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
 
      04-15-2012

On Sun, 15 Apr 2012 08:06:16 -0800, http://www.velocityreviews.com/forums/(E-Mail Removed) (Floyd L.
Davidson) wrote:

>TheRealSteve <(E-Mail Removed)> wrote:
>>On Sat, 14 Apr 2012 16:37:15 -0800, (E-Mail Removed) (Floyd L.
>>Davidson) wrote:
>>
>>>TheRealSteve <(E-Mail Removed)> wrote:
>>>>On Sat, 14 Apr 2012 21:20:26 +0200, Alfred Molon
>>>><(E-Mail Removed)> wrote:
>>>>
>>>>>In article <(E-Mail Removed)>, TheRealSteve
>>>>>says...
>>>>>> You can't have it both ways and flip flop between them when it suits
>>>>>> you.
>>>>>
>>>>>Flip flopping is good. It shows that you are not stuck religiously on
>>>>>one position, and are willing to accepts the other side's arguments if
>>>>>they are valid.
>>>>
>>>>The problem is he's arguing against himself. Shooting down his own
>>>>point in a later post doesn't lend much creedance to the point he was
>>>>making.

>
>>>
>>>The problem is that I understand what the benefits are, *and* what
>>>the negatives are too. Knowing both is not flip flopping and it is
>>>not arguing against oneself. It using you head!

>>
>>Excellent. So you then are agreeing that there is benefit to not
>>having an AA filter. Unless you want to flip flip again. lol

>
>Of course there is *a* benefit to not having an AA filter.
>There is also *a* benefit to having an AA filter.
>
>The question isn't is there benefit to either configuration, but
>which one is more significant in any given situation.


Thanks, thats all I was trying to say.

>That's where almost all of the arguments for removing the AA
>filter fail, because there is only a very narrow range of uses
>where such a camera will be more useful than an identical camera
>with an AA filter. Of course for the person doing exactly that


That used to be true. Not so much anymore when you're talking about
36MP and above cameras. Which, of course, is where this whole spat
started ... with the release of the two versions of the D800e.

>kind of work it makes no difference how unique they are!


I guess the uniqueness of the person is manifested by the fact they
believe they need 36MP or higher. When you get into that realm, having
an AA filter hurts way more than it helps. Which is exactly why very
high end cameras don't have them. It's certainly not due to cost for
those cameras.

Steve
 
Reply With Quote
 
 
 
 
TheRealSteve
Guest
Posts: n/a
 
      04-15-2012

On Sat, 14 Apr 2012 21:24:34 -0400, nospam <(E-Mail Removed)>
wrote:
[...]
>wrong. having 3 sensors increases chroma resolution (which is already
>more than you can see), not luma resolution which remains the same and
>is what matters.


That statement shows just how confused you are and why it's not worth
trying to explain to you why you're wrong anymore. Here's a few things
you need to brush up on before it's worth continuing:

First and most of all is sampling theory and aliasing. Then you'll
understand why the resolution of what you can see means absolutely
nothing. You'll learn that spatial resolutions far and above what you
can *see* are translated by aliasing into artifacts which you can
easily see. So the "resolution of the human eye" argument is
completely specious.

Second, you need to look at the pattern of a bayer sensor. You need to
understand how the substrate below the color filter has a liner
resolution twice that of the green channel and 4 times that of the red
and blue channels.

You need to understand how sampling at 1/2 the rate for green and at
1/4 the rate for red and blue greatly increases the odds of getting
aliasing into the individual color channels over and above a
monochromatic sensor with the same underlying pixel density as the
bayer sensor's substrate under the color filter.

You need to understand how, if any one of the color channels is
aliased then the resultant de-mosaiced image will have artifacts. And
those artifacts could be visible to the human eye even if the input
signal that caused the aliasing was not visible to the human eye.

And finally, you need to understand how a 3 sensor camera system works
as opposed to a bayer filter. How a 3 sensor system uses 3
monochromatic sensors, each of which samples it's assigned color at
the full rate of the substrate under the filter while a bayer sensor
samples each of it's assigned colors at only 1/2 or 1/4 the rate of
the substrate under the filter.

Once you understand those facts and learn about sampling. you may be
able to de-confuse yourself. But only if you can connect those facts
above and what you learn about sampling and aliasing in a coherent
way. We'll see how successful you are. If you can't then maybe someone
will connect the dots for you. But for now, it's obvious you don't
understand the basic underlying facts and/or haven't connected them
into a coherent thought yet.

Good luck and happy learning.

Steve
 
Reply With Quote
 
 
 
 
nospam
Guest
Posts: n/a
 
      04-15-2012
In article <(E-Mail Removed)>, Alfred
Molon <(E-Mail Removed)> wrote:

> > But let me just point out the one single most glaring flaw in
> > you whole argument: For any sensor with X number of sensor
> > locations, a "3 sensor" configuration produces a maximum pixel
> > resolution of X/3, but a Bayer Configuration produces X/1
> > pixels. Due to the Bayer pattern, as has been discussed, the
> > chroma resolution is lower that X/1, but it is higher than X/3,
> > and the luminance resolution equals the pixel resolution. Which
> > is to say that Bayer is going to out perform your "3 sensor"
> > configuration in terms of resolution.

>
> Floyd, do you realise that in a Bayer sensor you do not have luminance
> in every pixel? The green pixels may be a not too terrible approximation
> of luminance, but the red and green ones are a very bad approximation of
> luminance.


absolutely wrong.
 
Reply With Quote
 
Ryan McGinnis
Guest
Posts: n/a
 
      04-16-2012
On Thu, 12 Apr 2012, RichA wrote:

> On Apr 12, 6:29*am, Mxsmanic <(E-Mail Removed)> wrote:
>> Martin Brown writes:
>>> Rubbish. They don't have as many settings but the video quality is
>>> pretty good on the Canon 100is and Pentax D-5 that I have.

>>
>> Compare them side by side with video cameras.

>
> Rolling shutter cameras will always be inferior, even the Canon 5DIII
> and the GH2.


Indeed, but if it's "good enough" for 1080p broadcast television
productions, I'd say it's good enough for most applications -- and the
kinds of people that it's not good enough are probably using 2K or 4K cine
cameras that are well out of the budget of someone who was considering a
DSLR as a movie camera.


Ryan McGinnis * * * * * *(E-Mail Removed) * * * * ** *@bigstormpicture
Follow my storm chasing adventures at http://bigstormpicture.blogspot.com
Images@Getty: http://bit.ly/oDW1pT * * * *Images@AGE http://bit.ly/w4EuWB
The BIG Storm Picture: *http://bigstormpicture.com * *PGP Key: 0x65115E4C
 
Reply With Quote
 
nospam
Guest
Posts: n/a
 
      04-16-2012
In article <(E-Mail Removed)>, Mxsmanic
<(E-Mail Removed)> wrote:

> Three seprate sensors of resolution x produce a final resolution of x, for
> both luminance and chroma.


true.

> A single sensor with resolution x and a Bayer filter produces green resolution
> of slightly over x/2, red resolution of slightly over x/4, blue resolution of
> slightly over x/4,


x/2 + x/4 + x/4 = 1

> and luminance resolution of somewhat less than x.


it is not somewhat less. it's the same (actually, slightly and
insignificantly less).

> All
> luminance values are educated guesses, as are 1/2 the green values and 3/4 of
> the blue or red values.


they are not guesses. they are calculated and the error is very tiny,
so much so that it can be ignored in all but edge cases.

> Thus, for any given sensor resolution, three sensors perform far better than
> one with a Bayer filter.


it has twice the chroma resolution than bayer but you can't see the
difference so it doesn't matter.
 
Reply With Quote
 
nospam
Guest
Posts: n/a
 
      04-16-2012
In article <(E-Mail Removed)>, TheRealSteve
<(E-Mail Removed)> wrote:

> >wrong. having 3 sensors increases chroma resolution (which is already
> >more than you can see), not luma resolution which remains the same and
> >is what matters.

>
> That statement shows just how confused you are and why it's not worth
> trying to explain to you why you're wrong anymore. Here's a few things
> you need to brush up on before it's worth continuing:


i'm not confused, but you clearly are.

> First and most of all is sampling theory and aliasing. Then you'll
> understand why the resolution of what you can see means absolutely
> nothing. You'll learn that spatial resolutions far and above what you
> can *see* are translated by aliasing into artifacts which you can
> easily see. So the "resolution of the human eye" argument is
> completely specious.


you don't need to resolve more than what you can see. you do have to
avoid aliasing though.

> Second, you need to look at the pattern of a bayer sensor. You need to
> understand how the substrate below the color filter has a liner
> resolution twice that of the green channel and 4 times that of the red
> and blue channels.


think luma/chroma, not r/g/b channels.

> You need to understand how sampling at 1/2 the rate for green and at
> 1/4 the rate for red and blue greatly increases the odds of getting
> aliasing into the individual color channels over and above a
> monochromatic sensor with the same underlying pixel density as the
> bayer sensor's substrate under the color filter.
>
> You need to understand how, if any one of the color channels is
> aliased then the resultant de-mosaiced image will have artifacts. And
> those artifacts could be visible to the human eye even if the input
> signal that caused the aliasing was not visible to the human eye.


bayer doesn't work that way.

> And finally, you need to understand how a 3 sensor camera system works
> as opposed to a bayer filter. How a 3 sensor system uses 3
> monochromatic sensors, each of which samples it's assigned color at
> the full rate of the substrate under the filter while a bayer sensor
> samples each of it's assigned colors at only 1/2 or 1/4 the rate of
> the substrate under the filter.


you're very, very confused.

luminance is sampled at full resolution (same as a monochrome sensor of
the same pixel count) and chroma is sampled at half that.

> Once you understand those facts and learn about sampling. you may be
> able to de-confuse yourself.


i'm not confused.

> But only if you can connect those facts
> above and what you learn about sampling and aliasing in a coherent
> way. We'll see how successful you are. If you can't then maybe someone
> will connect the dots for you. But for now, it's obvious you don't
> understand the basic underlying facts and/or haven't connected them
> into a coherent thought yet.


except, i have and long ago.

> Good luck and happy learning.


same to you. you need it far more than i do.
 
Reply With Quote
 
TheRealSteve
Guest
Posts: n/a
 
      04-16-2012

On Sun, 15 Apr 2012 11:28:37 -0800, (E-Mail Removed) (Floyd L.
Davidson) wrote:

>TheRealSteve <(E-Mail Removed)> wrote:
>>On Sun, 15 Apr 2012 08:06:16 -0800, (E-Mail Removed) (Floyd L.
>>Davidson) wrote:
>>
>>>TheRealSteve <(E-Mail Removed)> wrote:
>>>>On Sat, 14 Apr 2012 16:37:15 -0800, (E-Mail Removed) (Floyd L.
>>>>Davidson) wrote:
>>>>
>>>>>TheRealSteve <(E-Mail Removed)> wrote:
>>>>>>On Sat, 14 Apr 2012 21:20:26 +0200, Alfred Molon
>>>>>><(E-Mail Removed)> wrote:
>>>>>>
>>>>>>>In article <(E-Mail Removed)>, TheRealSteve
>>>>>>>says...
>>>>>>>> You can't have it both ways and flip flop between them when it suits
>>>>>>>> you.
>>>>>>>
>>>>>>>Flip flopping is good. It shows that you are not stuck religiously on
>>>>>>>one position, and are willing to accepts the other side's arguments if
>>>>>>>they are valid.
>>>>>>
>>>>>>The problem is he's arguing against himself. Shooting down his own
>>>>>>point in a later post doesn't lend much creedance to the point he was
>>>>>>making.
>>>
>>>>>
>>>>>The problem is that I understand what the benefits are, *and* what
>>>>>the negatives are too. Knowing both is not flip flopping and it is
>>>>>not arguing against oneself. It using you head!
>>>>
>>>>Excellent. So you then are agreeing that there is benefit to not
>>>>having an AA filter. Unless you want to flip flip again. lol
>>>
>>>Of course there is *a* benefit to not having an AA filter.
>>>There is also *a* benefit to having an AA filter.
>>>
>>>The question isn't is there benefit to either configuration, but
>>>which one is more significant in any given situation.

>>
>>Thanks, thats all I was trying to say.

>
>But you didn't; and I did. Hmmmm...


If you look back though the history of many of these threads you'll
see that I've said exactly that is the beginning of each. It's only
when the posters who claim doom and gloom if there's no AA filterr
that I'm forced to defend that one side.

>
>>>That's where almost all of the arguments for removing the AA
>>>filter fail, because there is only a very narrow range of uses
>>>where such a camera will be more useful than an identical camera
>>>with an AA filter. Of course for the person doing exactly that

>>
>>That used to be true. Not so much anymore when you're talking about
>>36MP and above cameras. Which, of course, is where this whole spat
>>started ... with the release of the two versions of the D800e.

>
>I don't think a thing has changed due to the D800(E) having 36MP.
>That is far from the pixel density that will be needed to make
>an AA filter unnecessary. Something between 175 and 250 MP...


Now you're saying an AA filter *is* necessary below 175MP when just
before you were claiming to see both sides. Hmmmmm. Interesting that
the manufacturers and users of high end cameras don't agree with you.

>
>>>kind of work it makes no difference how unique they are!

>>
>>I guess the uniqueness of the person is manifested by the fact they
>>believe they need 36MP or higher.

>
>No, because they can have either 36 MP with or without the AA filter.
>
>>When you get into that realm, having
>>an AA filter hurts way more than it helps.

>
>Bullshit son. It just ain't so. The amount of aliasing
>distortion spread all through the recorded spectrum is far worse
>that the very slight loss if SNR at spatial frequencies very
>close to the Nyquist Limit. Or, that is true for most types of
>photography, and not true for a relatively limited set of
>circumstances.


And you womder why I need to only show defense of not having an AA
filter when you make outlandish statements like that which bely your
assertion that you can "see both sides."

>
>>Which is exactly why very
>>high end cameras don't have them. It's certainly not due to cost for
>>those cameras.

>
>There are several reasons that some cameras don't have them.
>Price probably has virtually nothing to do with it. Marketing
>probably has virtually everything (as evidence by the ability to
>convince people like you that technical facts are irrelevant).


I'm sure the users of Phase One, Leica, Mamiya and Hasselblad are all
easily fooled by marketing and that they all just don't "get it" like
you do. Their images are all trash, filled with aliasing artifacts but
no one who sees the images notice the artifacts because they are
blinded by marketing of those high end cameras. Yup, that must be it.
That's your story and you're sticking to it?
 
Reply With Quote
 
TheRealSteve
Guest
Posts: n/a
 
      04-16-2012

On Sun, 15 Apr 2012 11:40:27 -0800, (E-Mail Removed) (Floyd L.
Davidson) wrote:

>TheRealSteve <(E-Mail Removed)> wrote:
>>On Sat, 14 Apr 2012 21:24:34 -0400, nospam <(E-Mail Removed)>
>>wrote:
>>[...]
>>>wrong. having 3 sensors increases chroma resolution (which is already
>>>more than you can see), not luma resolution which remains the same and
>>>is what matters.

>>
>>That statement shows just how confused you are and why it's not worth
>>trying to explain to you why you're wrong anymore. Here's a few things
>>you need to brush up on before it's worth continuing:

>
>He's going to eat your lunch on that one...
>
>But let me just point out the one single most glaring flaw in
>you whole argument: For any sensor with X number of sensor
>locations, a "3 sensor" configuration produces a maximum pixel
>resolution of X/3, but a Bayer Configuration produces X/1
>pixels. Due to the Bayer pattern, as has been discussed, the
>chroma resolution is lower that X/1, but it is higher than X/3,
>and the luminance resolution equals the pixel resolution. Which
>is to say that Bayer is going to out perform your "3 sensor"
>configuration in terms of resolution.
>
>Where Bayer is not as good is color accuracy. But obviously for
>most photography the absolute accuracy of colors is not at all
>important.


The obvious flaw in your argument is that 3 monochromatic sensors of
the same size and substrate pixel density as your single bayer sensor
has 3x the sensels of the bayer sensor. So the max pixel density is
not X/3, but is X. everything from then on is flawed because it flows
from that piece of flawed information. The luminance resolution, while
it equals the "pixel" resolution, a "pixel" in the bayer sensor is
made up of 4 sensels, which is why it's 1/2 the linear resolution of
the substrate under the filter. A pixel of a 3 sensor system is made
up of 3 sensels, one on each of the 3 sensors. And the sensel
resolution is the same as the substrate under the filter, which is
twice the luminance linear resolution of the bayer sensor.

Steve
 
Reply With Quote
 
TheRealSteve
Guest
Posts: n/a
 
      04-16-2012

On Sun, 15 Apr 2012 17:04:37 -0800, (E-Mail Removed) (Floyd L.
Davidson) wrote:

>Alfred Molon <(E-Mail Removed)> wrote:
>>In article <(E-Mail Removed)>, Floyd L. Davidson says...
>>> But let me just point out the one single most glaring flaw in
>>> you whole argument: For any sensor with X number of sensor
>>> locations, a "3 sensor" configuration produces a maximum pixel
>>> resolution of X/3, but a Bayer Configuration produces X/1
>>> pixels. Due to the Bayer pattern, as has been discussed, the
>>> chroma resolution is lower that X/1, but it is higher than X/3,
>>> and the luminance resolution equals the pixel resolution. Which
>>> is to say that Bayer is going to out perform your "3 sensor"
>>> configuration in terms of resolution.

>>
>>Floyd, do you realise that in a Bayer sensor you do not have luminance
>>in every pixel? The green pixels may be a not too terrible approximation
>>of luminance, but the red and green ones are a very bad approximation of
>>luminance.

>
>You don't want to say things like that in polite company, because people
>will snicker.
>
> 1) Most, but not hardly all, of the luminance is recovered
> from the green sensors in a Bayer CFA.
>
> 2) Even the interpolation were to be done using only the
> green sensors for luminance, the simple fact is that
> X/2 is greater than X/3, and therefore a Bayer CFA will
> have better pixel resolution as describe above than
> will a "3 sensor" configuration.


That's just wrong. The Bayer sensor luminance resolution is X/2 where
X is the sensel resolution. The chrominance resolution is X/4. A 3
sensor system has a luminance and chrominace resolution of X, where X
is the same sensel resolution of the Bayer sensor.

Put another way, Placing a Bayer color filter over a 16MP sensor turns
that sensor into an 8MP luminance and 4MP chrominance sensor. On the
other hand, taking the same 16MP sensor, using 3 of them and putting a
single color filter in front of each turns that sensor into a 16MP
luminance and 16MP chrominance sensor. I'm sure you'll agree that 16MP
is better than 8 or 4MP

Steve
 
Reply With Quote
 
TheRealSteve
Guest
Posts: n/a
 
      04-16-2012

On Sun, 15 Apr 2012 18:29:53 -0800, (E-Mail Removed) (Floyd L.
Davidson) wrote:

>Mxsmanic <(E-Mail Removed)> wrote:
>>Floyd L. Davidson writes:
>>
>>> But let me just point out the one single most glaring flaw in
>>> you whole argument: For any sensor with X number of sensor
>>> locations, a "3 sensor" configuration produces a maximum pixel
>>> resolution of X/3, but a Bayer Configuration produces X/1
>>> pixels.

>>
>>Three seprate sensors of resolution x produce a final resolution of x, for
>>both luminance and chroma.

>
>It produces a pixel resolution of X/3.


Wrong. Simple math here. Take a sensor with X pixels and you have a
sensor has X sensels. 3 of them have 3X sensels. Put a bayer filter
over a single sensor and you have approximately X/4 "color pixels". A
3 sensor system has 3X/3 = X color pixels.

Steve
 
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