Forgot Password
Pentax Camera Forums Home
 

Reply
Show Printable Version Search this Thread
11-02-2013, 01:17 PM   #76
Lens Buying Addict
monochrome's Avatar

Join Date: Mar 2008
Location: Kirkwood (St. Louis) MO
Photos: Gallery | Albums
Posts: 19,461
This whole thread just begs for the first Firmware update.

11-02-2013, 01:44 PM   #77
Forum Member




Join Date: Aug 2009
Location: reno
Posts: 70
QuoteOriginally posted by jsherman999 Quote
Actually, everything Class A said was correct.



It's in fact telling you that for purposes of image SNR, the photosite size is not what matters because the SNR can be averaged over several photosites.

The 'larger photosite = better' myth has been debunked for a few years now.

.
please reread what i wrote.
11-02-2013, 03:34 PM   #78
Pentaxian




Join Date: Jun 2009
Location: GMT +10
Photos: Albums
Posts: 10,576
QuoteOriginally posted by Barry Pearson Quote
I have now done my own side-by-side tests with my own cameras, and come to significantly different conclusions from those I came to from processing the Imaging Resource DNGs.
Interesting results. The K-3 certainly comes up looking cleaner. The edges of the colours in your ColorMunki swatch also render more defined - eg the dark greens - than the colours in the IR swatch down the bottom of their test scene.

Barry have you run those same IR K-3 DNG's through the Pentax Digital Camera Utility 5 software that came with the camera? This is the most up-to-date K-3 DNG processor and also the 'official' tool for reading K-3 DNG's. I'd be interested to see if it knows some secrets about how to optimally render K-3 images, and if it can do better than ACR/Lightroom/RawTherapee.
11-02-2013, 04:52 PM   #79
Pentaxian
Class A's Avatar

Join Date: Aug 2008
Location: Wellington, New Zealand
Posts: 8,934
QuoteOriginally posted by azimuthal Quote
this is for dof equivalence, not particularly meaningful in this situation.
If you keep other parameters (shutter speed, ISO) equivalent then the total amount of light changes if the DOF changes. There is no way around that.

QuoteOriginally posted by azimuthal Quote
one stop of advantage in what?
SNR

QuoteOriginally posted by azimuthal Quote
you seem to think that taking a full-frame lens like the 70mm sigma off of an aps-c and putting it on a full-frame magically causes more light to go through it.
No, I don't think that.

QuoteOriginally posted by azimuthal Quote
this is not the case, it is true that more of the light is used in the case of the full-frame, this is what you might call an advantage of full-frame.
So we agree after all. But let's drop the "might".

Since IR move a full-frame camera closer to the scene compared to an APS-C camera, the field of view does not change. Perspective changes, but we can ignore that for a noise discussion.

So the full-frame camera captures the same scene using more total light (as it uses the full image circle of the Sigma 70/2.8, not just an APS-C crop of it). The surplus in total amount of light used gives the larger sensor an SNR advantage. One can level the playing field by stopping down the lens on the larger sensor. This not only brings the total amount of light used to the same level but also achieves the same DOF for both images.

QuoteOriginally posted by azimuthal Quote
promising news.
Arrogance is never nice, but combined with being wrong it becomes rather unfortunate.

QuoteOriginally posted by azimuthal Quote
so i'll try to help. this article says that if you keep the sensor size the same and increase the resolution while lowering the size of the photosites, you can still find an advantage in SNR, despite the smaller photosites.
There is no SNR advantage when you increase resolution.

You seem to be misinterpreting the pixel-level increase of SNR that stems from downscaling to be an absolute increase in terms of image noise. Note, however, that the higher resolution images starts with a worse pixel-level SNR so the downscaling just brings the pixel-level SNR to the same level as the low resolution image (not a better one). In summary, noise at the image level (i.e., what you see when you print to the same size, for instance) is not affected by resolution. [As a matter of fact, higher resolution has an advantage for noise distribution which also implies a slightly slower peak in the distribution shape, but that is a subtle point and it makes sense to simplify the discussion to the total noise energy here, and the latter does not change with resolution.]

QuoteOriginally posted by azimuthal Quote
the d600(24MP) has better snr than the k3(24MP) because although they have the same megapixels, the d600 has larger photosites. (assuming similar technology)
This is not correct. You could change the resolution of the D600 sensor so that its photosites have the same size as the K-3. As a result, the SNR would not change (please read the DxOMark article again).

The true underlying reason for the D600's better SNR (when using the same f-ratio, which isn't fair) is that its sensor is larger. The fact that its photosites are larger than those of the K-3 is just an indirect effect of the sensor being larger (and the resolution being the same).

QuoteOriginally posted by azimuthal Quote
the d800(36MP) has better (normalized) snr than the k5(16MP) because although they have the same pixel size, the D800 has higher resolution.
Again, you are claiming an incorrect cause-effect relationship.

Higher resolution does not imply better SNR. In your particular example, your conclusion can be observed, but only because having higher resolution while sharing the same photosite size is made possible by using a larger sensor. So again, the underlying reason for the SNR advantage is sensor size.

SNR advantages never come from photosite differences or resolution differences alone. SNR differences can only be observed when photosite-size / resolution differences are caused by a sensor size difference.


Last edited by Class A; 11-02-2013 at 05:26 PM.
11-02-2013, 05:01 PM   #80
Pentaxian
Class A's Avatar

Join Date: Aug 2008
Location: Wellington, New Zealand
Posts: 8,934
QuoteOriginally posted by Barry Pearson Quote
Important edit: I have now done my own side-by-side tests with my own cameras, and come to significantly different conclusions from those I came to from processing the Imaging Resource DNGs. See another post here.
@Brew1brew pointed out that you didn't use the same shutter speeds in your comparisons.

I personally will suspend judgement until I see the DxOMark figures for the K-3.
11-02-2013, 07:29 PM   #81
Forum Member




Join Date: Oct 2010
Location: Jakarta
Photos: Gallery
Posts: 96
QuoteOriginally posted by rawr Quote
I'd be interested to see if it knows some secrets about how to optimally render K-3 images, and if it can do better than ACR/Lightroom/RawTherapee.
Here's something interesting:

Recall that I use amaze demosaic algorithm which gives the best detail at lower ISOs but is noisy at high ISOs. There are actually 6 or 7 Algorithms to choose from in rawtherapee. So while most other software only gives you one algorithm and then tweak the hell out of the variables to give the best profile for your camera, Rawtherapee lets you to actually chose the engine to run your race with (this is of course how i perceive it. Corrections are welcome).

Here's K-3 Amaze vs K-3 igv demosaic:


For High ISO shots the igv demosaic algorithm fare better in that it reduces color blotches and refined the grain while preserving the color that we want to show. Of course that means other cameras also benefits from it, but i notice that the igv demosaic introduces aliasing on well defined diagonal lines on lower MP sensors such as the K-5. So there are trade offs.
11-02-2013, 08:23 PM   #82
Pentaxian




Join Date: Jun 2009
Location: GMT +10
Photos: Albums
Posts: 10,576
QuoteOriginally posted by wahid_satu Quote
So there are trade offs.
Always.
I still prefer the amaze shot though. Better pattern detail and contrast (on my monitor). But I can see what you mean by the colour splotches.

One day I'd like see a side-by-side high ISO test or 'shoot-out' between RT/LR/PDCU5/DxO/maybe Aperture, all rendering the same K-3 DNG. Maybe when PDCU 5 is available for download and all the other big name RAW processors are updated for the K-3.

I might give it a try since I have RT/LR/DxO, and should be able to get PDCU5 if they make it available for download. Who knows, if Capture One Express updates itself for the K-3, I could include that in the test too.

Last edited by rawr; 11-02-2013 at 08:29 PM.
11-02-2013, 09:56 PM   #83
Pentaxian
Class A's Avatar

Join Date: Aug 2008
Location: Wellington, New Zealand
Posts: 8,934
QuoteOriginally posted by rawr Quote
I might give it a try since I have RT/LR/DxO, and should be able to get PDCU5 if they make it available for download. Who knows, if Capture One Express updates itself for the K-3, I could include that in the test too.
That would be awesome!

11-03-2013, 01:21 AM   #84
Forum Member




Join Date: Aug 2009
Location: reno
Posts: 70
QuoteOriginally posted by Class A Quote

The true underlying reason for the D600's better SNR (when using the same f-ratio, which isn't fair) is that its sensor is larger. The fact that its photosites are larger than those of the K-3 is just an indirect effect of the sensor being larger (and the resolution being the same).
your argument is akin to saying if i increase the area of a rectangle, the indirect effect is at least one of the sides got larger, that is not an indirect effect that is what caused the increase in area.
area is dependent upon the two sides, not the other way around.

the size of a sensor is defined by the size of the pixels and the number of them, if, between two sensors, we keep one factor the same and increase the other we get an increase in sensor size, and thus SNR.

so once again, if we make two sensors with the same resolution, but one has a larger pixel size, the one with the larger pixel size will have higher SNR (it will also be larger in size, but this should be obvious)
if we make two sensors with the same pixel size but increase the resolution, the one with the larger resolution will have higher normalized SNR (it will also be larger in size, but this should be obvious)

it seems like you are using semantics to avoid your earlier statement in response to this:
"all other things being equal, the sensor with larger photosites will have higher snr"
the statement is still true. it is also true that it must be larger. but that was obvious from the description.

edit: regarding the quote: ok i see what you are saying, i thought that he implied that the sensor was larger but he was explicit about it and so you are right, a larger sensor with more megapixels and the same pixel size would also have a higher (normalized) SNR.

but still if we constrain megapixels and increase pixel size, it is impossible for the size not to increase, just as if we constrain the pixel size and increase megapixels the sensor size must increase.
the three things are proportional, given two we can find the third, constraining one while increasing another results in an increase in the third.

Last edited by azimuthal; 11-03-2013 at 02:00 AM.
11-03-2013, 02:12 AM   #85
Pentaxian
Class A's Avatar

Join Date: Aug 2008
Location: Wellington, New Zealand
Posts: 8,934
QuoteOriginally posted by azimuthal Quote
...it seems like you are using semantics to avoid your earlier statement in response to this:
"all other things being equal, the sensor with larger photosites will have higher snr"
the statement is still true. it is also true that it must be larger. but that was obvious from the description.
I'm not using semantics and please quote correctly (see below).

I took issue with your phrasing that made it look like as if a dependent variable were the actual cause of the effect.

When you make statements like "the SNR increases .... because the resolution is higher" or the "SNR increases ... because the photosites are larger" that's akin to saying "The flowers on my balcony were watered because the grass below them is wet". There was an underlying cause (it rained) but you are pointing to one of its effects to explain another effect. Note that the grass could be wet for other reasons (e.g., the sprinklers were turned on) -- just like it is possible for the resolution to change without the sensor size changing -- but you are claiming that the grass being wet is still the primary cause because it is obvious that there is only one water source for grass and flowers.

Now we could argue all day which comes first -- sensor size or photosite-size & resolution -- and which must be the dependent variable(s) respectively. An argument in my favour is that when comparing two sensors I don't need to know their actual resolutions and photosite-sizes. I only need to know which of them has the bigger surface in order to know which one has the SNR advantage (when using the same f-ratio). Resolutions and photosite-size do not need to enter the picture at all in that discussion and you can change them arbitrarily as long as they combine to the given sensor sizes.

Furthermore, the distinct problem with your kind of phrasing is that it could mislead people into thinking that changes to photosite-size or resolution are primary causes for affecting SNR.

Consider Cynog Ap Brychan statement (this time quoted correctly)
"I completely agree that the larger sensor should have a better signal to noise ratio if it has correspondingly larger photosites, all other things being equal (which they never are)"
I highlighted the fragment I referred to when I responded that the photosite-size (on its own, of course) does not matter. Note that his premise already was that the sensor is larger but he then added an extra condition about the photosite size. The latter was unnecessary which is why I responded with the DxOMark article reference.

Nevertheless, you found the need to suggest that I either didn't read the DxOMark article or did not understand it. This comment of yours was inappropriate in hindsight wasn't it? So I don't think that I got the wrong end of the semantics stick.

Anyhow, this discussion is neither interesting for us nor anyone else, so I won't engage in it any further.

Last edited by Class A; 11-03-2013 at 02:23 AM.
11-03-2013, 02:24 AM   #86
Forum Member




Join Date: Aug 2009
Location: reno
Posts: 70
QuoteOriginally posted by Class A Quote
When you make statements like "the SNR increases .... because the resolution is higher" or the "SNR increases ... because the photosites are larger" that's akin to saying "The flowers on my balcony were watered because the grass below them is wet". There was an underlying cause (it rained) but you are pointing to one of its effects to explain another effect. Note that the grass could be wet for other reasons (e.g., the sprinklers were turned on) -- just like it is possible for the resolution to change without the sensor size changing -- but you are claiming that the grass being wet is still the primary cause because it is obvious that there is only one water source for grass and flowers.
now you are committing the exact same crime you are condemning me for, misquoting or selectively quoting.

i qualified my statements by saying that the
SNR increases IF the pixel size is the same AND the resolution increases.
SNR increases IF the resolution is the same AND pixel size increases.
11-03-2013, 02:34 AM   #87
Pentaxian
Class A's Avatar

Join Date: Aug 2008
Location: Wellington, New Zealand
Posts: 8,934
QuoteOriginally posted by azimuthal Quote
now you are committing the exact same crime you are condemning me for, misquoting or selectively quoting.

i qualified my statements by saying that the
SNR increases IF the pixel size is the same AND the resolution increases.
SNR increases IF the resolution is the same AND pixel size increases.
I'm sorry that you feel misrepresented.

I deliberately used "..." to indicate that I left out parts of your quote. I left these parts out in order to highlight the incorrect causal relationship your phrasing was suggesting. I would have been fine if you had used "IF" and "AND" as you do in your new versions.

The versions I took issue you with were of the form
"SNR increases ALTHOUGH the pixel size is the same BECAUSE the resolution increases."

Here's your original: "the d800(36MP) has better (normalized) snr than the k5(16MP) because although they have the same pixel size, the D800 has higher resolution.".
11-03-2013, 02:47 AM   #88
Forum Member




Join Date: Aug 2009
Location: reno
Posts: 70
QuoteOriginally posted by Class A Quote
I'm sorry that you feel misrepresented.

I deliberately used "..." to indicate that I left out parts of your quote. I left these parts out in order to highlight the incorrect causal relationship your phrasing was suggesting. I would have been fine if you had used "IF" and "AND" as you do in your new versions.

The versions I took issue you with were of the form
"SNR increases ALTHOUGH the pixel size is the same BECAUSE the resolution increases."

Here's your original: "the d800(36MP) has better (normalized) snr than the k5(16MP) because although they have the same pixel size, the D800 has higher resolution.".
ok yea, for future reference don't quote somebody and take out qualifying statements and then complain that the qualifying statements aren't there -- even if you have a top-notch watering metaphor to go with it.
11-03-2013, 07:10 AM   #89
Pentaxian
jsherman999's Avatar

Join Date: Dec 2007
Photos: Gallery
Posts: 8,228
QuoteOriginally posted by azimuthal Quote
...
the size of a sensor is defined by the size of the pixels and the number of them, if, between two sensors, we keep one factor the same and increase the other we get an increase in sensor size, and thus SNR.
The size of the sensor is defined only by the size of the sensor, that is, the dimensions of the sensor.

QuoteQuote:
so once again, if we make two sensors with the same resolution, but one has a larger pixel size, the one with the larger pixel size will have higher SNR (it will also be larger in size, but this should be obvious)
if we make two sensors with the same pixel size but increase the resolution, the one with the larger resolution will have higher normalized SNR (it will also be larger in size, but this should be obvious)
But the size of pixels is not the determining factor in SNR, there. You're confusing causation will correlation, and attributing the change in SNR to the wrong factor.

The SNR gets better there because the area of the sensor gets bigger, simple as that. There are going to be upper and lower bounds for the number of pixels that should be in that area which are dependent on physics and current technology, but as long as you stay within those bounds, the pixel size doesn't matter for SNR - Just as originally stated by Class A in this thread, and implied by the DXOMark article you misread.





QuoteQuote:
it seems like you are using semantics to avoid your earlier statement in response to this:
"all other things being equal, the sensor with larger photosites will have higher snr"
the statement is still true. it is also true that it must be larger. but that was obvious from the description.
Actually what you just wrote above is a perfect example of clinging to semantics, except you were the one doing it. Writing "all other things being equal, the sensor with larger photosites will have higher snr" implied the causation lies with the photosite size, and it doesn't. The photosite size is coincidental to the are of the sensor getting bigger - which is what causes the SNR increase in that comparison.

Last edited by jsherman999; 11-03-2013 at 07:19 AM.
11-03-2013, 02:01 PM   #90
Forum Member




Join Date: Aug 2009
Location: reno
Posts: 70
QuoteOriginally posted by jsherman999 Quote
The size of the sensor is defined only by the size of the sensor, that is, the dimensions of the sensor.
But the size of pixels is not the determining factor in SNR, there. You're confusing causation will correlation, and attributing the change in SNR to the wrong factor.
there is no confusion here, as you cannot keep the number of pixels the same and increase their individual size without increasing sensor size.
an increase in sensor size must be caused by an increase in resolution or an increase in pixel size, or some ratio of the two.
the three are linked. the sensor size depends on the number of elements (megapixels) and the size that each element occupies (pixel size)
obviously there are different ways to arrive at a given sensor size, but it does not change the above fact.

you have four pixels, you increase the size of each pixel and you will increase snr, do you disagree with this statement?

to say that the d600 has better SNR than the K3 because it has the same number of megapixels but the pixels are larger in the case of the d600
is no less valid than saying the d600 has better SNR because it has a larger sensor.
the second statement is very much implied from the first.

oh, and, regarding the quote, i misread it, my mistake and my apologies, but i already addressed this. Class A was entirely right that if we know the sensor size increases the SNR will increase as well.
of course, regarding the specific quote
"I completely agree that the larger sensor should have a better signal to noise ratio if it has correspondingly larger photosites, all other things being equal (which they never are)"
if all other things being equal means megapixels? then the photosites absolutely must be larger for the sensor to be larger, this is a physical fact.
Reply

Bookmarks
  • Submit Thread to Facebook Facebook
  • Submit Thread to Twitter Twitter
  • Submit Thread to Digg Digg
Tags - Make this thread easier to find by adding keywords to it!
camera, cameras, d600, d7100, dslr, focus, image, images, iso, k-3, k3, pentax, pentax k-3, samples
Thread Tools Search this Thread
Search this Thread:

Advanced Search


Similar Threads
Thread Thread Starter Forum Replies Last Post
Pentax K 35mm f/3.5 samples altopiet Lens Sample Photo Archive 17 02-02-2017 08:21 AM
Pentax K-3 Review - many samples palacz Pentax K-3 9 10-22-2013 07:02 PM
K-30 Studio Samples Just Posted! DSims Pentax K-30 & K-50 45 09-22-2012 12:36 AM
Perhaps not Pentax news - D7000 samples on IR Big G Non-Pentax Cameras: Canon, Nikon, etc. 4 11-12-2010 06:20 AM
Updated info on removing the Pentax IR filter for Astro and IR imaging LeoTaylor Pentax DSLR Discussion 26 12-05-2008 03:59 PM



All times are GMT -7. The time now is 07:01 PM. | See also: NikonForums.com, part of our network of photo forums!
  • Red (Default)
  • Green
  • Gray
  • Dark
  • Dark Yellow
  • Dark Blue
  • Old Red
  • Old Green
  • Old Gray
  • Dial-Up Style
Hello! It's great to see you back on the forum! Have you considered joining the community?
register
Creating a FREE ACCOUNT takes under a minute, removes ads, and lets you post! [Dismiss]
Top