November 25, 2017, 04:49:19 AM

Author Topic: Smaller pixels = less DR?  (Read 4363 times)

YellowJersey

  • EOS Rebel T7i
  • ****
  • Posts: 138
Smaller pixels = less DR?
« on: September 15, 2015, 12:15:41 AM »
Hi guys, quick tech question.

 Do smaller pixels in higher resolution sensors result in less DR? I always thought that was the general rule.

 The reason I ask is that a guy in a local camera shop said that resolution has nothing to do with DR. It's been a long time since I'd thought about this, but I thought I'd ask the peanut gallery... I mean experts... yes, experts is what I meant. ;)

canon rumors FORUM

Smaller pixels = less DR?
« on: September 15, 2015, 12:15:41 AM »

3kramd5

  • EOS-1D X Mark II
  • *******
  • Posts: 1839
Re: Smaller pixels = less DR?
« Reply #1 on: September 15, 2015, 05:22:12 AM »
On a per-pixel basis, (well) size affects DR. At the sensor level, it's not that cut and dry.
Some Canon, some Nikon, some Sony, some Olympus

AcutancePhotography

  • EOS-1D X Mark II
  • *******
  • Posts: 1850
Re: Smaller pixels = less DR?
« Reply #2 on: September 15, 2015, 07:31:50 AM »
On a per-pixel basis, (well) size affects DR. At the sensor level, it's not that cut and dry.

And that is a very important distinction to keep in mind. 
I shoot with a Camera Obscura with an optical device attached that refracts and transmits light

K-amps

  • EOS-1D X Mark II
  • *******
  • Posts: 1781
Re: Smaller pixels = less DR?
« Reply #3 on: September 15, 2015, 07:53:28 AM »
The way I explain it to myself is using the Audio analogy. What is DR? I think it's the difference (or range) between highest signal level recorded without saturation and the lowest signal recorded without Noise.

So the Sensor has to be both sensitive, and be able to capture a lot of signal before saturating. This then makes one think, these would change based on the strength of the stimuli (signal or light entering the sensor). Therefore in order to standardize these measurements, a certain signal duration/strength is used to measure this, this signal should be standardized.

It would then also seem, that on the higher end, the amount of signal clip or full well capacity of the sensor will determine the upper limit, while read noise would limit the weak signal threshold of measuring DR. You then have at least 2 different phenomenon determining DR,   FWC, and Read noise (among other things).

Where does resolution come into this?

Other things being constant; Larger photo-sites will be more sensitive to the signal, helping DR, but on the other hand, will saturate quicker... this makes me think, the DR is not a function of resolution at the sensor level.

DR can be increased both by allowing a clean (noise free) signal to be recorded, which could be amplified (Exposure increase) without degradation. This is where EXMOR's are currently better. On the other hand, you could create a sensor that can record (or compress) highlights well beyond the saturation point of lessor sensors. These tricks go beyond the sensor, and are more about Signal processing... as more and more electronics are etched on to the analog Light Diode (sensor); and the more these signals and files are cooked, the more convoluted it gets...

Not sure if I am making sense or not. 
« Last Edit: September 15, 2015, 08:01:32 AM by K-amps »
EOS-80D & Pentax K1
EF Mount Rok 14mm F2.8; ∑ 24-105mm F4 A; 70-200 F/2.8L Mk.II; 100-400mm L Mk.II; EF 85mm L F/1.2 Mk. II; 100mm L F/2.8
Pentax D-FA 15-30 F2.8 & 28-105mm

C_Raven

  • PowerShot G7 X Mark II
  • **
  • Posts: 11
    • Ucronias
Re: Smaller pixels = less DR?
« Reply #4 on: September 15, 2015, 11:41:04 AM »
A short explanation is that DR depends on many things, pixel size being only a factor among many. A sensor with a higher resolution may "lose" DR due to pixel size, but may be gaining DR through other factors, so in the end, DR does not necessarily decrease with resolution.
1Ds Mark III | 24-105 f4L | 50mm 1,8 II

YellowJersey

  • EOS Rebel T7i
  • ****
  • Posts: 138
Re: Smaller pixels = less DR?
« Reply #5 on: September 15, 2015, 12:45:00 PM »
Interesting. Thanks!

 (and my apologies if this was posted in the wrong forum. None of forums seemed to clearly say, "this post should go here.")

9VIII

  • EOS-1D X Mark II
  • *******
  • Posts: 1666
Re: Smaller pixels = less DR?
« Reply #6 on: September 15, 2015, 01:02:20 PM »

Other things being constant; Larger photo-sites will be more sensitive to the signal, helping DR, but on the other hand, will saturate quicker... this makes me think, the DR is not a function of resolution at the sensor level.


A bit OT, but...
I've never read anything to say that larger photosites fill faster, and I doubt it would have any effect on DR. Filling faster wouldn't affect read noise or full-well-capacity, it would just change how long an exposure you need to reach saturation.

canon rumors FORUM

Re: Smaller pixels = less DR?
« Reply #6 on: September 15, 2015, 01:02:20 PM »

9VIII

  • EOS-1D X Mark II
  • *******
  • Posts: 1666
Re: Smaller pixels = less DR?
« Reply #7 on: September 15, 2015, 01:17:41 PM »
Edit: I'm just throwing around interesting and probably confusing numbers now, but it's cool to look at.

http://www.sensorgen.info/SonyA7R.html
http://www.sensorgen.info/SonyA7S.html

Same company, same technology, different pixel size. Notice the A7S keeps 8 stops up to ISO 102,400 vs. the A7R which only maintains 8 stops at ISO 6400.
Also note that the A7S has less DR at ISO 100 vs. the A7R.
The A7S and A7R both have more Dynamic Range.

The underlying principle that we can learn about here is that higher signal strength still gives more read noise, and possibly that you can't tune the sensor to perform optimally at reading both full and low well capacities.
« Last Edit: September 15, 2015, 01:32:33 PM by 9VIII »

rfdesigner

  • EOS 5DS R
  • ******
  • Posts: 867
Re: Smaller pixels = less DR?
« Reply #8 on: September 15, 2015, 03:01:11 PM »
Hi guys, quick tech question.

 Do smaller pixels in higher resolution sensors result in less DR? I always thought that was the general rule.

 The reason I ask is that a guy in a local camera shop said that resolution has nothing to do with DR. It's been a long time since I'd thought about this, but I thought I'd ask the peanut gallery... I mean experts... yes, experts is what I meant. ;)

Yes and no.


The sensor has "wells" or buckets to collect electrons displaced when a photon is sensed.

Larger pixels can collect more electrons before overflowing all else being equal

The dynamic range of the well equals the number of electrons it can hold.  A 5DII sensor can hold something in the region of 60,000 electrons..  that's about 15.5 bits of dynamic range.

http://www.astrosurf.com/buil/50d/test.htm

Note the well depth figure is at ISO400.. at ISO100 it's ~4x the ISO400 figure.

If you assume the 5Ds simply scales the 5DII pixels as measured above, then you'll have about 20000e / pixel which is still 14.2 bits of dynamic range (at 50Mpixels.. not scaled down to 8Mpix as with DXO)

You can use DxOmark results to work out Full well depth.

1. record ISO100 18%SNR in dB (graph must be in "screen" mode not "print" mode
2.  divide by 20
3.  Raise it to the power 10.
3.  Square it
4.  divide by 0.18

i.e. 40dB => 2 => 100 => 10000 => 55k electrons.

Result is full well depth in electrons.

The 5Ds comes in at ~26k electrons...  or 14.7 bits potential dynamic range if only the readout electronics were absolutely perfect.
« Last Edit: September 15, 2015, 03:28:51 PM by rfdesigner »
6D, 30D, 383L+, 28f1.8, 50STM, 100f2.0, 5.5"f8 scope, 12"f5 scope.

canon rumors FORUM

Re: Smaller pixels = less DR?
« Reply #8 on: September 15, 2015, 03:01:11 PM »