Forgot Password
Pentax Camera Forums Home
 

Reply
Show Printable Version Search this Thread
12-19-2014, 06:03 PM   #31
Pentaxian
Digitalis's Avatar

Join Date: Mar 2009
Location: Adelaide.
Posts: 8,530
QuoteOriginally posted by MJSfoto1956 Quote
I don't think you understand how a Bayer array works. The fact is, all derived "pixels" from the "raw" information supplied by a Bayer array are interpolations
I know that the blue and red channels are interpolated significantly more than green channel is, I just don't see how shuffling the filters around as it were is going to offer the claimed boost in image quality when it will have no effect on MTF - hence my statement on increasing color fidelity - however from what I know no consumer level DSLR has true 16bit raw files, so a majority of this data will be thrown away.

12-19-2014, 08:47 PM - 1 Like   #32
Site Supporter




Join Date: Jan 2010
Location: Michigan
Photos: Gallery
Posts: 2,069
QuoteOriginally posted by Digitalis Quote
I know that the blue and red channels are interpolated significantly more than green channel is,
It is too much for me to hold at one time, so, with 9YO son on the crayons, we did a "cheat sheet" to help when reading about the Bayer etc.
It is elementary and not fully correct or complete, also the various manufacturers do their own hardware and software variations, as I read.

https://app.box.com/s/sftstq56xv1tx7y5hxq6
12-20-2014, 09:01 AM   #33
Pentaxian
MJSfoto1956's Avatar

Join Date: Dec 2013
Location: Boston
Posts: 1,285
Original Poster
this article describes it nicely: how do digital image sensors work

The key take away is that every "pixel" in your final conversion is interpolated. On "average" most pixels have slightly less interpolation in their green channel data than either the red or blue channels (the key thing being "on average" -- every individual final pixel would depend on its neighbors). But in the end every pixel is a "best guess" in terms of what the RAW conversion algorithm thinks the data should be (since it is missing 3/4 of the data for any given pixel). It is for this reason that shifting the sensor around does indeed provide actual increase in resolution (i.e. better MTF) since the shifted pixels are basically "filling in the missing bits".

Michael
12-20-2014, 10:45 AM   #34
Pentaxian




Join Date: Nov 2013
Posts: 4,524
QuoteOriginally posted by MJSfoto1956 Quote
this article describes it nicely: how do digital image sensors work

The key take away is that every "pixel" in your final conversion is interpolated. On "average" most pixels have slightly less interpolation in their green channel data than either the red or blue channels (the key thing being "on average" -- every individual final pixel would depend on its neighbors). But in the end every pixel is a "best guess" in terms of what the RAW conversion algorithm thinks the data should be (since it is missing 3/4 of the data for any given pixel). It is for this reason that shifting the sensor around does indeed provide actual increase in resolution (i.e. better MTF) since the shifted pixels are basically "filling in the missing bits".

Michael
Color sensitivity - DxOMark

This is not 3/4 that is missing but at least naively 2/3 (2 colors out of 3). But because each photosite is not sensible just to his own color but to more colors in term of light intensity alone (without color) you have more than just 1/3 of the information available, even more so for the green chanel that has 50% of the photosites.

So we miss between 1/2 or 2/3 of the information but still reducing the resolution would mean loosing some details we have anyway. Counter productive. So we keep the resolution, some data is interpolated, some data is perfectly known... And we get a final image that way.

As for shifting the sensor, as long as the sensor shift is not exactly of the distance between pixels it will add more resolution regardless and of course decrease noise.



Add to that

12-20-2014, 02:07 PM   #35
Pentaxian
MJSfoto1956's Avatar

Join Date: Dec 2013
Location: Boston
Posts: 1,285
Original Poster
unfortunately in real world shooting it often is more complicated than that. For example, when you shoot in Tungsten light, the blue channel basically has nothing but noise in it. This messes up both the color fidelity as well as the luminance. Something analogous happens when you shoot underwater -- there the red channel is full of noise. And then there is night photography with all sorts of color temperature lighting, with each color temperature affecting the different channels in different ways. One of the key things I've noticed with DxO Optics Pro vs any other RAW converter is just how much detail they "put back" into the blue channel under tungsten lighting. How they pull this off is anyone's guess. But it is immediately noticeable when you compare an indoor shot processed with DxO vs. ACR -- just look at the mess that is the blue channel in ACR then see the difference in DxO. Night and day. I don't think DxO is "faking" it here (but perhaps they are doing some "smoke and mirrors"). Rather, I believe they have data about each sensor which allows their algorithms to be smarter -- especially in less than ideal lighting.

Or put another way: there really is nothing particularly "raw" about the data stored in any RAW format -- to complete the data accurately with the utmost fidelity requires knowledge of the sensor's behaviors under all lighting conditions. Otherwise, just converting the data to RGB without knowledge of the sensor's behavior is leaving some of the color fidelity as well as the luminance noise on the table.

YMMV

Michael
12-21-2014, 07:28 AM   #36
Veteran Member
whk1992's Avatar

Join Date: Aug 2014
Photos: Gallery
Posts: 674
QuoteOriginally posted by ChristianRock Quote
How many people really need such high MP count though? I know I don't, but I don't print wall-sized photos.
everyone who crops photos will appreciate a higher resolution without paying extra.

Imagine yourself using a prime lens and you want pseudo-zoom.

Last edited by whk1992; 12-21-2014 at 07:45 AM.
12-21-2014, 07:44 AM   #37
Pentaxian




Join Date: Nov 2013
Posts: 4,524
QuoteOriginally posted by whk1992 Quote
everyone who crop photos will appreciate a higher resolution without paying extra.

Imagine yourself using a prime lens and you want pseudo-zoom.
+1, but not especially with constraint of tripod and taking 4 photos

I suppose it depend of one preference but I have seen noticable improvement in usability using K3 vs K5 for the added resolution. Would I use 40MP ? Certainly if it would come without drawbacks. And well that's exactly finaly what Olympus does. You may have to make some effort to get 40MP... but it doesn't change anything for your other shots. Would I buy a camera for such feature alone? Higly unlikely. But I mean micr 4/3 is too small to afford much higher resolution than what they have today without big progress in sensor technology... And we might not be there yet. They kinda limited on how to bump the pixel count.

Last edited by Nicolas06; 12-21-2014 at 08:00 AM.
12-21-2014, 07:47 AM   #38
Veteran Member
whk1992's Avatar

Join Date: Aug 2014
Photos: Gallery
Posts: 674
QuoteOriginally posted by Nicolas06 Quote
+1, but not especially with constraint of tripod and taking 4 photos

I suppose it depend of one preference but I have seen noticable improvement in usability using K3 vs K5 for the added resolution. Would I use 40MP ? Certainly if it would come without drawbacks. Ans well that's exactly finaly what Olympus does. You may have to make some effort to get 40MP... but it doesn't change anything for your other shots.
Also, I suppose a 40MP source make noise much less visible after it is scaled down. That's super useful in taking long-exposed night photos.

---------- Post added 12-21-14 at 06:50 AM ----------

I wonder if Pentax is going to add this feature to existing cameras in future firmware updates... The chances are Pentax will put this feature in new cameras only to boost profits if they are considering it.

12-21-2014, 08:03 AM   #39
Pentaxian




Join Date: Nov 2013
Posts: 4,524
QuoteOriginally posted by whk1992 Quote
Also, I suppose a 40MP source make noise much less visible after it is scaled down. That's super useful in taking long-exposed night photos.

---------- Post added 12-21-14 at 06:50 AM ----------

I wonder if Pentax is going to add this feature to existing cameras in future firmware updates... The chances are Pentax will put this feature in new cameras only to boost profits if they are considering it.
That make sense. Why spending money in R&D for old models, in particular in a shriking market? Easier to advertise as new functionnality.
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, dslr, feature, olympus, pentax, photography, photos, sensor, sensor shift, shift, shift for 40mp
Thread Tools Search this Thread
Search this Thread:

Advanced Search


Similar Threads
Thread Thread Starter Forum Replies Last Post
Sensor shift chiane Pentax K-5 47 12-04-2014 10:40 AM
Will Ricoh use this new Sony sensor in future bodies? Winder Pentax DSLR Discussion 22 11-19-2014 05:17 AM
Patents: Sensor shift AF, 80-160 for 645 JPT Pentax News and Rumors 35 07-31-2014 04:24 PM
Rebuttal: My Image Sensor Captures Images DavidSKAF3 General Talk 34 01-07-2014 11:37 AM



All times are GMT -7. The time now is 06:15 AM. | 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