Pages: [1]   Go Down

Author Topic: Resolution vs image quality Mark II  (Read 2133 times)

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10625
    • Echophoto
Resolution vs image quality Mark II
« on: February 23, 2017, 02:04:02 AM »

Hi,

Pixel density goes up and image quality goes up, is that possible?

To begin with, how do we define image quality? A definition would be faithful rendition of the original. A rendition should add, remove or modify as little as possible to original image.

 
  • Proper resolution is obviously is a part of the equation. We need a sensor that can resolve the image cast by the lens.
  • Having a sensor that cannot resolve all the detail the lens yields will produce artefacts, known as aliases.
  • Noise is also a modification of the image. Just as an example, has anyone seen a grainy sky in real life?
  • We may also need to reproduce the deepest shadows, that requires a sensor with good dynamic range.

So we want a sensor with high resolution, matching the lens. If we buy a world class lens we probably also want to have a world class sensor. But it has been shown that sensor resolution also helps lenses that are not so great. The explanation is easy:

We can look at the MTF of the system. It is something like this:

MTF_system = MTF_lens * MTF_aperture *  MTF_olp_filter * MTF_pixel_aperture


  • MTF_aperture corresponds to diffraction
  • MTF_pixel_aperture corresonds to the smearing represented by sensitive area of the pixel

Reducing pixel pitch will increase the MTF_pixel_aperture factor and also the MTF_olp_filter will increase. So whatever a lens, MTF will improve with smaller pixels.

Reducing pixel size will affect noise at pixel level. Pixel noise is mostly determined by the number of photons detected by the pixel. If we reduce pixel dimension to half, the area will be a fourth of the larger pixel. Shot noise is proportional to the square root (SQRT) of the detected photons. So a pixel of half diameter will have twice the noise. But we have four times the number of pixels. So noise level over a sensor will not be affected.

There is a limitation to this. Part of the pixel area is made up from transistors and connections and those areas don't capture light. What happens is that design rules are reduced. Older sensors may have been produced for 0.050 micron design rules but now 0.018 micron design rules are common. Also, the electrical design is simplified, so that much less of the pixel area  is used by gates and interconnects.

All this makes that optimum pixel size shrinks.

But, there is a limitation and that is affecting both DR and high ISO. In the darkest parts, noise is dominated by readout noise and readout noise acts differently from shot noise. Halving the diameter of a pixel we could replace a large pixel with four small ones and still have the same shot noise but at a cost of one EV in DR.

Many photographers are skeptical of things measurable, but DxO produces a lot of quite solid data. The first attachment shows SNR for three different cameras. The Nikon D700, D750 and D810. The cameras have 12, 24 and 36 megapixels. The D750 and the D810 are quite close, indicating that they capture the same number of photons for a given exposure. The D810 reaches lower ISO, which indicates it has higher full well capacity.

So, in this case development makes the sensor better although it has significant increase in megapixels.

If we look at DR the lead of the Nikon D810 is even cleaner. DR depends both on number of captured photons (Full Well Capacity) and readout noise. The major development is in readout noise. The D700 uses old technology and it has high readout noise at low ISO. The short explanation is external converters. At high ISO analogue gain is increased before the external converters, which gives good high ISO performance. It is the low ISO end that suffers for old technology.

The DxO figures shown here are normalised for a given print size. So, this would be what we would see in a print, or viewing an image on screen. Would we view the image at actual pixels the trend would be different.

Check also the link below:

http://photonstophotos.net/Charts/PDR.htm#Nikon%20D700,Nikon%20D750,Nikon%20D810

Best regards
Erik



shadowblade

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1835
Re: Resolution vs image quality Mark II
« Reply #1 on: February 23, 2017, 02:25:56 AM »

Just as an example, has anyone seen a grainy sky in real life?


I've been in a sandstorm...
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10625
    • Echophoto
Re: Resolution vs image quality Mark II
« Reply #2 on: February 23, 2017, 02:30:00 AM »

Thanks! Good example! Gotta any great pictures?

Best regards
Erik

I've been in a sandstorm...

shadowblade

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1835
Re: Resolution vs image quality Mark II
« Reply #3 on: February 23, 2017, 02:32:58 AM »

Thanks! Good example! Gotta any great pictures?

Best regards
Erik

Nope - I was hunkered down inside a tent.
Logged

Telecaster

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2093
Re: Resolution vs image quality Mark II
« Reply #4 on: February 23, 2017, 04:35:57 PM »

In the Middle East a Khamsīn (dry, dusty/sandy windstorm) can be fierce. A friend and I got caught out by one while on an afternoon run c. spring 1984. We headed straight for a field of sorghum, ducked down and turned our backs to the wind, then emerged 5–10 minutes later soaked in dust. With patches of red, abraded skin.

Grain, OTOH, is a photographic artifact I've always liked.  :)

-Dave-
Logged

razrblck

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 479
  • Chill
    • Portfolio
Re: Resolution vs image quality Mark II
« Reply #5 on: February 24, 2017, 04:41:26 AM »

Grain, OTOH, is a photographic artifact I've always liked.  :)

Same here, especially on very flat patches of color it gives them a more interesting and pleasant look in my view.
Logged
Instagram (updated often)

Mousecop

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 73
Re: Resolution vs image quality Mark II
« Reply #6 on: March 19, 2017, 07:00:58 PM »

Hi,

Pixel density goes up and image quality goes up, is that possible?

Yes.

Kind of a proven fact by now.

Over the years, we've increased MP counts, while increasing resolution, dynamic range, noise performance, high ISO performance, speed and more. E.g. Canon went from 13mp to 50mp in its 35mm sensor cameras, and there is no question that the results have improved significantly by almost every measure.

A lot of this is that sensor and other technology has advanced during this time; and in most cases, manufacturers won't increase MP counts unless it at least breaks even in terms of other parameters. If Canon put out a 100mp sensor using the same technology as the current 5DS, it would likely experience more shot noise and more diffraction.

If they continue to increase MP counts, they will either need to classify those cameras as speciality items, or admit that it will show more noise and diffraction. As it stands, Canon is already marketing the 5DS as a speciality camera, and Nikon's D500 deliberately kept the sensor to 21mp rather than max out the MP count.

So yes, it is theoretically possible that manufacturers could reduce pixel pitch and resolution, while sacrificing other qualities. In practice, I seriously doubt they will increase MP counts for consumer grade models unless they can mitigate most of those issues.
Logged

rdonson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2355
Re: Resolution vs image quality Mark II
« Reply #7 on: March 20, 2017, 11:59:35 AM »

Erik, isn't there some point of diminishing returns?  For example, would you think a M4/3 sensor with a 100 MP and a great lens really do a great job?
Logged
Regards,
Ron

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10625
    • Echophoto
Re: Resolution vs image quality Mark II
« Reply #8 on: March 20, 2017, 03:04:49 PM »

Hi,

It may depend on priorities a bit… Let's look at some graphs

http://photonstophotos.net/Charts/PDR.htm#Canon%20EOS%205D%20Mark%20III,Canon%20EOS%205DS%20R,Leica%20S-E%20(Typ%20006),Sony%20ILCE-6300


So, here we compare two generations of the Canon 5D, a medium format camera using a 45x30 mm CCD sensor and an APS-C camera with a Sony Exmoor sensor.

Just to say, Bill Claff's photographic DR is just one measure of image quality. On the other hand, it may be argued that DR is the parameter most affected by pixel size.

What we see in the above figure is that the Canon 5DsR has better DR than the older 5DIII. It is also mostly superior to the Leica S2 that uses a larger CCD device of older design.

The APS-C sized Sony Exmoor outperforms the larger sensors up to 200 ISO.

Now, let's drop some more sensors in that comparison:
http://photonstophotos.net/Charts/PDR.htm#Canon%20EOS%205D%20Mark%20IV,Canon%20EOS%205DS%20R,Sony%20ILCE-7RII,Sony%20ILCE-7S


So, now we can see that the Canon 5DIV is playing in the same division as the Sony's albeit having less pixels.

Let's now ignore the money aspect and get some MFD stuff in the equation:
http://photonstophotos.net/Charts/PDR.htm#Canon%20EOS%205DS%20R,Pentax%20645Z,Phase%20One%20IQ260,Phase%20One%20IQ3100,Sony%20ILCE-7RII


Here we can see that old CCD backs struggle to keep up with the EOS 5DsR. The Pentax 645Z using a small CMOS sensor is keeping ahead the the smaller sensor competition. The large Phase One IQ3100 sensor plays in a different division although it has relatively small pixels.

Now, let's look at an actual pixel view of the DPReview studio scene:
http://echophoto.dnsalias.net/ekr/Articles/PixelSizes/DPReview1.png


The IQ3 100MP back outresolves the sensor and the subject and yields proper rendition.

Or we can look at some fine detail in the comparison view at high ISO:
http://echophoto.dnsalias.net/ekr/Articles/PixelSizes/DPReview2.png


In the near future (1-2 years) I guess we will see 70+ MP sensors in 24x36 mm and 100+ MP sensors in 44x33mm, that would be my guess.

Best regards
Erik








Erik, isn't there some point of diminishing returns?  For example, would you think a M4/3 sensor with a 100 MP and a great lens really do a great job?

rdonson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2355
Re: Resolution vs image quality Mark II
« Reply #9 on: March 20, 2017, 05:10:26 PM »

As pixel size decreases in an effort to increase the number of pixels on a given size sensor doesn't the noise increase?  Isn't noise another measure of photo quality? 

These charts seem at odds with DXO results or did I miss something?
Logged
Regards,
Ron

pcgpcg

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 397
    • paulglasser
Re: Resolution vs image quality Mark II
« Reply #10 on: March 20, 2017, 06:03:41 PM »

...isn't there some point of diminishing returns?...
I'm no expert, but it appears from looking at the equation, from a purely mathematical perspective, returns do diminish because the MTF of the sensor can never be greater than one (100%), at least at the spatial frequency for which it is normalized. In other words, there is a ceiling which you approach, and no amount of sensor advancement will get you past that point. If you are already at an MTF of 80%, for example, the best you can ever do is to approach 100%.
Is this correct?
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10625
    • Echophoto
Re: Resolution vs image quality Mark II
« Reply #11 on: March 21, 2017, 02:03:27 AM »

Hi,

The noise per pixel increases, but we have more pixels and that essentially compensates for that. So noise levels are not affected until wiring to photosensitive area increases to much. So, say facial features in decently exposed picture will have the same noise on say a 42 MP sensor as long as you look at the same magnification. If you pixel peep at actual pixels on the screen the small pixel image will be more noisy.

The included samples indicate this clearly, the tonal range of the A7rII exceeds the tonal range of the A7sII up to 25600 ISO in the "Print mode" while the A7sII has a great advantage in the per pixel "Screen mode".

Bill Claffs data doesn't report "engineering DR", like DxO-mark does but more like a DR based on Signal Noise Ratio like (SNR = 8) and it is also normalised. It would be more like the tonality curve on DxO.

Also enclosed are two comparisons from the DPReview Studio scene, one at actual pixel crops the other at comparison scale. Note that the 6400 ISO images in the "comparison sample" are pretty close in noise.

Bill Claffs data

As pixel size decreases in an effort to increase the number of pixels on a given size sensor doesn't the noise increase?  Isn't noise another measure of photo quality? 

These charts seem at odds with DXO results or did I miss something?

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10625
    • Echophoto
Re: Resolution vs image quality Mark II
« Reply #12 on: March 21, 2017, 02:46:00 AM »

Hi,

MTF is always a function of frequency. So a system that perhaps reaches 90% MTF at 10 lp/mm may only reach say 70% MTF at 40 lp/mm.

Regarding sensors, it is the pixel aperture that decides the MTF. So, small fill factor increases MTF and so does decreasing pixel size. But, most of that change is probably at high frequencies.

The great benefit of decreasing pixel sizes may that aliasing is reduced.

There is real downside to small pixels, and that is that the amount of data increases. That may be one of the reasons that Canon 1DXII and Nikon 5D have moderate pixel counts you cannot achieve 16 FPS on 50 MP SLR.

Another way to make use of smaller pixels is to reduce the sensor size. Making the sensor smaller reduces SNR, but make it possible to build small or flexible systems.

Best regards
Erik


I'm no expert, but it appears from looking at the equation, from a purely mathematical perspective, returns do diminish because the MTF of the sensor can never be greater than one (100%), at least at the spatial frequency for which it is normalized. In other words, there is a ceiling which you approach, and no amount of sensor advancement will get you past that point. If you are already at an MTF of 80%, for example, the best you can ever do is to approach 100%.
Is this correct?
Pages: [1]   Go Up