Nagging questions
Eric Miller
that the questions themselves may be misguided, so feel free to interpret.
Chip size, noise and other stuff.
All other things being equal, will a larger sensor have more noise? I
ask this for two, interrelated reasons, the first of which is that I see
discussions of the end of the Megapixel wars for APS-C sensor due to the
loss of image quality with higher pixel pitch. The second is that I have
recently seen some pretty good images from a Canon G10 which has a much
higher pixel pitch albeit on a much smaller chip.
Actual resolution increase with higher pixel pitch.
It certainly seems that the G10 has much higher resolution than my 5D
(but certainly not better image quality) considering the pixel
pitch/sensor size. What actual resolution increase could I expect with
that kind of pixel pitch on an APS-C or FF/35mm sensor? Is there some
formula that describes the diminishing resolution gains due to added noise?
Size of pixel/photosite where resolution increase would not equal
additional pixels.
Is there a certain way in which noise is related to the size of the
individual photosite? Or is that dependent upon other factors that can
be addressed through better engineering/manufacture of the sensor?
Conversion of Bayer pattern only affect colors of image.
Finally, do the algorithms used to convert the output of a Bayer-type
sensor independently process luminance and color information? I guess
that would be, does the algorithm produce color and luminance
information independently of each other such that a grayscale value
could be produced for each pixel without using information collected at
more than one photosite?
Eric Miller
www.dyesscreek.com
Alan Browne
> Just some random, mostly resolution-related questions. I fully accept
> that the questions themselves may be misguided, so feel free to interpret.
>
> Chip size, noise and other stuff.
>
> All other things being equal, will a larger sensor have more noise? I
> ask this for two, interrelated reasons, the first of which is that I see
> discussions of the end of the Megapixel wars for APS-C sensor due to the
> loss of image quality with higher pixel pitch. The second is that I have
> recently seen some pretty good images from a Canon G10 which has a much
> higher pixel pitch albeit on a much smaller chip.
Basically it is the sensor "site size" that has the most affect on
signal but noise does not vary that much. Of course with less signal
the noise becomes more important.
Canon has developed remarkable noise reduction s/w as well, so they can
make less than optimal hardware look very good.
>
> Actual resolution increase with higher pixel pitch.
Lens and anti-alias filter limited.
> It certainly seems that the G10 has much higher resolution than my 5D
> (but certainly not better image quality) considering the pixel
> pitch/sensor size. What actual resolution increase could I expect with
> that kind of pixel pitch on an APS-C or FF/35mm sensor? Is there some
> formula that describes the diminishing resolution gains due to added noise?
>
> Size of pixel/photosite where resolution increase would not equal
> additional pixels.
>
> Is there a certain way in which noise is related to the size of the
> individual photosite? Or is that dependent upon other factors that can
> be addressed through better engineering/manufacture of the sensor?
Yes. Roughly speaking noise does not change much with site size; but
signal goes up as the site size increases thereby increasing S/N.
>
> Conversion of Bayer pattern only affect colors of image.
>
> Finally, do the algorithms used to convert the output of a Bayer-type
> sensor independently process luminance and color information? I guess
> that would be, does the algorithm produce color and luminance
> information independently of each other such that a grayscale value
> could be produced for each pixel without using information collected at
> more than one photosite?
Good question.
nospam
<miller_no...@bellsouth.net> wrote:
> Just some random, mostly resolution-related questions. I fully accept
> that the questions themselves may be misguided, so feel free to interpret.
>
> Chip size, noise and other stuff.
>
> All other things being equal, will a larger sensor have more noise?
no. for a given sensor technology and pixel count, a larger sensor will
have larger pixels, and therefore have less noise.
<http://www.clarkvision.com/imagedetail/digital.sensor.performance.summa
ry/>
> I ask this for two, interrelated reasons, the first of which is that I see
> discussions of the end of the Megapixel wars for APS-C sensor due to the
> loss of image quality with higher pixel pitch. The second is that I have
> recently seen some pretty good images from a Canon G10 which has a much
> higher pixel pitch albeit on a much smaller chip.
> Actual resolution increase with higher pixel pitch.
>
> It certainly seems that the G10 has much higher resolution than my 5D
> (but certainly not better image quality) considering the pixel
> pitch/sensor size. What actual resolution increase could I expect with
> that kind of pixel pitch on an APS-C or FF/35mm sensor? Is there some
> formula that describes the diminishing resolution gains due to added noise?
resolution depends on a lot of factors, including the sensor, the lens,
how accurately you focus, etc. you can have the best sensor in the
world but if you slap a cheapo lens on the camera, it won't matter
much.
if you want to know how a g10 compares with a 5d, you need to test them
both. dpreview has done that, and the g10 will have slightly higher
resolution, but a lot more noise.
> Size of pixel/photosite where resolution increase would not equal
> additional pixels.
>
> Is there a certain way in which noise is related to the size of the
> individual photosite? Or is that dependent upon other factors that can
> be addressed through better engineering/manufacture of the sensor?
the bigger the better, but better sensor technology counts too. a small
sensor of today might be better than a large sensor of a few years ago,
but you can always make a larger sensor with the latest technology so
it's moot.
> Conversion of Bayer pattern only affect colors of image.
>
> Finally, do the algorithms used to convert the output of a Bayer-type
> sensor independently process luminance and color information? I guess
> that would be, does the algorithm produce color and luminance
> information independently of each other such that a grayscale value
> could be produced for each pixel without using information collected at
> more than one photosite?
there are many many ways to demosaic bayer images, and you will need
more than one photosite no matter what you do.
NameHere
>In article <hf6em7$oq2$1...@news.eternal-september.org>, Eric Miller
><miller_no...@bellsouth.net> wrote:
>
>> Just some random, mostly resolution-related questions. I fully accept
>> that the questions themselves may be misguided, so feel free to interpret.
>>
>> Chip size, noise and other stuff.
>>
>> All other things being equal, will a larger sensor have more noise?
>
>no. for a given sensor technology and pixel count, a larger sensor will
>have larger pixels, and therefore have less noise.
>
><http://www.spamsite.com/imagedetail/digital.sensor.performance.summa
>ry/>
>
I guess that's why this 1/2.5" sized sensor has more dynamic range than
most APS-C sized DSLR sensors which only have 7-8 EV stops.
http://farm4.static.flickr.com/3142/2861257547_9a7ceaf3a1_o.jpg
You're a pretend-photographer troll, one that loves nothing better than to
spout biased misinformation posted by another DSLR-Troll who posted tests
on his website to justify why he wasted all that money on his DSLR gear.
I.e. Clarkblindness.com
nospam
<wh...@address.info> wrote:
> I guess that's why this 1/2.5" sized sensor has more dynamic range than
> most APS-C sized DSLR sensors which only have 7-8 EV stops.
most dslr sensors are capable of over 12 stops, limited by the d/a
converter.
RichA
wrote:
> Just some random, mostly resolution-related questions. I fully accept
> that the questions themselves may be misguided, so feel free to interpret.
>
> Chip size, noise and other stuff.
>
> All other things being equal, will a larger sensor have more noise? I
> ask this for two, interrelated reasons, the first of which is that I see
> discussions of the end of the Megapixel wars for APS-C sensor due to the
> loss of image quality with higher pixel pitch. The second is that I have
> recently seen some pretty good images from a Canon G10 which has a much
> higher pixel pitch albeit on a much smaller chip.
>
> Actual resolution increase with higher pixel pitch.
>
> It certainly seems that the G10 has much higher resolution than my 5D
> (but certainly not better image quality) considering the pixel
> pitch/sensor size.
Stop using Canon lenses and you could reverse that.
Alfred Molon
miller_no...@bellsouth.net says...
> Finally, do the algorithms used to convert the output of a Bayer-type
> sensor independently process luminance and color information?
No, they work on RGB lebel - not on luminance/chrominance level.
> I guess
> that would be, does the algorithm produce color and luminance
> information independently of each other such that a grayscale value
> could be produced for each pixel without using information collected at
> more than one photosite?
No, because luminance information is not available at single photosites.
You either have red, green or blue light intensity measurements, but not
luminance at the single pixel level. To have that you would have to
remove the colour filters.
--
Alfred Molon
------------------------------
Olympus E-series DSLRs and micro 4/3 forum at
http://tech.groups.yahoo.com/group/MyOlympus/
http://myolympus.org/ photo sharing site
Bob Larter
> On Wed, 02 Dec 2009 16:02:21 -0500, nospam <nos...@nospam.invalid> wrote:
>
>> In article <hf6em7$oq2$1...@news.eternal-september.org>, Eric Miller
>> <miller_no...@bellsouth.net> wrote:
>>
>>> Just some random, mostly resolution-related questions. I fully accept
>>> that the questions themselves may be misguided, so feel free to interpret.
>>>
>>> Chip size, noise and other stuff.
>>>
>>> All other things being equal, will a larger sensor have more noise?
>> no. for a given sensor technology and pixel count, a larger sensor will
>> have larger pixels, and therefore have less noise.
>>
>> <http://www.spamsite.com/imagedetail/digital.sensor.performance.summa
>> ry/>
>>
>
> I guess that's why this 1/2.5" sized sensor has more dynamic range than
> most APS-C sized DSLR sensors which only have 7-8 EV stops.
>
> http://farm4.static.flickr.com/3142/2861257547_9a7ceaf3a1_o.jpg
>
> You're a pretend-photographer troll, one that loves nothing better than to
Don't mind this loon. He has no idea what he's talking about, & delights
in wasting everybody's time.
--
W
. | ,. w , "Some people are alive only because
\|/ \|/ it is illegal to kill them." Perna condita delenda est
---^----^---------------------------------------------------------------
Peter Chant
> most dslr sensors are capable of over 12 stops, limited by the d/a
> converter.
Is this the limit? Then why did Fuji have their fancy "super" sensor with
differing sizes of sites? I wonder why someone does not come out with a 16
bit a/d chip?
Pete
Bob Larter is Lionel Lauer - Look it up.
wrote:
>NameHere wrote:
>> On Wed, 02 Dec 2009 16:02:21 -0500, nospam <nos...@nospam.invalid> wrote:
>>
>>> In article <hf6em7$oq2$1...@news.eternal-september.org>, Eric Miller
>>> <miller_no...@bellsouth.net> wrote:
>>>
>>>> Just some random, mostly resolution-related questions. I fully accept
>>>> that the questions themselves may be misguided, so feel free to interpret.
>>>>
>>>> Chip size, noise and other stuff.
>>>>
>>>> All other things being equal, will a larger sensor have more noise?
>>> no. for a given sensor technology and pixel count, a larger sensor will
>>> have larger pixels, and therefore have less noise.
>>>
>>> <http://www.spamsite.com/imagedetail/digital.sensor.performance.summa
>>> ry/>
>>>
>>
>> I guess that's why this 1/2.5" sized sensor has more dynamic range than
>> most APS-C sized DSLR sensors which only have 7-8 EV stops.
>>
>> http://farm4.static.flickr.com/3142/2861257547_9a7ceaf3a1_o.jpg
>>
>> You're a pretend-photographer troll, one that loves nothing better than to
>[snip]
>
>Don't mind this loon. He has no idea what he's talking about, & delights
>in wasting everybody's time.
Bob Larter's legal name: Lionel Lauer
Home news-group, an actual group in the "troll-tracker" hierarchy:
alt.kook.lionel-lauer (established on, or before, 2004)
Registered Description: "the 'owner of several troll domains' needs a group where he'll stay on topic."
<http://groups.google.com/groups/search?hl=en&num=10&as_ugroup=alt.kook.lionel-lauer>
"Results 1 - 10 of about 2,170 for group:alt.kook.lionel-lauer."
nospam
<pet...@MpeteOzilla.Vco.ukE> wrote:
> I wonder why someone does not come out with a 16
> bit a/d chip?
recent slrs use 14 bit d/a converters and some medium format backs have
16 bit converters.
Alan Browne
> nospam wrote:
>
>
>> most dslr sensors are capable of over 12 stops, limited by the d/a
>> converter.
>
> Is this the limit? Then why did Fuji have their fancy "super" sensor with
> differing sizes of sites? I wonder why someone does not come out with a 16
> bit a/d chip?
The fuji sensor solved the highlights issue, not the shadow issue that
"deeper" a/d should help with.
Several of the MF cameras use 16 bit a/d. It's diminishing returns
after 12 or 14 bits.
Alan Browne
> nospam wrote:
>
>
>> most dslr sensors are capable of over 12 stops, limited by the d/a
>> converter.
>
> Is this the limit? Then why did Fuji have their fancy "super" sensor with
> differing sizes of sites? I wonder why someone does not come out with a 16
> bit a/d chip?
The fuji sensor solved the highlights issue, not the shadow issue that
Alfred Molon
nos...@nospam.invalid says...
> most dslr sensors are capable of over 12 stops, limited by the d/a
> converter.
Not true. Most DSLRs have less than 9 stops of dynamic range, according
to the tests of dpreview. See for instance the Nikon D300s:
http://www.dpreview.com/reviews/nikond300s/page17.asp
The best it achieves is 8.6 EV. Even the D3, with its bigger pixels is
not better:
http://www.dpreview.com/reviews/NikonD3/page20.asp
Also here the maximum is 8.6EV.
JimKramer
news:MPG.258390796...@news.supernews.com...
> In article <021220091653162593%nos...@nospam.invalid>,
> nos...@nospam.invalid says...
>> most dslr sensors are capable of over 12 stops, limited by the d/a
>> converter.
>
> Not true. Most DSLRs have less than 9 stops of dynamic range, according
> to the tests of dpreview. See for instance the Nikon D300s:
> http://www.dpreview.com/reviews/nikond300s/page17.asp
>
> The best it achieves is 8.6 EV. Even the D3, with its bigger pixels is
> not better:
> http://www.dpreview.com/reviews/NikonD3/page20.asp
>
> Also here the maximum is 8.6EV.
Reading all the way down to "RAW headroom" on the D3 link, 2nd paragraph:
"As usual the default Adobe Camera RAW conversion delivers less dynamic
range than JPEG from the camera (a more contrasty tone curve and very little
noise reduction in shadows). Simply switching to 'Auto' in the ACR
conversion dialog reaps huge rewards (we measured the result to have exactly
12 stops of dynamic range), and in our tests with real world shots produced
superb results with images that seemed to be over exposed beyond
redemption."
-Jim
nospam
Molon <alfred...@yahoo.com> wrote:
> > most dslr sensors are capable of over 12 stops, limited by the d/a
> > converter.
>
> Not true. Most DSLRs have less than 9 stops of dynamic range, according
> to the tests of dpreview. See for instance the Nikon D300s:
> http://www.dpreview.com/reviews/nikond300s/page17.asp
dpreview tests dynamic range by comparing jpegs which is bogus, to say
the least.
> The best it achieves is 8.6 EV. Even the D3, with its bigger pixels is
> not better:
> http://www.dpreview.com/reviews/NikonD3/page20.asp
>
> Also here the maximum is 8.6EV.
scroll to the bottom where they measured 12 stops with raw.
dxo labs measured it at 12.2 here:
<http://www.dxomark.com/index.php/eng/Image-Quality-Database/Nikon/D3>
and roger clark measured 13.7 here, but he measures the capability of
the sensor itself, not the rest of the system.
<http://www.clarkvision.com/imagedetail/digital.sensor.performance.summa
ry/>
Eric Miller
> In article <hf6em7$oq2$1...@news.eternal-september.org>, Eric Miller
> <miller_no...@bellsouth.net> wrote:
>
>> Just some random, mostly resolution-related questions. I fully accept
>> that the questions themselves may be misguided, so feel free to interpret.
>>
>> Chip size, noise and other stuff.
>>
>> All other things being equal, will a larger sensor have more noise?
>
> no. for a given sensor technology and pixel count, a larger sensor will
> have larger pixels, and therefore have less noise.
>
> <http://www.clarkvision.com/imagedetail/digital.sensor.performance.summa
> ry/>
>
Okay, but if the larger sensor has pixels of the same size, will it have
more noise?
Eric Miller
www.dyesscreek.com
nospam
<miller_no...@bellsouth.net> wrote:
> Okay, but if the larger sensor has pixels of the same size, will it have
> more noise?
if the pixels are the same size, the larger sensor will have more of
them, which is also beneficial.
Ofnuts
Adding more bits makes sense only if you get meaningful data, i.e., if
the low bits are still above noise level. IIRC each bit earned requires
6dB more signal/noise power ratio. For 14 bits, the electronics should
be as quiet as those of an audio device with 84db of signal/noise ratio,
and for 16 bits if goes to 96dB. Not undoable, but not that easy.
--
Bertrand
Alan Browne
I should even add, that IIRC, DXO tests of Hasselblad sensors at 16 bit
a/d show them to not be as good as Nikon, Canon and Sony cameras in
noise terms.
Alfred Molon
nos...@nospam.invalid says...
> > The best it achieves is 8.6 EV. Even the D3, with its bigger pixels is
> > not better:
> > http://www.dpreview.com/reviews/NikonD3/page20.asp
> >
> > Also here the maximum is 8.6EV.
>
> scroll to the bottom where they measured 12 stops with raw.
He is writing that with RAW usually you have one more stop of headroom.
The D3 achieves up to 12, but it is also a full-frame camera, pretty
much the best performing one at the moment. Most DSLRs use smaller
sensors and have therefore less dynamic range. You claimed that "most
dslr sensors are capable of over 12 stops" which is simply not true.
nospam
Molon <alfred...@yahoo.com> wrote:
> He is writing that with RAW usually you have one more stop of headroom.
> The D3 achieves up to 12, but it is also a full-frame camera, pretty
> much the best performing one at the moment. Most DSLRs use smaller
> sensors and have therefore less dynamic range. You claimed that "most
> dslr sensors are capable of over 12 stops" which is simply not true.
yes it is.
<http://www.clarkvision.com/imagedetail/digital.sensor.performance.summa
ry/>
Wolfgang Weisselberg
> Most DSLRs use smaller
> sensors and have therefore less dynamic range.
Bzzzt.
Sensor size *as such* has nothing to do with dynamic range.
Pixel size, however ...
Small sensors tend to have smaller pixels, that's all.
> You claimed that "most
> dslr sensors are capable of over 12 stops" which is simply not true.
http://clarkvision.com/articles/digital.sensor.performance.summary/index.html#data
indicates that the sensors are perfectly capable of 12+ stops,
based on their full well capacity and read noise, but excluding
A/D noise (which is not really tied to the sensor, anyway).
The rest of the page makes it clear that the camera performance
is limited by
a) 12 bit A/D converters,
b) noise in A/D converters,
c) *much* less improvement for A/D converters of 14 bit than would
be assumed by the additional bits. High megaherz speeds seem
to do that, at least for now.
-Wolfgang