. But they *do* have
better resolution if you consider they have the ability to produce a
color image where a higher spatial frequency can be captured without
alias artifacts than a bayer sensor of equivalent pixel spacing. And
*that* is the only point I'm arguing and it's a correct one as much as
you and Wolfy want to deny it.

>the entire foveon scam is nothing more than leave off the antialias
>filter for lots of false (and higher frequency) detail that gives the
>illusion of higher resolution, then crank the sharpening and boost
>contrast, resulting in an image that has a lot of snap to it and makes
>people say 'whoa', but an image with a very inaccurate representation
>of the original subject.


Bayer also gives a very inaccurate representation of the original
subject. If the rest of the camera outresolves the sensor, you either
have to blur it with a strong AA filter or put up with color
artifacts. Neither of which faithfully represents the original subject
any more or less than a Foveon. And for both of them, if the camera
does not outresolve the sensor, you don't need an AA filter. But the
Foveon can reach that point sooner than the bayer because for the same
overall pixel density, the foveon has better full color image spatial
resolution.
 
Reply With Quote
 
 
 
 
TheRealSteve
Guest
Posts: n/a
 
      05-13-2012

On Sat, 12 May 2012 11:17:49 -0400, nospam <(E-Mail Removed)>
wrote:

>In article <(E-Mail Removed)>, TheRealSteve
><(E-Mail Removed)> wrote:
>
>> >It's you who masturbates on Bayer and Foveon pixels, and the
>> >latter being "better" in some way. I really don't care about that,
>> >since Foveon is *way* worse for what I do and need. Which is
>> >a lot of very high ISO.

>>
>> Foveon is better in some way. Just because it's not better for what
>> you do, you have to masturbate on how Bayer is the end all and be all
>> sensor for everyone.

>
>foveon is not better in any way that matters to humans. it does have
>higher chroma resolution but people can't see it. for luminance
>resolution, it is much worse than what competing cameras offer, and it
>costs more to boot.


You keep saying that when it doesn't matter at all. People can
absolutely see the alias artifacts that occur because of the lower
sampling rate of the bayer sensor.

>> >> I'm using the exact same definition in
>> >> terms of the physical reality of the image being captured, not
>> >> marketing hpye.
>> >
>> >The image is not even being captured by pixels. It's captured by
>> >sensels. That's the physical reality of the image being captured.

>>
>> Of which the bayer sensor has only one at each pixel location while
>> the foveon and 3-sensor has 3. And since it's *locations* that matter
>> when it comes to spatial resolution and sampling an image, the fact
>> that the bayer has fewer locations that capture each color and then
>> tries to merge these under-sampled locations into a final image is why
>> it has the problems it has.

>
>bayer has far more spatial locations than foveon could ever dream of.
>
>until last year, foveon was stuck at 4.7 megapixels, while bayer was as
>much as 24 mp (or more if you go to medium format). that's a *huge*
>gap.


But we're talking about comparing the same thing. A 4.7MP Foveon will
outresolve a 4.7MP Bayer. A 24MP Foveon will outresolve a 24MP Bayer.
A 4.7MP 3 sensor system will outresolve a 4.7MP Bayer. A 24MP 3 sensor
system will outresolve a 24MP Bayer. Whatever resolution you pick, the
Foveon or 3 sensor system will outresolve the bayer at that
resolution. If Bayer sensors go way above the resolution than any
available Foveon or 3 sensor systems, then the comparision can only be
theoretical. But the theoretical results will be the same. A
theoretical 100MP Foveon or 3 sensor system will outresolve a 100MP
bayer.

>the sd1 upped foveon to 15 mp which was a bit closer to the best bayer,
>but at a ridiculous price. you could get a *real* 40 mp camera for
>slightly more money. nobody bought it, thus the $5000 price cut, and
>it's *still* overpriced, just not as bad.


But since overall sensor resolution at the same pixel spatial
resoltion is the only point I'm comparing on, all of what you're
saying is specious.

>> >So the Foveon marketing hype is much closer to the physical
>> >reality of the image being captured than your interpretation ...

>>
>> You're falling for it again. If foveon use different marketing, such
>> as claiming full color information at each pixel vs. only partial
>> color information at each pixel and having to guess the rest, that
>> would be less hype and more reality.

>
>actually sigma/foveon does claim they don't guess and it's yet another
>lie. foveon guesses far more than bayer ever did. their ads show r/g/b
>but that's not what the sensor captures. not even close.


That part's not a lie. And it doesn't matter whether it's rgb or
something else. As long as they capture the spectrum they require to
produce the final color image at each pixel location, they are doing
better than bayer which captures only part of the required spectrum at
each pixel location and "algorithmically comes up with" (PC for
guesses) the part not captured.

>> >> The difference being shown is that the bayer sensor
>> >> doesn't capture the full color information for each pixel while the
>> >> Foveon and 3 sensor systems do, using the same definition of a pixel
>> >> without falling into marketing traps.
>> >
>> >But they still don't capture it with pixels, they capture it
>> >with sensels.

>>
>> Of course. The Foveon and 3 sensor systems have 3 sensels for each
>> pixel to physically sample all 3 colors that make up each pixel. The
>> bayer cfa has only one sensels and captures only one color at each
>> pixel. It has to algorithmically figure out the rest. This simple
>> concept apparently is too difficult for you to understand, which is
>> why you fall for Sigma's marketing hype. Sigma's not stupid. They're
>> fooling people like you.

>
>looks like they fooled you. foveon does *not* capture rgb, but rather
>it has three overlapping samples which it then tries to calculate rgb,
>but there are so many variables that there is quite a bit of
>uncertainty to it.


As I said above, it doesn't matter how they break the spectrum down.
What matters is that what's needed for the final image is sampled at
each pixel. And that's why the resolution is better.

>bayer measures one colour per location and accurately calculates the
>other two, so it's already *well* ahead of the game.


No, bayer doesn't do either of those things. Bayer also captures a
band of wavelenghts and not strictly r/g/b at each sensel location. So
in that regard, it's the same as Foveon. And if Bayer *could*
accurately calculate the missing information at each pixel, it would
not have color artifacts. The fact that even you admit it does have
color artifacts proves you're wrong when you way it "accurately
calculates the other two", and that it's not *well* ahead of the game.

>the conversion from foveon space to rgb is nonlinear and the samples
>have a lot of overlap, which means there's actually a whole lot of
>guessing going on in foveon, which is why there are odd colour casts,
>blotches and weird effects (e.g., overexposed reds turning pink). every
>once in a while there are psychadelic effects. it's a mess.


You mean just like bayer does, but for different reasons. And the only
thing I'm talking about is resolution, and the bayers psychadelic
effects, creating a mess in the image, are a result of lower
resolution than the Foveon. That the Foveon has other problems is
immaterial. As far as my argument is concerned, Foveon was just an
analog for a 3 sensor system anyway. Both of them sample what they
need at each pixel and the bayer does not. And that's the *only* thing
that matters for my position.

>> >2) You're the one who wants "the physical reality of the
>> > image being captured" ...

>>
>> Which is exactly what I'm doing. The physical reality is that Foveon
>> and 3 sensor systems have the full color information of 3 sensels for
>> each pixel location. The Bayer cfa does not.

>
>nope. foveon definitely doesn't have full colour information (see
>above). only 3 sensor systems do and at a huge cost and weight penalty.


Even 3 sensor systems don't necessarily have *full* color information.
But like the Foveon, they sample what they need to create a final
image that the eye can see as full color at each pixel location. The
bayer does not. And sampling everything that's needed at each pixel
location vs. some things at every other pixel location and other
things at even less than every other pixel location is why the Foveon
and 3 sensor systems outresolve bayer for the same pixel spacing. And
that's the *only* point I'm making. Anything else you bring up that's
not related to that point (price, weight, cost of 3 sensor systems,
other problems Foveon may have) is specious for arguing against that
point.

>> The physical reality is
>> that each color channel of the Foveon and 3 sensor systems are sampled
>> at the full spatial resolution of the sensor. They are not sampled at
>> that full rate for the Bayer cfa.

>
>bayer samples at the full rate just like foveon does. you still don't
>get it.


Ah, you finally bring up a point that's actually not specious. It's
factually wrong, but at least it's not specious.

>> The bayer cfa has plenty of advantages over both Foveon and 3-sensor
>> systems. But this whole discussion has been focusing on only
>> resolution and preventing aliasing artifacts and that's probably the
>> only areas that the Bayer lags behind the other two. But you have such
>> a hardon for bayer that you can't even admit the fact that bayer does
>> lag behind the other two in this simple point of comparison. You'd
>> like to think Bayer is better at *everything* but it's not.

>
>bayer doesn't lag behind. it's *well* ahead of foveon and laughably so.


It does lag behind on the one point of comparison I'm making. It's
well ahead on most other areas of comparison. The fact that you can't
see that means you have your space hat and blinders on.

>like i said, until last year, foveon was stuck at under 5 mp while
>bayer had 24 mp. with the sd1, it's a little closer now but bayer still
>has over twice as many pixels and canon's entry level slrs at 1/4 the
>price of the sd1m (and 1/10th the price of the original sd1) has *more*
>pixels!


Back to making specious arguments because the one you tried to make
that wasn't specious was factually wrong.

>the real question is just how long will sigma keep up this game.


It all depends on how much longer people buy them. I'm surprised
people still are. If Sigma were smart, they would bring a small Foveon
to market for things like cellphones and tablets. They may have a
chance with that.
 
Reply With Quote
 
TheRealSteve
Guest
Posts: n/a
 
      05-13-2012

On Sun, 13 May 2012 03:32:14 +0200, Wolfgang Weisselberg
<(E-Mail Removed)> wrote:

>TheRealSteve <(E-Mail Removed)> wrote:
>> On Fri, 11 May 2012 22:35:38 +0200, Wolfgang Weisselberg
>>>TheRealSteve <(E-Mail Removed)> wrote:
>>>> On Thu, 10 May 2012 18:09:03 +0200, Wolfgang Weisselberg
>>>>>TheRealSteve <(E-Mail Removed)> wrote:
>>>>>> On Tue, 8 May 2012 00:36:49 +0200, Wolfgang Weisselberg
>>>>>>>TheRealSteve <(E-Mail Removed)> wrote:
>>>>>>>> On Sat, 5 May 2012 22:27:42 +0200, Wolfgang Weisselberg

>
>>>>>> [...]
>>>>>>>> But if you want to use their marketing hype to compare, then you have
>>>>>>>> to use it fully.

>
>>>>>>>I do.
>>>>>>>Bayer -- megapixels as used by the people who make and market
>>>>>>>such cameras.
>>>>>>>Foveon -- megapixels as used by the people who make and market
>>>>>>>such cameras.

>
>>>>>No answer is also an answer.

>
>>>> Just because you didn't read or understand my answer doesn't mean I
>>>> didn't answer it.

>
>>>You didn't answer how I didn't use the marketing hype fully
>>>to compare.

>
>> The answer was how you *did* use the marketing hype fully to compare.

>
>I used it correctly, of course.
>
>> Why should I say you didn't use marketing hype to compare when that's
>> exactly what you did?

>
>"But if you want to use their marketing hype to compare, then
>you have to use it fully." Said you. I did. Your point?


No, you absolutely did not. To use their marketing hype fully without
cherry picking, you would have to compare their 45MP camera to a bayer
15MP camera, not to a 45MP bayer camera.


>> Sigma's defiition isn't the same definition as
>> everyone else's.

>
>Sigma's *SENSOR* isn't the same sensor as everyone else's.
>So of course it hast to be counted in other ways.


No it doesn't *have* to be counted in other ways. They do count it in
other ways to fool people like you.

>> They triple it for marketing hype.

>
>They count their sensels. Which is only fair: Bayer does it
>too.


And they have 3 sensels at each pixel location while bayer has only
one. So? All that means is that they're sampling everything they need
to generate a full color pixel at each location. A bayer sensor does
not. That doesn't mean you have to change the definition of things
like spatial frequency and nyquist spatial samplng rate. Those are the
things you have to look at when comparing resolution, not the
whimsical definition of a pixel by someone trying to trick you into
thinking they have more pixels, of which you've obviously fallen for.


>> You simply *do* compare different
>> definitions of pixels as it relates to what goes into the image being
>> captured and presented.

>
>I simply *do* compare different sensor types --- i.e. what
>relates to what goes into the image being captured.


Too bad that's not what you're doing.

>I don't care about the presentation one bit, since that
>depends on printers, paper, processes ... which are simply
>not comparable.


Ah, but those things are also comparable as long as you define
consistent parameters for comparison... something you are unable to
do.

>>>> If a camera manufacturer defined a
>>>> pixel as the full bucket count of electrons that could be captured at
>>>> each sensel location, you would believe them when they said they have
>>>> a 10000GP sensor and would use that marketing tripe to compare to a
>>>> bayer 10000GP sensor.

>
>>>Nope. I would merely notice that with a TheRealSteve sensor 25GP
>>>would equal about 1 MP on a Bayer pattern sensor.

>
>> Lol, which is exactly what you *should* be doing with the Foveon vs.
>> Bayer.

>
>This is what I am doing.


But it's not what you did do when you tried to compare a 45MP "sigma
pixel" camera with a 45MP bayer. As above, you should simply notice
that with a Sigma sensor, the 45MP would equal about 15MP on a bayer
pattern sensor. Yet one more way you are inconsistent.

>> Sigma's definition of a 45MP sensor would equal about 15MP on a
>> Bayer pattern sensor.

>
>WRONG! Sigma's definition of a 45MP sensor would equal about 21
>MP on a Bayer pattern sensor. You're falling for Bayer pattern
>marketing hype and don't even notice it. (And for the record,
>nospam's wrong about that, but he could easily find out what
>frequencies (measured in pixel lengths) appear in real world
>images.)


Ah, so you admit you're wrong when you were trying to compare a 45MP
Foveon with a 45MP bayer. Excellent. And whether you want to say 21
because of pixel pitch or 15 because of 3 sensels vs. 1 sensel = a
pixel is immaterial. My only argument is that it's less, which you
finally agree with.

>> if you use the same definition of
>> pixels.

>
>You don't. That's why you think Foveon is better.


Obviously you think so also if, using your comparison, two sensors
with equivalent pixel pitch would work out to 45MP for a Foveon and
21MP for a bayer. Whether or not that makes sense doesn't matter
because it's your definition.

>>>>>> But here's a
>>>>>> perfect example of yours. You're trying to compare two different
>>>>>> definitions of megapixels which shows you don't really know what a
>>>>>> megapixel is and will believe whatever Sigma is feeding you.

>
>>>>>That's not cherry picking, that's just using the definitions as
>>>>>used by their respective proponents. That doesn't change anything
>>>>>in the results, except for your nya nya nya fetishism.

>
>>>> So you even admit you're using different definitions (the different
>>>> ones used by their respective proponents) in your comparison.

>
>>>No, I admit to using the same definition. "As used by the people
>>>who make and market such cameras".

>
>> "As used by the people who make and market such cameras" leads to
>> different definitions because one manufacturer who makes and markets
>> such cameras uses a different definition than everyone else.

>
>They also use a different sensor than everyone else.


And their sensor has better color spatial resolution for the same
spatial sampling rate because they sample what they need for a color
image at each location while a bayer does not. So yes, it's different.
And whatever definition you use for a color pixel in order to compare
them had better show that difference or else it's not a valid
comparison. If you choose to use Sigma's simplification for marketing
hype because you don't understand sampling theory, fine. Then use it.
But if you do understand what's going on and define your pixels
equivalently, then you'd see that for the same pixel count, the Sigma
won't alias as much at the same input spatial frequencies below
nyquist. If you go *above* nyquist, then all bets are off.

>> But you
>> have such a fetish for nya nya nya (your terms) that you can't see
>> that.

>
>You must be really upset I found out about *your* fetish that
>you try to hang it on me.


You're projecting again.

>>>> The fact
>>>> that you know you're doing something wrong just to say nya nya nya
>>>> shows your fetish.

>
>>>The fact that I am even talking to a crackpot like you is
>>>indeed proof that I do something wrong.

>
>> The fact that you think I'm the crackpot when you're the one who can't
>> even keep the definition of a pixel straight and falls for simple
>> misleading marketing tripe is indeed proof that you're the crackpot.

>
>Yes, everyone is a crackpot but you.


No, just you.

>>>>>>>> And using their marketing hype
>>>>>>>> fully, you would have to compare the 45MP Foveon to a 15MP Bayer since
>>>>>>>> both are counting sensel locations.

>
>>>>>>>Why would I have to do that?
>>>>>>>Sensels and locations aren't even used in the marketing hype.
>>>>>>>Only megapixels.

>
>>>>>> Exactly! And that's why you can't compare megapixels between Sigma's
>>>>>> definition for Foveon and Bayer.

>
>>>>>Of course you can!
>>>>>It just destroys your nya nya nya.

>
>>>> You really have a hard on about nya nya nya.

>
>>>It's you who masturbates on Bayer and Foveon pixels, and the
>>>latter being "better" in some way. I really don't care about that,
>>>since Foveon is *way* worse for what I do and need. Which is
>>>a lot of very high ISO.

>
>> Foveon is better in some way.

>
>Yes, it gives money to Sigma if you buy one.


Yes, and it has better resolution for the same pixel spacing.

>> Just because it's not better for what
>> you do, you have to masturbate on how Bayer is the end all and be all
>> sensor for everyone.

>
>Interesting, how you project your own insecurities and
>motives on other people.


Interesting how when I use your stupid arguments against you, you
think it's projecting. That really shows your insecurities.

>>>> So much so that even you
>>>> know you're (at least I think you know) that you're not making sense
>>>> in your comparison just to say nya nya nya.

>
>>>You're really an idiot, you cannot even come up with your own
>>>arguments, you just steal them.

>
>> No, I just take your stupid ones (nya nya nya, masturbates over
>> sensors, etc.) and use them against you just to show how inane they
>> are.

>
>So you say stupid things. That makes you an idiot.


Quoting your stupidity to show you how stupid you are doesn't make me
an idiot. But arguing your stupidity with an idiot like you certainly
lowers me to some extent. I'll readily admit that.

>> The fact that you think they are idiotic proves just what an idiot you are.

>
>But I don't!


Ok, you called them stupid things, not idiotic. My mistake. So you say
stupid things and I merely point out to you how stupid they are and
you agree they are stupid things. Too funny!

>They're wrong the way you apply them to me, that's all.
>And you really cannot come up with something original?


Stupid things are stupid things and you say them and I point out how
stupid they are. If you think your saying of "nya nya nya" and
"masturbates" are worthy of being called original arguments in an
adult discussion, then you really are as stupid as you sound.

>>>>>> It's precisely because selsels and
>>>>>> locations aren't used in the marketing hype, they are fooling you.

>
>>>>>You change the megapixel definition for Foveon, why not for
>>>>>Bayer, too?

>
>>>> Ok, new definition for Bayer. A bayer sensor has as many pixels as the
>>>> full well capacity of the entire sensor. Now compare them. Hey, it's
>>>> just as valid as what you're trying to do above.

>
>>>Yes, now you finally have a way to really trample Bayer into
>>>the ground. Congratulations.

>
>> Your sarcasm is misplaced since I'm only taking to extreme what you're
>> doing with using Sigma's definition of pixels to compare with everyone
>> elses.

>
>The day you manufacture and sell cameras with your own sensor
>type is the day you can define how to count your pixels.
>I guess hell freezes over first, though.


I see. So we have to take your definition because you manufacture and
sell cameras with your own sensor. All hail the great stupid one.

>> At least now you see how stupid your argument is.

>
>At least now I see how stupid you are.


Good comeback. Very original. Well thought out and presented. Yes, you
are very smart.

>>>>>>>So, I compare straight: megapixels to megapixels.

>
>>>>>> And there's where you're falling into the trap of their marketing
>>>>>> hype. You're comparing Sigma's definition of megapixels to the rest of
>>>>>> the world's definition of megapixels. And they are different.

>
>>>>>You're comparing 3 sensels pixels to one sensel pixels.
>>>>>You're falling for your own hype, comparing *your* definition of
>>>>>megapixels to the rest of the world's definition of megapixels.

>
>>>> But that's where you're wrong.

>
>>>I'm exactly right.

>
>> You've already proven to yourself that your wrong when your own
>> argument was used against you and you treated it with sarcasm.

>
>Yep, that was the day I invented my own definition of pixels.
>
>What was it again? Ah, yes, a camera has 100.000.000/IQ of the
>user pixels. Which would mean you get all the 50+ MPix cameras,
>and normal people like me have to make do with roughly 1 MPix.


Original, but specious.

>Oh, that's as stupid as your own definition. And completely
>useless, since you're on the physics on how the image is
>recorded trip. Which means you need to count sensels.


That just shows you don't know what spatial sampling rate means. You
can have 100 sensels all stacked up on eachother (or on 100 separate
sensors that are aligned in image spatial space) and it's still
capturing only one spatial location. When you want to compare spatial
resolution, you have to compare spatial sampling locations and not
sensels. You're so hung up on trying to understand what pixels are and
Sigma vs. bayer definitions that you miss the spatial sampling
fundamentals of capturing the image in the first place.

>>>> I'm using the exact same definition in
>>>> terms of the physical reality of the image being captured, not
>>>> marketing hpye.

>
>>>The image is not even being captured by pixels. It's captured by
>>>sensels. That's the physical reality of the image being captured.

>
>> Of which the bayer sensor has only one at each pixel location while
>> the foveon and 3-sensor has 3.

>
>Yep.
>
>> And since it's *locations* that matter
>> when it comes to spatial resolution and sampling an image,

>
>Assuming identical resolution from lenses, assuming identical
>beam splitting from AA filters, etc --- all of which is not
>given --- ...


YES, assuming everything else is identical. YES, it's not given
everything else will be identical. But if you want to compare the
sensors, you make everything else the same. Or didn't they teach you
that in high school science class.

>> the fact
>> that the bayer has fewer locations that capture each color

>
>Foveon doesn't capture colour at all. Which is why it has so
>many problems.
>
>The upper layer of a Foveon sensor captures all frequencies, but
>not all of the photons that are actually converted to electrons ,
>only some. How much it captures depends e.g. on random happenings
>of manufacturing of that pot. The lower pots also are larger,
>and thus capture again all frequencies, but not all of the
>photons (which are actually converter), in a slightly different
>composition. And so on.
>
>So Foveon captures 3 sets of luminosity --- however, these sets
>are not panchromatic or even orthochromatic. There's no "this
>is blue" and "this is green" like in a Bayer sensor.
>
>There's 3 very noisy signals that are guessed into some
>approximation of some RGB stimuli. Sometimes the guesses are
>quite right, sometimes they are way off.


Yes, Foveon has problems. In fact, I don't even know how we got onto
Foveon in the first place when I was only comparing 3 sensor systems.
The only point of comparison I'm making regarding Foveon is that it
does physically sample everything it needs at each photosite while a
bayer does not. That's really the only way it's comparable to a 3
sensor system and it's the only point of comparison I'm making.

>> and then
>> tries to merge these under-sampled locations into a final image is why
>> it has the problems it has.

>
>Bayer doesn't have any relevant problems that can be solved
>by Foveon.


Relevant is a relative term. What's not relevant to you might be to
someone else. But at least you added the qualifier "relevant."


>> Yes, they have 3x the sensels for the same pixel density. But not 3x
>> the pixels.

>
>Since *you* want "the physical eality of the image being captured"
>you *have* to count sensels. No matter if you decide that
>they have or have not 3x the pixels.


No, I am counting the spatial resolution of everything that goes into
making the final image. It doesn't matter whether it's sensels, pixels
or whatever else you want to come up with.

[...]
>I'll not even go into luminance carrying way more information than
>chrominance for the human eye and into the fact that chrominance
>is *way* *way* *way* undersampled in the human eye, and is placed
>in random locations on a 2D surface, not stacked like Foveon.
>Such details would only give you a migraine attack.


None of which, of course, matters when it comes to aliasing artifacts.

>Actually, nope. Bayer counts electrons (-> photons) passing
>through a (broad) bandpass after an AA filter. I.e. Bayer
>captures luminosity values after a bandpass. This is not the
>same as "one color". Lay people may think this, however.
>
>> It has to algorithmically figure out the rest.

>
>Just as Foveon has to algorithmically figure out all 3
>colours (and does not manage well).


Which, again, is a different argument not related to alias artifacts.

>>>2) You're the one who wants "the physical reality of the
>>> image being captured" ...

>
>> Which is exactly what I'm doing.

>
>... has to COUNT SENSELS.


When it comes to spatial resolution, it absolutely does not have to
count sensels. Counting sensels gives you the wrong answer.

>> The Bayer cfa does not. The physical reality is
>> that each color channel of the Foveon and 3 sensor systems are sampled
>> at the full spatial resolution of the sensor.

>
>Only true for 3-sensor systems, but there the sensels don't
>align.


Good, then lets go back to only 3 sensor systems, which is what I was
originally talking about before someone else brought up Foveon. The
only way Foveon is equivalent is that it captures what it needs to
create a color image at every spatial photosite. Other than that,
there's not much comparison.

>> They are not sampled at
>> that full rate for the Bayer cfa.

>
>And still Bayer produces perfectly good RGB triples ...


No it does not. There are color artifacts that would not be there if
it produced "perfectly good" rgb triples.

>> The bayer cfa has plenty of advantages over both Foveon and 3-sensor
>> systems. But this whole discussion has been focusing on only
>> resolution and preventing aliasing artifacts and that's probably the
>> only areas that the Bayer lags behind the other two.

>
>Bayer has more resolution per sensel than the other two. So
>it does not lag behind. As for aliasing artifacts, they are
>worse with Foveon --- no AA filter.


Now you're not keeping things equal. That figures, since you don't
think you have to.
 
Reply With Quote
 
nospam
Guest
Posts: n/a
 
      05-14-2012
In article <(E-Mail Removed)>, TheRealSteve
<(E-Mail Removed)> wrote:

> >> >It's you who masturbates on Bayer and Foveon pixels, and the
> >> >latter being "better" in some way. I really don't care about that,
> >> >since Foveon is *way* worse for what I do and need. Which is
> >> >a lot of very high ISO.
> >>
> >> Foveon is better in some way. Just because it's not better for what
> >> you do, you have to masturbate on how Bayer is the end all and be all
> >> sensor for everyone.

> >
> >foveon is not better in any way that matters to humans. it does have
> >higher chroma resolution but people can't see it. for luminance
> >resolution, it is much worse than what competing cameras offer, and it
> >costs more to boot.

>
> You keep saying that when it doesn't matter at all. People can
> absolutely see the alias artifacts that occur because of the lower
> sampling rate of the bayer sensor.


you're so hung up on aliasing that you didn't even read what i wrote.

nowhere did i say anything about aliasing. i'm talking about chroma
resolution, which is foveon's sole advantage, and one which humans
won't ever notice because they can't see the additional chroma
resolution it provides. bayer is already *well* above what the eye can
resolve. it's a waste.

> >> >> I'm using the exact same definition in
> >> >> terms of the physical reality of the image being captured, not
> >> >> marketing hpye.
> >> >
> >> >The image is not even being captured by pixels. It's captured by
> >> >sensels. That's the physical reality of the image being captured.
> >>
> >> Of which the bayer sensor has only one at each pixel location while
> >> the foveon and 3-sensor has 3. And since it's *locations* that matter
> >> when it comes to spatial resolution and sampling an image, the fact
> >> that the bayer has fewer locations that capture each color and then
> >> tries to merge these under-sampled locations into a final image is why
> >> it has the problems it has.

> >
> >bayer has far more spatial locations than foveon could ever dream of.
> >
> >until last year, foveon was stuck at 4.7 megapixels, while bayer was as
> >much as 24 mp (or more if you go to medium format). that's a *huge*
> >gap.

>
> But we're talking about comparing the same thing. A 4.7MP Foveon will
> outresolve a 4.7MP Bayer.


no it won't, but let's say it does just for fun.

the 4.7mp sd14 came out in 2007, well after 10mp bayer sensors were
widely available, and for twice the price too! it's no surprise that
the sd14 price dropped to $350 in a year or so.

sigma *still* uses that same sensor in several cameras (dp1 & dp2
variants and the sd15), except that entry level slrs are now 18 mp and
the best 35mm was 24 until the d800 came along and now it's 36mp.

so even if a 4.7 mp foveon could somehow outresolve a 4.7 mp bayer, it
doesn't actually matter because bayer is *so* far ahead.

even in an absolute worst case scenario, an 18 mp canon slr will
outperform a foveon 4.7mp camera in every possible way, for *less*
money.

> A 24MP Foveon will outresolve a 24MP Bayer.


again, no.

> A 4.7MP 3 sensor system will outresolve a 4.7MP Bayer. A 24MP 3 sensor
> system will outresolve a 24MP Bayer. Whatever resolution you pick, the
> Foveon or 3 sensor system will outresolve the bayer at that
> resolution. If Bayer sensors go way above the resolution than any
> available Foveon or 3 sensor systems, then the comparision can only be
> theoretical. But the theoretical results will be the same. A
> theoretical 100MP Foveon or 3 sensor system will outresolve a 100MP
> bayer.


that's a flawed comparison. the 15 mp sd1 costs $2400, yet for around
$600 or so, you can get a canon entry level slr with 18 mp and for just
a little more than what the sd1 costs, you can get 21-36mp, either of
which will do better than foveon.

> >the sd1 upped foveon to 15 mp which was a bit closer to the best bayer,
> >but at a ridiculous price. you could get a *real* 40 mp camera for
> >slightly more money. nobody bought it, thus the $5000 price cut, and
> >it's *still* overpriced, just not as bad.

>
> But since overall sensor resolution at the same pixel spatial
> resoltion is the only point I'm comparing on, all of what you're
> saying is specious.


it's not specious at all. $600 buys a camera that has more spatial
resolution than the $2400 sd1m.

> >> >So the Foveon marketing hype is much closer to the physical
> >> >reality of the image being captured than your interpretation ...
> >>
> >> You're falling for it again. If foveon use different marketing, such
> >> as claiming full color information at each pixel vs. only partial
> >> color information at each pixel and having to guess the rest, that
> >> would be less hype and more reality.

> >
> >actually sigma/foveon does claim they don't guess and it's yet another
> >lie. foveon guesses far more than bayer ever did. their ads show r/g/b
> >but that's not what the sensor captures. not even close.

>
> That part's not a lie. And it doesn't matter whether it's rgb or
> something else. As long as they capture the spectrum they require to
> produce the final color image at each pixel location, they are doing
> better than bayer which captures only part of the required spectrum at
> each pixel location and "algorithmically comes up with" (PC for
> guesses) the part not captured.


call it whatever you want, but the fact remains that bayer is *more*
accurate than foveon, so those 'guesses,' as you call it, are pretty
damned good guesses. so much so that it's either wildly amazing luck or
it's not guessing at all.

> >> >> The difference being shown is that the bayer sensor
> >> >> doesn't capture the full color information for each pixel while the
> >> >> Foveon and 3 sensor systems do, using the same definition of a pixel
> >> >> without falling into marketing traps.
> >> >
> >> >But they still don't capture it with pixels, they capture it
> >> >with sensels.
> >>
> >> Of course. The Foveon and 3 sensor systems have 3 sensels for each
> >> pixel to physically sample all 3 colors that make up each pixel. The
> >> bayer cfa has only one sensels and captures only one color at each
> >> pixel. It has to algorithmically figure out the rest. This simple
> >> concept apparently is too difficult for you to understand, which is
> >> why you fall for Sigma's marketing hype. Sigma's not stupid. They're
> >> fooling people like you.

> >
> >looks like they fooled you. foveon does *not* capture rgb, but rather
> >it has three overlapping samples which it then tries to calculate rgb,
> >but there are so many variables that there is quite a bit of
> >uncertainty to it.

>
> As I said above, it doesn't matter how they break the spectrum down.


actually it does matter, since what foveon does is very inaccurate and
also contradicts sigma's bogus claims about bayer guessing.

> What matters is that what's needed for the final image is sampled at
> each pixel. And that's why the resolution is better.


except, what is sampled at each foveon pixel isn't what's needed for
the final image. there's a lot more that goes into the 'guess.'

> >bayer measures one colour per location and accurately calculates the
> >other two, so it's already *well* ahead of the game.

>
> No, bayer doesn't do either of those things.


yes it most certainly does. seriously, go read some papers on bayer. or
just look at the photos.

> Bayer also captures a
> band of wavelenghts and not strictly r/g/b at each sensel location. So
> in that regard, it's the same as Foveon.


it's nothing at all like foveon. not even remotely close. where did you
come up with that rubbish?

> And if Bayer *could*
> accurately calculate the missing information at each pixel,


and in fact, it does, more so than foveon.

> it would
> not have color artifacts.


that's an entirely separate issue.

> The fact that even you admit it does have
> color artifacts proves you're wrong when you way it "accurately
> calculates the other two", and that it's not *well* ahead of the game.


you're confusing two different things.

> >the conversion from foveon space to rgb is nonlinear and the samples
> >have a lot of overlap, which means there's actually a whole lot of
> >guessing going on in foveon, which is why there are odd colour casts,
> >blotches and weird effects (e.g., overexposed reds turning pink). every
> >once in a while there are psychadelic effects. it's a mess.

>
> You mean just like bayer does, but for different reasons. And the only
> thing I'm talking about is resolution, and the bayers psychadelic
> effects, creating a mess in the image, are a result of lower
> resolution than the Foveon.


nope.

> That the Foveon has other problems is
> immaterial.


it's not immaterial at all. what matters is the final image, which for
foveon, is not very good. bayer consistently produces excellent images,
without the blotching, noise, weird colours, etc. that plague foveon.

> As far as my argument is concerned, Foveon was just an
> analog for a 3 sensor system anyway. Both of them sample what they
> need at each pixel and the bayer does not. And that's the *only* thing
> that matters for my position.


except, it's wrong.

> >> >2) You're the one who wants "the physical reality of the
> >> > image being captured" ...
> >>
> >> Which is exactly what I'm doing. The physical reality is that Foveon
> >> and 3 sensor systems have the full color information of 3 sensels for
> >> each pixel location. The Bayer cfa does not.

> >
> >nope. foveon definitely doesn't have full colour information (see
> >above). only 3 sensor systems do and at a huge cost and weight penalty.

>
> Even 3 sensor systems don't necessarily have *full* color information.


what?

wasn't that they have full colour their advantage? now they don't have
full colour? make up your mind.

> But like the Foveon, they sample what they need to create a final
> image that the eye can see as full color at each pixel location. The
> bayer does not.


bayer most certainly does sample what it needs, otherwise the pictures
would look crappy and they do not. you just don't understand how it
works.

> And sampling everything that's needed at each pixel
> location vs. some things at every other pixel location and other
> things at even less than every other pixel location is why the Foveon
> and 3 sensor systems outresolve bayer for the same pixel spacing. And
> that's the *only* point I'm making.


then your only point is *wrong*.

> Anything else you bring up that's
> not related to that point (price, weight, cost of 3 sensor systems,
> other problems Foveon may have) is specious for arguing against that
> point.


it's not specious. even if a 3 sensor system was better, it comes at a
high cost (money, size and weight) and something almost as good is much
cheaper, smaller and lighter.

in other words, it's not worth it.

> >> The physical reality is
> >> that each color channel of the Foveon and 3 sensor systems are sampled
> >> at the full spatial resolution of the sensor. They are not sampled at
> >> that full rate for the Bayer cfa.

> >
> >bayer samples at the full rate just like foveon does. you still don't
> >get it.

>
> Ah, you finally bring up a point that's actually not specious. It's
> factually wrong, but at least it's not specious.


it's exactly correct.

> >> The bayer cfa has plenty of advantages over both Foveon and 3-sensor
> >> systems. But this whole discussion has been focusing on only
> >> resolution and preventing aliasing artifacts and that's probably the
> >> only areas that the Bayer lags behind the other two. But you have such
> >> a hardon for bayer that you can't even admit the fact that bayer does
> >> lag behind the other two in this simple point of comparison. You'd
> >> like to think Bayer is better at *everything* but it's not.

> >
> >bayer doesn't lag behind. it's *well* ahead of foveon and laughably so.

>
> It does lag behind on the one point of comparison I'm making. It's
> well ahead on most other areas of comparison. The fact that you can't
> see that means you have your space hat and blinders on.


the *only* area in which bayer is behind foveon is in chroma resolution
and once again, humans can't see the difference.

> >like i said, until last year, foveon was stuck at under 5 mp while
> >bayer had 24 mp. with the sd1, it's a little closer now but bayer still
> >has over twice as many pixels and canon's entry level slrs at 1/4 the
> >price of the sd1m (and 1/10th the price of the original sd1) has *more*
> >pixels!

>
> Back to making specious arguments because the one you tried to make
> that wasn't specious was factually wrong.


nothing specious about any of that.

> >the real question is just how long will sigma keep up this game.

>
> It all depends on how much longer people buy them. I'm surprised
> people still are. If Sigma were smart, they would bring a small Foveon
> to market for things like cellphones and tablets. They may have a
> chance with that.


actually, that would be incredibly stupid.

foveon tried cellphone sensors once already and it failed miserably, so
much so, that it was the cause of their near bankruptcy, which is how
sigma ended up buying them for pennies on the dollar.
 
Reply With Quote
 
nospam
Guest
Posts: n/a
 
      05-14-2012
In article <(E-Mail Removed)>, TheRealSteve
<(E-Mail Removed)> wrote:

> And they have 3 sensels at each pixel location while bayer has only
> one. So? All that means is that they're sampling everything they need
> to generate a full color pixel at each location. A bayer sensor does
> not.


bayer definitely samples everything it needs. just look at the photos.

if it didn't sample what it needed, they would not look as good as they
do.

> And their sensor has better color spatial resolution for the same
> spatial sampling rate because they sample what they need for a color
> image at each location while a bayer does not. So yes, it's different.


not true, but regardless, the higher chroma resolution isn't anything
humans can see, so why bother capturing it? it's a waste of storage
space as well as the time to process it.

> >Oh, that's as stupid as your own definition. And completely
> >useless, since you're on the physics on how the image is
> >recorded trip. Which means you need to count sensels.

>
> That just shows you don't know what spatial sampling rate means. You
> can have 100 sensels all stacked up on eachother (or on 100 separate
> sensors that are aligned in image spatial space) and it's still
> capturing only one spatial location.


that part is true.

> When you want to compare spatial
> resolution, you have to compare spatial sampling locations and not
> sensels. You're so hung up on trying to understand what pixels are and
> Sigma vs. bayer definitions that you miss the spatial sampling
> fundamentals of capturing the image in the first place.


in other words, a 15 mp foveon sensor (45mp in sigma-speak) has the
same spatial sampling as a 15 mp bayer.

you can't keep your own story straight.

> >Foveon doesn't capture colour at all. Which is why it has so
> >many problems.
> >
> >The upper layer of a Foveon sensor captures all frequencies, but
> >not all of the photons that are actually converted to electrons ,
> >only some. How much it captures depends e.g. on random happenings
> >of manufacturing of that pot. The lower pots also are larger,
> >and thus capture again all frequencies, but not all of the
> >photons (which are actually converter), in a slightly different
> >composition. And so on.
> >
> >So Foveon captures 3 sets of luminosity --- however, these sets
> >are not panchromatic or even orthochromatic. There's no "this
> >is blue" and "this is green" like in a Bayer sensor.
> >
> >There's 3 very noisy signals that are guessed into some
> >approximation of some RGB stimuli. Sometimes the guesses are
> >quite right, sometimes they are way off.

>
> Yes, Foveon has problems. In fact, I don't even know how we got onto
> Foveon in the first place when I was only comparing 3 sensor systems.
> The only point of comparison I'm making regarding Foveon is that it
> does physically sample everything it needs at each photosite while a
> bayer does not. That's really the only way it's comparable to a 3
> sensor system and it's the only point of comparison I'm making.


so how is it that bayer is more accurate than foveon, if bayer doesn't
sample 'everything it needs' ?

obviously, not only does bayer sample everything it needs, but it does
so with greater accuracy.

> >I'll not even go into luminance carrying way more information than
> >chrominance for the human eye and into the fact that chrominance
> >is *way* *way* *way* undersampled in the human eye, and is placed
> >in random locations on a 2D surface, not stacked like Foveon.
> >Such details would only give you a migraine attack.

>
> None of which, of course, matters when it comes to aliasing artifacts.


he didn't say anything about aliasing. try to stay on topic.

> >Actually, nope. Bayer counts electrons (-> photons) passing
> >through a (broad) bandpass after an AA filter. I.e. Bayer
> >captures luminosity values after a bandpass. This is not the
> >same as "one color". Lay people may think this, however.
> >
> >> It has to algorithmically figure out the rest.

> >
> >Just as Foveon has to algorithmically figure out all 3
> >colours (and does not manage well).

>
> Which, again, is a different argument not related to alias artifacts.


alias artifacts don't matter that much when the colours are wrong and
the image is full of blotches and noise.
 
Reply With Quote
 
nospam
Guest
Posts: n/a
 
      05-14-2012
In article <(E-Mail Removed)>, TheRealSteve
<(E-Mail Removed)> wrote:

> >> >> >the photos may *look* better to some people because alias artifacts are
> >> >> >mistaken for real detail, and sigma adds a lot of sharpening and boosts
> >> >> >the contrast too. it's all smoke and mirrors.
> >> >> >
> >> >> >meanwhile, the colours are off and there's all sorts of ugly blotching
> >> >> >in the shadows and banding in the highlights.
> >> >>
> >> >> The resolution is better.
> >> >
> >> >it's not, as any test chart shows.
> >>
> >> It is, as test charts do show.

> >
> >test charts do *not* show foveon has higher resolution.

>
> But they do.


i don't know what you're looking at but they definitely do *not* show
that.

what they show is that foveon and bayer resolve roughly the same, with
aliasing becoming a problem at about the same point below nyquist.

> >> Bayer sensors (without a strong AA
> >> filter that defeats the purpose of a resolution test of the sensor)
> >> will show alias artifacts at below nyquist. Foveon or 3 sensor systems
> >> will not. It's pretty simple, but I guess not simple enough for you to
> >> understand.

> >
> >*any* test chart will show that alias artifacts can occur below nyquist.

>
> Ok, I'll play your game. But bayer sensors have alias artifacts at
> much further below the nyquist spatial resolution than Foveon


they do not.

> >if you can't see that, you're either blind or don't understand what
> >you're looking at (maybe both).

>
> I know you don't know waht you're looking at.


resolution charts, which you or anyone else can look at.

<http://www.dpreview.com> has them for every camera they reviewed.

> >> >> But there's plenty of things that can affect
> >> >> an image quality other than resolution. They're probably faltering at
> >> >> those other things.
> >> >
> >> >and they do. foveon has a lot of problems and it doesn't solve anything
> >> >that needs to be solved either. it's one of the biggest scams.
> >>
> >> There's yet another area you're confused about. Either that or you
> >> just don't care enough about the problems with bayer cfa to worry
> >> about them. Floyd's paper described very well some of the problems of
> >> the bayer sensor that Foveon solves. You just choose to discount it,
> >> which explains a lot.

> >
> >first of all it wasn't floyd's paper, it was his *link* to someone
> >elses paper. you're so confused you can't even get *that* right.

>
> Well, since you were the first one to call it "floyds paper", I went
> with that so you would know what I was reffering to. I admit I was
> wrong for not correcting you on yet something else you were wrong
> about.


i never called it floyd's paper. you are making stuff up. again.

> >second, foveon doesn't solve a problem that needs to be solved. humans
> >can't see the additional chroma resolution that foveon has. it's a
> >complete waste. bayer is a *much* better solution for human viewing.

>
> Yet another thing you are confused about. You seem to think that human
> vision has anything at all to do with aliasing artifacts of different
> sensor technology. The only place where human vision is involved is in
> seeing the artifacts, which has nothing to do with chroma resolution
> of the human eye and has everything to do with spatial resolution of
> the sensor.


i didn't say anything about aliasing.

i'm talking specifically about chroma resolution and that humans can't
see it very well, so the lone advantage of foveon is a waste.

> We already know what the overall resolution of the human eye is, and
> not just the chroma resolution. Apple even trademarked a term for it,
> a retina display. So according to your specious argument, a photo
> sensor doesn't need to have any more resolution than a retina display.
> And that's total nonsense.


it's not nonsense. there is no point in capturing what you can't see.

> >foveon might sound good on paper, but the reality is, humans can't see
> >any difference and it's full of shortcomings they *can* see, such as
> >noise, blotching, inaccurate colour and being limited to low iso. the
> >cameras are slow, buggy and don't have anywhere near the features of
> >competing cameras.

>
> I agree with the other limitations of Foveon. But they *do* have
> better resolution if you consider they have the ability to produce a
> color image where a higher spatial frequency can be captured without
> alias artifacts than a bayer sensor of equivalent pixel spacing. And
> *that* is the only point I'm arguing and it's a correct one as much as
> you and Wolfy want to deny it.


it's wrong, no matter how much you keep yapping.

> >the entire foveon scam is nothing more than leave off the antialias
> >filter for lots of false (and higher frequency) detail that gives the
> >illusion of higher resolution, then crank the sharpening and boost
> >contrast, resulting in an image that has a lot of snap to it and makes
> >people say 'whoa', but an image with a very inaccurate representation
> >of the original subject.

>
> Bayer also gives a very inaccurate representation of the original
> subject.


nonsense. bayer is extremely accurate, much more so than foveon is.

> If the rest of the camera outresolves the sensor, you either
> have to blur it with a strong AA filter or put up with color
> artifacts. Neither of which faithfully represents the original subject
> any more or less than a Foveon. And for both of them, if the camera
> does not outresolve the sensor, you don't need an AA filter. But the
> Foveon can reach that point sooner than the bayer because for the same
> overall pixel density, the foveon has better full color image spatial
> resolution.


nonsense.
 
Reply With Quote
 
Wolfgang Weisselberg
Guest
Posts: n/a
 
      05-14-2012
Alfred Molon <(E-Mail Removed)> wrote:
> In article <(E-Mail Removed)>, Wolfgang Weisselberg
> says...


>> They count their sensels. Which is only fair: Bayer does it
>> too.


> A "sensel" is not a pixel. A pixel is a separate spatial location


A pixel is a picture element. It's a (usually) separate spatial
location ... in the *result*.

> - the
> Sigma only has 15 million of it.


.... applying some definition of pixel.


It can have more pixels with digital zoom or enlargements.
(Same for Bayer).

(Sounds like hair splitting, right? Isn't, though: the resulting
image does contain more picture elements if it is enlarged.
Not more *information*, just more pixels.)

-Wolfgang
 
Reply With Quote
 
Wolfgang Weisselberg
Guest
Posts: n/a
 
      05-14-2012
TheRealSteve <(E-Mail Removed)> wrote:

> But we're talking about comparing the same thing. A 4.7MP Foveon will
> outresolve a 4.7MP Bayer. A 24MP Foveon will outresolve a 24MP Bayer.
> A 4.7MP 3 sensor system will outresolve a 4.7MP Bayer. A 24MP 3 sensor
> system will outresolve a 24MP Bayer.


So ... you're comparing the same thing. What Bayer sizes
were typical when Foveon had a "4.7 MPix" sensor? Compare
against that.

As for the 24 MPix Foveon and 24 MPix 3-sensor system versus
the 24 MPix Bayer ... imaginary cameras are always better
than real cameras, however the imaginary cameras need lots of
unobtainium.


> Whatever resolution you pick, the
> Foveon or 3 sensor system will outresolve the bayer at that
> resolution.


No, they won't. Imaginary cameras don't outresolve existing
cameras in the real world.


>>the sd1 upped foveon to 15 mp which was a bit closer to the best bayer,
>>but at a ridiculous price. you could get a *real* 40 mp camera for
>>slightly more money. nobody bought it, thus the $5000 price cut, and
>>it's *still* overpriced, just not as bad.


> But since overall sensor resolution at the same pixel spatial
> resoltion is the only point I'm comparing on, all of what you're
> saying is specious.


So basically you are yammering on and on about something that
has no relevance in the real world.

You might as well tell us that pigs fly more economically and
quieter than conventional planes.


>>actually sigma/foveon does claim they don't guess and it's yet another
>>lie. foveon guesses far more than bayer ever did. their ads show r/g/b
>>but that's not what the sensor captures. not even close.


> That part's not a lie. And it doesn't matter whether it's rgb or
> something else. As long as they capture the spectrum they require to
> produce the final color image at each pixel location,


They don't. They capture 3 values. These do not compose
a spectrum.

> they are doing
> better than bayer which captures only part of the required spectrum at
> each pixel location and "algorithmically comes up with" (PC for
> guesses) the part not captured.


So basically, Foveon has to guess spatial locations which Bayer
already has, Foveon has to guess the colour which Bayer already
has and for that reason Foveon does better.



>>looks like they fooled you. foveon does *not* capture rgb, but rather
>>it has three overlapping samples which it then tries to calculate rgb,
>>but there are so many variables that there is quite a bit of
>>uncertainty to it.


> As I said above, it doesn't matter how they break the spectrum down.


It does. That's so trivial ...

> What matters is that what's needed for the final image is sampled at
> each pixel. And that's why the resolution is better.


So a '15 MPix' Foveon is better than a, say 36 MPix Bayer.
Just comparing same with same: the current highest resolution
available in FF sensor cameras.


>>bayer measures one colour per location and accurately calculates the
>>other two, so it's already *well* ahead of the game.


> No, bayer doesn't do either of those things. Bayer also captures a
> band of wavelenghts and not strictly r/g/b at each sensel location. So
> in that regard, it's the same as Foveon.


Actually, nope. Foveon captures rather random values, Bayer
captures well defined data.

> And if Bayer *could*
> accurately calculate the missing information at each pixel, it would
> not have color artifacts.


With a correct AA filter, that's exactly what happens.

> The fact that even you admit it does have
> color artifacts proves you're wrong when you way it "accurately
> calculates the other two", and that it's not *well* ahead of the game.


'15 MPix' Foveon vs 36 MPix Bayer, or 60 MPix (medium format), ...
Well, and the colour fidelity.
Well, and the high ISO capability.
Well, and the aliasing Foveon has.

Sure, Foveon is very close to Bayer ...


> As far as my argument is concerned, Foveon was just an
> analog for a 3 sensor system anyway.


Which only shows that you haven't understood Foveon.


>>nope. foveon definitely doesn't have full colour information (see
>>above). only 3 sensor systems do and at a huge cost and weight penalty.


> Even 3 sensor systems don't necessarily have *full* color information.


Yes, even a 3-sensor system only works as a 3-stimulus
measuring instrument.


>>bayer doesn't lag behind. it's *well* ahead of foveon and laughably so.


> It does lag behind on the one point of comparison I'm making.


And airplanes lag behind on the point of price versus flying pigs.

-Wolfgang
 
Reply With Quote
 
Wolfgang Weisselberg
Guest
Posts: n/a
 
      05-14-2012
TheRealSteve <(E-Mail Removed)> wrote:
> <(E-Mail Removed)> wrote:
>>TheRealSteve <(E-Mail Removed)> wrote:
>>> On Fri, 11 May 2012 22:35:38 +0200, Wolfgang Weisselberg
>>>>TheRealSteve <(E-Mail Removed)> wrote:
>>>>> On Thu, 10 May 2012 18:09:03 +0200, Wolfgang Weisselberg
>>>>>>TheRealSteve <(E-Mail Removed)> wrote:
>>>>>>> On Tue, 8 May 2012 00:36:49 +0200, Wolfgang Weisselberg
>>>>>>>>TheRealSteve <(E-Mail Removed)> wrote:
>>>>>>>>> On Sat, 5 May 2012 22:27:42 +0200, Wolfgang Weisselberg



>>> Why should I say you didn't use marketing hype to compare when that's
>>> exactly what you did?


>>"But if you want to use their marketing hype to compare, then
>>you have to use it fully." Said you. I did. Your point?


> No, you absolutely did not. To use their marketing hype fully without
> cherry picking, you would have to compare their 45MP camera to a bayer
> 15MP camera, not to a 45MP bayer camera.


Please show me where Foveon or Sigma said in their marketing hype
that their 45 MPix cameras need to be compared against 15 MPix
Bayer cameras.


>>> Sigma's defiition isn't the same definition as
>>> everyone else's.


>>Sigma's *SENSOR* isn't the same sensor as everyone else's.
>>So of course it hast to be counted in other ways.


> No it doesn't *have* to be counted in other ways.


OK, than I shall count 4 Bayer sensels as one pixel.
So the 45 MSensel ("15 MPix" according to you) Foveon has to be
compared to a 60 MSensel ("15 MPix" as I shall count it) Bayer ---
and Bayer clearly wins again.

That should make you happy now.

> They do count it in
> other ways to fool people like you.


*I*'m not fooled, I simply adhere to your "terms of the physical
reality of the image being captured".

You're the one who is inconsistent.


>>> They triple it for marketing hype.


>>They count their sensels. Which is only fair: Bayer does it
>>too.


> And they have 3 sensels at each pixel location while bayer has only
> one. So?


So you need to compare sensels to sensels.

> All that means is that they're sampling everything they need
> to generate a full color pixel at each location. A bayer sensor does
> not.


So basically Bayer sensor images do not consist of RGB triples.
Interesting.

And Foveon doesn't have any colour problems. Interesting.

Hook, line *and* sinker.


>>> You simply *do* compare different
>>> definitions of pixels as it relates to what goes into the image being
>>> captured and presented.


>>I simply *do* compare different sensor types --- i.e. what
>>relates to what goes into the image being captured.


> Too bad that's not what you're doing.


You think so? Just because I don't agree to your irrelevant,
specious, never applicable in real life claims?


>>I don't care about the presentation one bit, since that
>>depends on printers, paper, processes ... which are simply
>>not comparable.


> Ah, but those things are also comparable as long as you define
> consistent parameters for comparison...


Oh! Consisten parameters! Let's take ... opacity ... and
flourescent whitening agent of the paper. Yeah, that should
make everything comparable. Just as your spasmodic tries to
compare sweet water fish to sunlight. After all, both are
starting with an 's' and are written with ASCII symbols.

> something you are unable to do.


Oh, I just use comparable parameters, unlike you.


>>>>> If a camera manufacturer defined a
>>>>> pixel as the full bucket count of electrons that could be captured at
>>>>> each sensel location, you would believe them when they said they have
>>>>> a 10000GP sensor and would use that marketing tripe to compare to a
>>>>> bayer 10000GP sensor.


>>>>Nope. I would merely notice that with a TheRealSteve sensor 25GP
>>>>would equal about 1 MP on a Bayer pattern sensor.


>>> Lol, which is exactly what you *should* be doing with the Foveon vs.
>>> Bayer.


>>This is what I am doing.


> But it's not what you did do when you tried to compare a 45MP "sigma
> pixel" camera with a 45MP bayer.


Correct.
You are unable to say straight how many Bayer pixels a '45MP
"sigma pixel" camera' compares to. The only thing you are
interested in is somehow claiming "Foveon is better than Bayer",
no matter how you have to distort reality to get there. I
merely try to insert *some* sanity into your claims.


> As above, you should simply notice
> that with a Sigma sensor, the 45MP would equal about 15MP on a bayer
> pattern sensor. Yet one more way you are inconsistent.


So you say that they are equivalent? Fine. No more "Fovepn
is better". That we can live with.


>>> Sigma's definition of a 45MP sensor would equal about 15MP on a
>>> Bayer pattern sensor.


>>WRONG! Sigma's definition of a 45MP sensor would equal about 21
>>MP on a Bayer pattern sensor. You're falling for Bayer pattern
>>marketing hype and don't even notice it. (And for the record,
>>nospam's wrong about that, but he could easily find out what
>>frequencies (measured in pixel lengths) appear in real world
>>images.)


> Ah, so you admit you're wrong when you were trying to compare a 45MP
> Foveon with a 45MP bayer.


Let me rephrase that:
Sigma's definition of a 45MP sensor would equal about 21 MP
on a Bayer pattern sensor in resolution.

> Excellent. And whether you want to say 21
> because of pixel pitch or 15 because of 3 sensels vs. 1 sensel = a
> pixel is immaterial. My only argument is that it's less, which you
> finally agree with.


The number 45 is bigger than the number 21.
And *your* claim was that it equals a 15 MPix Bayer sensor in
resolution ...

>>> if you use the same definition of
>>> pixels.


>>You don't. That's why you think Foveon is better.


> Obviously you think so also if, using your comparison, two sensors
> with equivalent pixel pitch would work out to 45MP for a Foveon and
> 21MP for a bayer.


Huh? Eqivalent pixel pitch? What? Who?

> Whether or not that makes sense doesn't matter
> because it's your definition.


If you don't even understand what others are saying, no
wonder you're terribly confused.


>>>>No, I admit to using the same definition. "As used by the people
>>>>who make and market such cameras".


>>> "As used by the people who make and market such cameras" leads to
>>> different definitions because one manufacturer who makes and markets
>>> such cameras uses a different definition than everyone else.


>>They also use a different sensor than everyone else.


> And their sensor has better color spatial resolution for the same
> spatial sampling rate because they sample what they need for a color
> image at each location while a bayer does not.


And some sensors are UV capable or x-ray capable. That's however
not interesting for displaying things as the eye sees it.
Just as a "better color spatial resolution". In fact, there
*are* Bayer sensors that outrank the colour resolution of the best
Foveon sensors.

So, yes, you're right, but your claim is irrelevant. The ...
ah, _suboptimal_ colour rendition and blotching with Foveon are
much more relevant (and real drawbacks) in real life.

> So yes, it's different.


And, as we have learned, worse in real life.


> And whatever definition you use for a color pixel in order to compare
> them had better show that difference or else it's not a valid
> comparison.


Any comparison *you* make "in terms of the physical reality of
the image being captured" better shows the difference in sensels
per spatial location or else it's not a valid comparison. Up to
now I've seen a lot of invalid comparisons from you.


> If you choose to use Sigma's simplification for marketing
> hype because you don't understand sampling theory, fine. Then use it.
> But if you do understand what's going on and define your pixels
> equivalently, then you'd see that for the same pixel count, the Sigma
> won't alias as much at the same input spatial frequencies below
> nyquist. If you go *above* nyquist, then all bets are off.


The Sigma will alias *more*. No AA filter, more alias (unless
the sigma lenses are even worse than coke bottle bottoms).

You seem to keep forgetting that. Or you seem to imply that there
are relevant scenes where there are no high frequency contents.


>>> But you
>>> have such a fetish for nya nya nya (your terms) that you can't see
>>> that.


>>You must be really upset I found out about *your* fetish that
>>you try to hang it on me.


> You're projecting again.


You learned a new word from me.


>>>>> The fact
>>>>> that you know you're doing something wrong just to say nya nya nya
>>>>> shows your fetish.


>>>>The fact that I am even talking to a crackpot like you is
>>>>indeed proof that I do something wrong.


>>> The fact that you think I'm the crackpot when you're the one who can't
>>> even keep the definition of a pixel straight and falls for simple
>>> misleading marketing tripe is indeed proof that you're the crackpot.


>>Yes, everyone is a crackpot but you.


> No, just you.


No, just you.


>>>>>>>>> And using their marketing hype
>>>>>>>>> fully, you would have to compare the 45MP Foveon to a 15MP Bayer since
>>>>>>>>> both are counting sensel locations.


>>>>>>>>Why would I have to do that?
>>>>>>>>Sensels and locations aren't even used in the marketing hype.
>>>>>>>>Only megapixels.


>>>>>>> Exactly! And that's why you can't compare megapixels between Sigma's
>>>>>>> definition for Foveon and Bayer.


>>>>>>Of course you can!
>>>>>>It just destroys your nya nya nya.


>>>>> You really have a hard on about nya nya nya.


>>>>It's you who masturbates on Bayer and Foveon pixels, and the
>>>>latter being "better" in some way. I really don't care about that,
>>>>since Foveon is *way* worse for what I do and need. Which is
>>>>a lot of very high ISO.


>>> Foveon is better in some way.


>>Yes, it gives money to Sigma if you buy one.


> Yes, and it has better resolution for the same pixel spacing.


Yes, Sigma doesn't use an AA filter.
So it also has way more false resolution for the same pixel
spacing.


>>The day you manufacture and sell cameras with your own sensor
>>type is the day you can define how to count your pixels.
>>I guess hell freezes over first, though.


> I see. So we have to take your definition because you manufacture and
> sell cameras with your own sensor. All hail the great stupid one.


I don't define how to count pixels. You try to do that
however. Hail TheRealSteve who doesn't even have a proper
name, nor a working email adress. Probably doesn't want his
parents to see how he behaves.


>>> You've already proven to yourself that your wrong when your own
>>> argument was used against you and you treated it with sarcasm.


>>Yep, that was the day I invented my own definition of pixels.


>>What was it again? Ah, yes, a camera has 100.000.000/IQ of the
>>user pixels. Which would mean you get all the 50+ MPix cameras,
>>and normal people like me have to make do with roughly 1 MPix.


> Original, but specious.


Let me translate 'specious' for the othe readers:
| TheRealSteve agrees I didn't invent my own definition of pixels,
| but he did. So the argument worked against him, but not against
| me. Thus I didn't prove myself I am or was wrong.
|
| But lil Stevie cannot ever admit he was wrong.


>>Oh, that's as stupid as your own definition. And completely
>>useless, since you're on the physics on how the image is
>>recorded trip. Which means you need to count sensels.


> That just shows you don't know what spatial sampling rate means.


I do know, do you?

> You
> can have 100 sensels all stacked up on eachother (or on 100 separate
> sensors that are aligned in image spatial space) and it's still
> capturing only one spatial location.


The location in any way is an area, not a point 'location'.

Stacked like Foveon each lower layer is larger than the layer
above it. On a 100-sensor system, the alignment *in reality*
is a misalignment, so again you are sampling a larger area.

> When you want to compare spatial
> resolution, you have to compare spatial sampling locations and not
> sensels.


Actually, the physics on how the image is recorded is that
it's not recorded with spatial locations but with sensels.

Sometimes the sensels are arranged in a grid pattern.

> You're so hung up on trying to understand what pixels are and
> Sigma vs. bayer definitions that you miss the spatial sampling
> fundamentals of capturing the image in the first place.


You're so hung up on "Foveon is better than Bayer" that you
don't even understand what pixels are.

For example, if you upsize an image coming from a 45 MSensel
Foveon sensor to 20 M spatial locations --- and don't say you
cannot do that --- how many pixels does the result have?


>>> And since it's *locations* that matter
>>> when it comes to spatial resolution and sampling an image,


>>Assuming identical resolution from lenses, assuming identical
>>beam splitting from AA filters, etc --- all of which is not
>>given --- ...


> YES, assuming everything else is identical. YES, it's not given
> everything else will be identical. But if you want to compare the
> sensors, you make everything else the same. Or didn't they teach you
> that in high school science class.


AH! THAT explains your fixation of 'same pixel counts by
TheRealSteve's definition'. You didn't grasp that to compare
real life systems, you have to use real life assumptions.
Like "what sensors were available in 2010". Like "missing AA
filters" when that's what's in real life.

But yet you are completely resistant to "make everything the
same: sensel count". How comes?

>>> the fact
>>> that the bayer has fewer locations that capture each color


>>Foveon doesn't capture colour at all. Which is why it has so
>>many problems.


>>The upper layer of a Foveon sensor captures all frequencies, but
>>not all of the photons that are actually converted to electrons ,
>>only some. How much it captures depends e.g. on random happenings
>>of manufacturing of that pot. The lower pots also are larger,
>>and thus capture again all frequencies, but not all of the
>>photons (which are actually converter), in a slightly different
>>composition. And so on.


>>So Foveon captures 3 sets of luminosity --- however, these sets
>>are not panchromatic or even orthochromatic. There's no "this
>>is blue" and "this is green" like in a Bayer sensor.


>>There's 3 very noisy signals that are guessed into some
>>approximation of some RGB stimuli. Sometimes the guesses are
>>quite right, sometimes they are way off.


> Yes, Foveon has problems. In fact, I don't even know how we got onto
> Foveon in the first place when I was only comparing 3 sensor systems.
> The only point of comparison I'm making regarding Foveon is that it
> does physically sample everything it needs at each photosite


No, it doesn't. If it got all that, Foveon wouldn't have to
guess and too often guesses wrong. Bad colour fidelity,
blotching ... Proof that Foveon does *not* record enough.
OK, there's a _slight_ chance that the state of the art of
software and algorithms isn't good enough.

> while a bayer does not.


And yet Bayer sensor cameras manage to capture enough that
each photosite gets assigned an RGB triple --- with results
that are *better* than what Foveon manages.

How comes?

> That's really the only way it's comparable to a 3
> sensor system and it's the only point of comparison I'm making.


A 3-sensor system has too many drawbacks in the wild.


>>> and then
>>> tries to merge these under-sampled locations into a final image is why
>>> it has the problems it has.


>>Bayer doesn't have any relevant problems that can be solved
>>by Foveon.


> Relevant is a relative term. What's not relevant to you might be to
> someone else.


Name one relevant problem.


>>> Yes, they have 3x the sensels for the same pixel density. But not 3x
>>> the pixels.


>>Since *you* want "the physical eality of the image being captured"
>>you *have* to count sensels. No matter if you decide that
>>they have or have not 3x the pixels.


> No, I am counting the spatial resolution of everything that goes into
> making the final image.


So where's the MTF curve of the lens? That surely goes into
making the final image. And where's the upscaling, downscaling,
cropping and rotating (by a few degrees)? That does go into the
final image. Where's the retouching and the denoising?

All that does go into the final image ... and yet you completely
ignore it.

> It doesn't matter whether it's sensels, pixels
> or whatever else you want to come up with.


Yep, in that case it doesn't matter, since a lot of other
changes are being made to the resolution.

>>I'll not even go into luminance carrying way more information than
>>chrominance for the human eye and into the fact that chrominance
>>is *way* *way* *way* undersampled in the human eye, and is placed
>>in random locations on a 2D surface, not stacked like Foveon.
>>Such details would only give you a migraine attack.


> None of which, of course, matters when it comes to aliasing artifacts.


So I am perfectly right and you don't have any counter
argument.

Funny that you should mention aliasing artifacts, since Foveon
cameras don't have AA filters ... and presto -> Aliasing!

>>Actually, nope. Bayer counts electrons (-> photons) passing
>>through a (broad) bandpass after an AA filter. I.e. Bayer
>>captures luminosity values after a bandpass. This is not the
>>same as "one color". Lay people may think this, however.


>>> It has to algorithmically figure out the rest.


>>Just as Foveon has to algorithmically figure out all 3
>>colours (and does not manage well).


> Which, again, is a different argument not related to alias artifacts.


So I am perfectly right and you don't have any counter
argument.


>>>>2) You're the one who wants "the physical reality of the
>>>> image being captured" ...


>>> Which is exactly what I'm doing.


>>... has to COUNT SENSELS.


> When it comes to spatial resolution, it absolutely does not have to
> count sensels. Counting sensels gives you the wrong answer.


So spatial resolution with Foveon is as good as with Bayer,
by your claim of pixel counting.

But then you're moving goalposts from spatial resolution to colour
resolution and back to spatial resolution, as you are continously
being beat.


>>> The Bayer cfa does not. The physical reality is
>>> that each color channel of the Foveon and 3 sensor systems are sampled
>>> at the full spatial resolution of the sensor.


>>Only true for 3-sensor systems, but there the sensels don't
>>align.


> Good, then lets go back to only 3 sensor systems,


And you're shifting goalposts from 3-sensor to Foveon back
to 3-sensor.

> which is what I was
> originally talking about before someone else brought up Foveon. The
> only way Foveon is equivalent is that it captures what it needs to
> create a color image at every spatial photosite.


So does Bayer. Oh, not at a single isolated spatial photosite.

But there aren't any photos consisting of one pixel. Even small
images are at least a few hundred by a few hundred pixels.
And most sensors contain at least a million photosites. Heck,
the prototype sensor had 100x100 spatial photosites.

So the "at a single isolates spatial photosite" claim is obviously
not relevant for anything but wanking.


>>> They are not sampled at
>>> that full rate for the Bayer cfa.


>>And still Bayer produces perfectly good RGB triples ...


> No it does not.


Yet it does. Otherwise there wouldn't be colour photography by
Bayer pattern sensors. There are, of course, a few suboptimal
implementations.

> There are color artifacts that would not be there if
> it produced "perfectly good" rgb triples.


There are also colour artifacts --- worse colour artifacts
--- with Foveon. So Foveon produces worse RGB triples than
Bayer ... even though you claim it gets everything in each
single spatial location.


>>> The bayer cfa has plenty of advantages over both Foveon and 3-sensor
>>> systems. But this whole discussion has been focusing on only
>>> resolution and preventing aliasing artifacts and that's probably the
>>> only areas that the Bayer lags behind the other two.


>>Bayer has more resolution per sensel than the other two. So
>>it does not lag behind. As for aliasing artifacts, they are
>>worse with Foveon --- no AA filter.


> Now you're not keeping things equal.


What? Sensel for sensel ... that's equal.

> That figures, since you don't
> think you have to.


What's equal is for you to decide?
Interesting.
Who died and made you king?

-Wolfgang
 
Reply With Quote
 
 
 
Reply

Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
Re: The death of the Bayer filter? Maybe not. Bruce Digital Photography 3 04-04-2012 08:04 PM
Death, even when expected, is hard for anyone to deal with, but asuicidal death asim malik Perl Misc 1 09-13-2009 07:42 AM
Re: Crocodile Hunter - DEATH VIDEO - Steve Irwin - crocodile hunter - death video - steve irwin.exe (1/1) Mikey DVD Video 3 09-14-2006 05:52 PM
VoIP ebay & skype = DEATH (Offtopic ? Maybe) codefixer@gmail.com Cisco 0 09-08-2005 09:27 PM
Of Intrest ? Maybe Yes, Maybe No Old Gringo Computer Support 0 10-11-2004 01:35 AM



Advertisments