Forgot Password
Pentax Camera Forums Home
 

Reply
Show Printable Version Search this Thread
07-22-2015, 12:29 PM   #1
Site Supporter
Site Supporter




Join Date: Jan 2013
Location: Hampshire, UK
Posts: 1,647
DR or megapixels

Wouldn't it be nice if Pentax introduced 2 variants of their new ff ? Canon and Nikon have done this before.

A two sensor option would be intriguing... One aimed at those who have the consumer urge for the highest number of pixels and version two to be a high Dynamic Range variant, at , say, 24mp. All styling and features otherwise to be the same, maybe frame rate related differences would be necessary and probably a few other firmware tuning differences too ...

Yes, I realise this is simplistic idea and is probably not possible, but it would probably get a very broad appeal and some notice outside our world, especially with pros.

I'd go for the high DR variant. 24mp is sufficient, but a DR of 2 to 3 stops more than the K-3, less noise and generally more latitude would get my vote over huge pixel count. But, having a second variant would satisfy the 50% (???) of others as well ..

07-22-2015, 12:51 PM   #2
Pentaxian




Join Date: Feb 2015
Photos: Gallery
Posts: 12,145
DR or megapixels can be done with a single high pixel count sensor and software selectable decimation filter to get either the high resolution raw file or the lower resolution higher DR raw file, or both.
07-22-2015, 01:03 PM   #3
Site Supporter
Site Supporter
luftfluss's Avatar

Join Date: Jun 2011
Location: NJ
Photos: Gallery | Albums
Posts: 11,581
DR is the new MP, it seems.
07-22-2015, 01:13 PM   #4
Loyal Site Supporter
Loyal Site Supporter




Join Date: Mar 2009
Location: Gladys, Virginia
Photos: Gallery
Posts: 27,572
The D810 has higher dynamic range than the D610/A7 cameras based on DXO Mark testing. Not sure if this is an either/or situation. The A7s, which has quite low megapixels has lower dynamic range than any of them up till about iso 3200.

07-22-2015, 01:24 PM - 1 Like   #5
Veteran Member
Na Horuk's Avatar

Join Date: Feb 2012
Location: Slovenia, probably
Photos: Gallery | Albums
Posts: 11,186
QuoteOriginally posted by BarryE Quote
A two sensor option would be intriguing... One aimed at those who have the consumer urge for the highest number of pixels and version two to be a high Dynamic Range variant, at , say, 24mp. All styling and features otherwise to be the same, maybe frame rate related differences would be necessary and probably a few other firmware tuning differences too ...
OR, do what Magic Lantern can do with Canon cameras: Allow various pixels on the sensor to use different ISO, thus making a single image with much higher DR, but smaller MP. So you have 36MP with regular DR, or you can use this function to get lower MP, but much higher DR. Kind of like HDR, but without any bracketing. Just one shot.
07-22-2015, 02:10 PM   #6
ACG
Forum Member




Join Date: Apr 2014
Photos: Gallery | Albums
Posts: 61
The dynamic range from the a7r, to me, is more impressive than the resolution. The recovery of detail from highlights and shadows is awesome.
I would be more than happy with pentax if they settle with a 24mp high dynamic range sensor than a higher resolution and lower dynamic range sensor for their full frame offering.
07-22-2015, 04:52 PM - 2 Likes   #7
Veteran Member
falconeye's Avatar

Join Date: Jan 2008
Location: Munich, Alps, Germany
Photos: Gallery
Posts: 6,871
Awaiting DxO's testing of the A7rII sensor, the current class leader for both dynamic range and pixel count is the same: D810.

There are technical reasons why a high pixel count facilitates a high dynamic range.

If any differentiation is needed, then a high burst rate, low pixel count, high iso capable sensor would be a possible second option.

07-23-2015, 02:45 AM   #8
Site Supporter
Site Supporter




Join Date: Jan 2013
Location: Hampshire, UK
Posts: 1,647
Original Poster
Just trying to understand this DR V pixel count. Here goes:
(I'll call each pixel a photo well, or PW)

For the same physical sensor size, there will be larger PWs on a sensor that has a lower resolution. A larger PW will fill up to max with received photons, slower than smaller PWs. The photon count range received and maps to the DR. So a 50mp sensor will have a lower DR than a 25mp sensor. However, now the engineering/software kicks in and this is what distinguishes different manufacturers using the same sensors.

I assume the ISO used is effectively irrelevant, as the handling/counting of the photons received is multiplied/amplified at higher ISOs which translates into a numeric that the software uses later. The PWs are likely to fill slower at high ISO as there is less light, but if the image is being overexposed as the ISO is too high, the software will just interpret the photon count then as white. So discounting ISO. Similarly for aperture and shutter speed.

So how might the software handle small PWs that fill up quickly ? Perhaps by emptying a full PW, registering it and allowing it to refill. At high shutter speeds, this might become a limiting factor. If the image is largely exposed within the sensors DR, and only a few PW were needing to be refilled, then there will be fewer operations required. But at high shutter speeds and with a high key image, then processor speed may become an issue. I suppose the algorithms would perhaps have to allow some images to burn out in these circumstances.

So I conclude that, yes a high pixel count can still have an improving DR, but that the physical size of the smaller PWs has to be handled in some way in software. This may be limited by processor speeds and what else it has to do. So inherently, a smaller pixel count, surely must have a better DR, must it not ? Where might I have gone wrong ?
07-23-2015, 03:37 AM   #9
Loyal Site Supporter
Loyal Site Supporter




Join Date: Mar 2009
Location: Gladys, Virginia
Photos: Gallery
Posts: 27,572
QuoteOriginally posted by BarryE Quote
Just trying to understand this DR V pixel count. Here goes:
(I'll call each pixel a photo well, or PW)

For the same physical sensor size, there will be larger PWs on a sensor that has a lower resolution. A larger PW will fill up to max with received photons, slower than smaller PWs. The photon count range received and maps to the DR. So a 50mp sensor will have a lower DR than a 25mp sensor. However, now the engineering/software kicks in and this is what distinguishes different manufacturers using the same sensors.

I assume the ISO used is effectively irrelevant, as the handling/counting of the photons received is multiplied/amplified at higher ISOs which translates into a numeric that the software uses later. The PWs are likely to fill slower at high ISO as there is less light, but if the image is being overexposed as the ISO is too high, the software will just interpret the photon count then as white. So discounting ISO. Similarly for aperture and shutter speed.

So how might the software handle small PWs that fill up quickly ? Perhaps by emptying a full PW, registering it and allowing it to refill. At high shutter speeds, this might become a limiting factor. If the image is largely exposed within the sensors DR, and only a few PW were needing to be refilled, then there will be fewer operations required. But at high shutter speeds and with a high key image, then processor speed may become an issue. I suppose the algorithms would perhaps have to allow some images to burn out in these circumstances.

So I conclude that, yes a high pixel count can still have an improving DR, but that the physical size of the smaller PWs has to be handled in some way in software. This may be limited by processor speeds and what else it has to do. So inherently, a smaller pixel count, surely must have a better DR, must it not ? Where might I have gone wrong ?
I don't know the answer, but photon wells are not necessarily related to pixel size. The D7200 has significantly better dynamic range throughout its range than even the old 16 megapixel Sony sensors did, despite having 24 megapixels in an APS-C camera. The D810 once again has better dynamic range up to iso 3200 than other cameras out there right now.
07-23-2015, 03:54 AM - 1 Like   #10
Pentaxian




Join Date: Feb 2015
Photos: Gallery
Posts: 12,145
QuoteOriginally posted by BarryE Quote
A larger PW will fill up to max with received photons, slower than smaller PWs
Actually not. As the well capacity increase, the light sensitive exposed area also increases by the same amount. That's why for a given set of aperture and iso sensitivity, the exposure time is roughly the same regardless what the sensor size is. So, adding up signals coming from two adjacent pixel or having a double sized pixel is roughly the same, except a few percent of sensitive area lost for the additional insulation area and additional interconnects needed with smaller pixels. It's also possible to increase DR by adding up well capacity but then departing from standardized ISO values (what Nikon are doing basically). Now, keeping actual ISO constant to its standardized values, by having one large pixel instead of several adjacent smaller pixels, the gain in DR versus pixel merging is only a few percents. On the other hand, having small pixels and combining them by matrix configuration or software has the advantage of having a single techo plateform (lower NRE) and for the final camera product design, the flexibility to chose the tradeoff between better resolution or better efficiency (slightly better noise, or DR). Furthermore, having more smaller pixel, eliminate the need of optical anti-alias filter, and even after decimation, contains slightly more image information then an equivalent lower pixel count would contain.

QuoteOriginally posted by BarryE Quote
This may be limited by processor speeds and what else it has to do.
Regarding processing speed, image processors have embedded specialized DSP hardware computing blocks, hence some standard operations such as JPEG compression can be performed by hardware instead of going through a sequential machine. Typically, digital decimation is easy to implement in a pipelined FIR (Finite Impulse Response) filter fashion, so basically, there is not computing time involved, only data latency added, and since the amount of data is reduced after the decimation filter. In your current DSLR, you actually have this already implemented, for instance on the K-3, you can select JPEG 24Mpixels, 14Mpixels, 2Mpixels, and you'll see that you can even achieve a larger number of shots in burst mode.

---------- Post added 23-07-15 at 13:00 ----------

QuoteOriginally posted by Rondec Quote
I don't know the answer, but photon wells are not necessarily related to pixel size. The D7200 has significantly better dynamic range throughout its range than even the old 16 megapixel Sony sensors did, despite having 24 megapixels in an APS-C camera. The D810 once again has better dynamic range up to iso 3200 than other cameras out there right now.
Given the same sensor techno, more dynamic range with same pixel size, is achieved via a decrease of sensitivity (=different min ISO setting).

---------- Post added 23-07-15 at 13:04 ----------

QuoteOriginally posted by falconeye Quote
There are technical reasons why a high pixel count facilitates a high dynamic range.
Actually, the larger the pixel the lower the noise and the better the DR (thermal noise = K*T/C). So, increasing pixel count on same sensor area actually decreases DR. In order to counter this issue, adding extra cap in the signal path helps reduce noise , increase DR, but at the same time decreases cell sensitivity. In order to achieve more DR, D810 used lower ISO, but then the user need a tripod or image stabilization more often...

---------- Post added 23-07-15 at 13:18 ----------

Last edited by biz-engineer; 07-23-2015 at 04:20 AM.
07-23-2015, 04:26 AM   #11
Site Supporter
Site Supporter




Join Date: Jan 2013
Location: Hampshire, UK
Posts: 1,647
Original Poster
Ok Biz-engineer. That brings the understanding up a notch or two. I was trying to keep the PW idea simple and not add in any adjacent well adjustments via matrices etc. The idea that the software is interpreting the photon count in one PW, based on it's neighbour, must run the risk of areas of an image blocking up to white, without the more gradual transitions in tone that was once seen in film (I understand). I guess any image is an interpretation via hardware and software, but the aesthetics of a large sensor, as in the 645Z as I've experienced briefly, do lead me to believe that we can be a little complacent about what the software is doing, rather than relying on the purer capture of a large sensor/larger PWs, as can be seen when trying to recover image data.

Agree with your mage processor comment. Again I was keeping the 'processing' argument simple. Offloading calculations to dedicated hardware, will handle many procedures, but I guess not all and this I assume may sometimes become a restriction.

Good to be able to learn here when we are not snapping ...or post processing ... or thinking about the next lens etc.
07-23-2015, 04:34 AM - 1 Like   #12
Veteran Member
falconeye's Avatar

Join Date: Jan 2008
Location: Munich, Alps, Germany
Photos: Gallery
Posts: 6,871
QuoteOriginally posted by BarryE Quote
a smaller pixel count, surely must have a better DR, must it not ? Where might I have gone wrong ?
You're misled and long contorted answers typically won't help.

DR (of a sensor) is determined by the ratio of TOTAL number of electrons a sensor can store (total well capacity) and TOTAL read noise in electrons at some GIVEN resolution (total noise, not pixel noise).

total well capacity = #pixels * FWC

where FWC = a pixel's full well capacity and FWC is typically proportional to a pixel's surface area. Therefore total well capacity depends on the sensor size but doesn't depend on #pixels.

The Nikon D810 has a total well capacity of about 3.3 billion electrons per square mm. Which is class leading due to its ISO 64 capability. A more typical value across the industry are 2 billion electrons per square mm.

Think of the total read noise at a given resolution as a constant to keep the argument simple enough. If anything, it decreases with increasing #pixels (because of oversampling a signal), i.e., if anything, DR increases with increasing #pixels.

Last edited by falconeye; 07-23-2015 at 04:40 AM.
07-23-2015, 04:49 AM   #13
Senior Member




Join Date: Dec 2013
Posts: 223
from a cine perspective "Dynamic Range" is more important since the end result is gonna be graded which always sacrifices a bit of range to achieve a certain look, then depending on the destination will either be downsized to 1080 for broadcast, or left as is for the silver screen (tv series mr robot was recorded with a red one cine camera, thanks to its 16.5 ev dynamic range, the low light scenes look devoid of noise which is epic in my books)

personally i think colour gradients and contrast are more important than canvas size and detail, but that does not mean the guy who lives across the street from me values the same

I do notice a trend with Dynamic Range, the more a camera has it the more that camera is gonna cost
07-23-2015, 05:11 AM   #14
Veteran Member
falconeye's Avatar

Join Date: Jan 2008
Location: Munich, Alps, Germany
Photos: Gallery
Posts: 6,871
QuoteOriginally posted by Ratcheteer Quote
I do notice a trend with Dynamic Range, the more a camera has it the more that camera is gonna cost
Well, that trend ...

Today, sensors are starting to hit the wall of physical feasibility wrt quantum efficiency which reaches a stunning 65% for the A7s.

Therefore, it isn't but logical that vendors turn to dynamic range. After all, there is no physical limit to DR, it can become as high as 100 EV

That's like processors which turned from GHz to #cores ...
07-23-2015, 11:13 AM   #15
Pentaxian




Join Date: Feb 2015
Photos: Gallery
Posts: 12,145
QuoteOriginally posted by falconeye Quote
Think of the total read noise at a given resolution as a constant to keep the argument simple enough. If anything, it decreases with increasing #pixels (because of oversampling a signal), i.e., if anything, DR increases with increasing #pixels.
This statement would be true is the quantization noise + the sens amp noise would be the main contributors. It's actually the reverse situation, therefore, having more pixels on the same sensor does not increase the DR. There is still a small advantage to having larger pixels, that's why the A7s made sense. When we design a pixel, we can increase the well capacity to exceed the base value given by the exposed pixel area (decrease of actual iso), we can even switch in cap to compress the high lights and achieve a slightly better DR. If the DR increase as the pixel size decreases, then point & shots would have a higher DR then FF.

QuoteOriginally posted by falconeye Quote
Therefore, it isn't but logical that vendors turn to dynamic range. After all, there is no physical limit to DR, it can become as high as 100 EV
If dynamic range was a fallacy, then we could make a cell phone call from the earth to the moon. Unfortunately, this is not at all the case.
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!
24x36mm, camera, count, d7200, dr, filter, full-frame, hardware, image, iso, noise, pentax, pixel, pixels, post, range, size, variant
Thread Tools Search this Thread
Search this Thread:

Advanced Search


Similar Threads
Thread Thread Starter Forum Replies Last Post
Does Size Matter? A question of Megapixels. Equus17 Pentax K-30 & K-50 24 02-05-2015 07:27 PM
Question on histogram / DR LensBeginner Photographic Technique 25 02-03-2015 09:19 PM
Computer monitors only 4 megapixels or less Michael Piziak General Photography 11 08-09-2014 03:36 PM
Print size vs. megapixels of camera fevbusch Pentax K-30 & K-50 29 12-18-2013 09:21 AM
Which film has most DR & which classic Pentax cameras have 1/2000 or faster shutter? brkl Film SLRs and Compact Film Cameras 38 04-08-2009 08:16 AM



All times are GMT -7. The time now is 04:38 AM. | See also: NikonForums.com, CanonForums.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