Pages: [1] 2 3   Go Down

Author Topic: Is there a reason for the blue/cyan bias on the Sony A7RII images?  (Read 16723 times)

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto

Hi,

This question was posted on the MFD threads:

"Is there a reason for the blue/cyan bias on the Sony A7RII images? I have seen it not just in Erik's samples but in many other's as well."

I know that some MFD users now work with Sony A7r, and would be interested to hear about their experience.

Here is my personal opinion:

  • White balance is a most important factor in colour reproduction, Sony's AWB may be cooler than other makes.
  • Colour profiles are probably more important than sensor characteristics.
  • The Alpha 900 was known to have "good colour", but I don't know if that applies to current models. There are some interesting articles discussing this: https://www.onlandscape.co.uk/2013/01/camera-colour-first-tests/

Best regards
Erik
Logged
Erik Kaffehr
 

pdp11

  • Newbie
  • *
  • Offline Offline
  • Posts: 15
Re: Is there a reason for the blue/cyan bias on the Sony A7RII images?
« Reply #1 on: December 28, 2015, 08:33:35 am »

I never used a MFD and I never use a A7R2.
My opinion is that the A7R2 color filters is really optimized for high signal at the price of color selectivity.
EX: 1ds3 and 5d2 have probabilly the same sensor, but with different color array: the first has better colors, the second lesser noise.
The same things for a5000 (better noise) and X100 (better colors).

I think that most of currents models use less selective filter array for better high iso, noise and dinamic range (and better DXO mark and dpreview).
I buyed (and quickly sold) the D800, but I found that the image quality was a big step back from D700.
An example: with d800 color folliage form different trees is flat and pritty the same, with D700 or 1ds3 is more rich.
With merrils is even richer, but at the price of some random colors.

The D800 DR is great, but colors are flats and full of shifts.

And there isn't a AWB problem: D800 has a good AWB compared to D700.

I think that color reprodution is mostly dependand form:
1) subject lighting
2) camera color filter
3) camera profiling (and photographer color management skills)

the Alpha 900 has a great color reprodution, but very poor high iso performance, that confirm my opinion.

PS:
I'm start to think that an high dxomark (that is only a noise benchmark) rating means poor sensor camera: all my preferred camera are around 80.
The color reprodution is now the most important (and less discuted) factor in a camera:
resolution, noise and dynamic range are now almost goods for all cameras, but color reprodution that is the first thing that we notice in a photo.
« Last Edit: December 28, 2015, 08:52:44 am by pdp11 »
Logged

Peter_DL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 544
Re: Is there a reason for the blue/cyan bias on the Sony A7RII images?
« Reply #2 on: December 28, 2015, 07:08:25 pm »

I never used a MFD and I never use a A7R2.
My opinion is that the A7R2 color filters is really optimized for high signal at the price of color selectivity.

EX: 1ds3 and 5d2 have probabilly the same sensor, but with different color array: the first has better colors, the second lesser noise.
The same things for a5000 (better noise) and X100 (better colors).

I think that most of currents models use less selective filter array for better high iso, noise and dinamic range (and better DXO mark and dpreview).
I buyed (and quickly sold) the D800, but I found that the image quality was a big step back from D700.
An example: with d800 color folliage form different trees is flat and pritty the same, with D700 or 1ds3 is more rich.
With merrils is even richer, but at the price of some random colors.

The D800 DR is great, but colors are flats and full of shifts.
And there isn't a AWB problem: D800 has a good AWB compared to D700.
...

PS: I'm start to think that an high dxomark (that is only a noise benchmark) rating means poor sensor camera: all my preferred camera are around 80. The color reprodution is now the most important (and less discuted) factor in a camera: resolution, noise and dynamic range are now almost goods for all cameras, but color reprodution that is the first thing that we notice in a photo.

Interesting thought,
which seems to be basically supported by the Sensitivity Metamerism Index which is reported as well by DxO (see measurements > color response), and which is 78 for the Nikon D800 vs higher 83 for the D700. Likewise with the mentioned Canon 5DII, SMI 80 vs the 1DsIII, SMI 86. But then, the Sony A7RII yields a high SMI of 85 as well.

Although DxO does not seem to be really convinced that the numbers would be meaningful: >> In practice, the SMI for DSLRs ranges between 75 and 85, and is not very discriminating. It is different for low-end cameras (such as camera phones), which typically have a SMI of about 40. For this reason, we give this measurement as an indication but do not integrate it in DxO Mark.<<

http://www.dxomark.com/About/In-depth-measurements/Measurements/Color-sensitivity

--
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: Is there a reason for the blue/cyan bias on the Sony A7RII images?
« Reply #3 on: December 29, 2015, 01:52:10 am »

Hi,

The DxO rating is probably most influenced by DR. Not really because it is more important than the other factors, but because a lot of progress has been made in that area.

The SMI is a way to measure how close the FGA (Filter Grid Array) is to Luther Ives condition for a given illuminant. A high SMI means close to Luther Ives. A good description is given here: http://dougkerr.net/Pumpkin/articles/Metameric_Error.pdf

If the spectral response of the FGA/sensor combination is known, the colour separation can be calculated. The image below is shows the colour separation of the Canon 5DII. It shows the Adobe RGB, and Pointer's gamut. Pointer's gamut is essentially all the non specular colours occurring in nature.

Taken from Anders Torger's: http://www.ludd.luth.se/~torger/dcamprof.html

What I essentially want to say is that it is to easy to make assumptions, but reality is more complex.

Anders Torger, the author of DCamProf, takes an interesting approach, he first creates a "neutral tone operator" and than adds a specific look to profile:
http://www.ludd.luth.se/~torger/dcamprof.html#design_look

I also add two images shot at the same time with a Canon 5DIII and a Sony A99, I did not have my A7rII at that time, but I would expect that the A7rII is pretty similar to the A99 in colour rendition. Sony left and Canon right.

Best regards
Erik






Interesting thought,
which seems to be basically supported by the Sensitivity Metamerism Index which is reported as well by DxO (see measurements > color response), and which is 78 for the Nikon D800 vs higher 83 for the D700. Likewise with the mentioned Canon 5DII, SMI 80 vs the 1DsIII, SMI 86. But then, the Sony A7RII yields a high SMI of 85 as well.

Although DxO does not seem to be really convinced that the numbers would be meaningful: >> In practice, the SMI for DSLRs ranges between 75 and 85, and is not very discriminating. It is different for low-end cameras (such as camera phones), which typically have a SMI of about 40. For this reason, we give this measurement as an indication but do not integrate it in DxO Mark.<<

http://www.dxomark.com/About/In-depth-measurements/Measurements/Color-sensitivity

--
Logged
Erik Kaffehr
 

Peter_DL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 544
Re: Is there a reason for the blue/cyan bias on the Sony A7RII images?
« Reply #4 on: December 29, 2015, 06:46:16 am »

If the spectral response of the FGA/sensor combination is known, the colour separation can be calculated. The image below is shows the colour separation of the Canon 5DII. It shows the Adobe RGB, and Pointer's gamut. Pointer's gamut is essentially all the non specular colours occurring in nature.
http://www.ludd.luth.se/~torger/img/ssf_csep.png
Taken from Anders Torger's: http://www.ludd.luth.se/~torger/dcamprof.html

What I essentially want to say is that it is to easy to make assumptions, but reality is more complex.

Hi,

I think it basically quite simple:

There is a fundamental antagonism between: a) low noise and b) meeting the Luther-Ives condition,
and the better b) is fulfilled the easier it is for Raw converter like LR/ACR to reconstruct "accurate" color (scene-referred, colorimetric) and finally to derive what is called: good color.

What pdp11 describes above is essentially in line with my experience with LR/ACR and different cameras over the past decade. Some cameras "interact" better with LR/ACR than others in terms of good color which finally means less or more work in post processing (and with dng-profiling, if you want so).

Does it correlate with DxO's simple Sensitivity metamerism index ? You are probably right that we need a more sophisticated measure here, maybe like the u'v' heat map as shown above. But then, I also see some remarkable correlation of DxO's SMI with my empirical rating of my cameras' colors in LR/ACR.

--


Logged

Jack Hogan

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 798
    • Hikes -more than strolls- with my dog
Re: Is there a reason for the blue/cyan bias on the Sony A7RII images?
« Reply #5 on: December 30, 2015, 04:36:34 am »

You are probably right that we need a more sophisticated measure here, maybe like the u'v' heat map as shown above. But then, I also see some remarkable correlation of DxO's SMI with my empirical rating of my cameras' colors in LR/ACR.

Why don't they just measure each channel's relative absolute QE vs wavelength and publish that instead of these milkshake indicators?  It can't be that hard to do it professionally if a hobbyist can give it a rough and tumble try for $10.

Jack
« Last Edit: December 30, 2015, 04:42:58 am by Jack Hogan »
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: Is there a reason for the blue/cyan bias on the Sony A7RII images?
« Reply #6 on: December 30, 2015, 06:05:23 am »

Jack,

Impressed by your ingenuity…

Best regards
Erik

Why don't they just measure each channel's relative absolute QE vs wavelength and publish that instead of these milkshake indicators?  It can't be that hard to do it professionally if a hobbyist can give it a rough and tumble try for $10.

Jack
Logged
Erik Kaffehr
 

BrianVS

  • Full Member
  • ***
  • Offline Offline
  • Posts: 164
Re: Is there a reason for the blue/cyan bias on the Sony A7RII images?
« Reply #7 on: December 30, 2015, 08:06:30 am »

The A7R2 CMOS sensor is Back-Side illuminated, meaning that light does not have to pass through the wiring layer of the chip before being absorbed. Shorter wavelength light is absorbed first in the silicon wafer, red light is absorbed deepest in the sensor. I am guessing that BSI sensors are more efficient in acquiring shorter wavelengths of light than front-side illuminated sensors, and this may produce a bias towards blue/cyan.

PDP-11/35, PDP-11/34... used the latter for image processing. PDP-8, first film scanner. But did much more with the VAX 11/780, 11/750, and 11/730.
« Last Edit: December 30, 2015, 08:10:42 am by BrianVS »
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: Is there a reason for the blue/cyan bias on the Sony A7RII images?
« Reply #8 on: December 30, 2015, 09:22:31 am »

Hi,

Would that be the case, I would think WB and profiling would take care of that.

My guess is that it may be another myth, or it could be that auto WB on Sony's may be a bit on the cold side. As an old Sony user I have found that doing WB on a grey card generally makes the images "warmer" than Sony's WB.

Best regards
Erik

The A7R2 CMOS sensor is Back-Side illuminated, meaning that light does not have to pass through the wiring layer of the chip before being absorbed. Shorter wavelength light is absorbed first in the silicon wafer, red light is absorbed deepest in the sensor. I am guessing that BSI sensors are more efficient in acquiring shorter wavelengths of light than front-side illuminated sensors, and this may produce a bias towards blue/cyan.

PDP-11/35, PDP-11/34... used the latter for image processing. PDP-8, first film scanner. But did much more with the VAX 11/780, 11/750, and 11/730.
Logged
Erik Kaffehr
 

BrianVS

  • Full Member
  • ***
  • Offline Offline
  • Posts: 164
Re: Is there a reason for the blue/cyan bias on the Sony A7RII images?
« Reply #9 on: December 30, 2015, 11:07:01 am »

Looking at the spectral response of other Sony BSI sensors, the peaks are shifted towards shorter wavelengths as compared with their Front-Side illuminated sensors. Blue response is increased, red response is decreased. What I would expect given the technology. You can compensate using white balance, which would scale down the Blue channel compared with Red. This is probably where Sony decided to get as many photons as possible out of the sensor and let the user color balance the image later. Easier to decrease channel values after the light has been collected.

http://www.sony.net/Products/SC-HP/new_pro/april_2014/imx214_e.html

There certainly is a blue/cyan shift for the Sony BSI sensor that they actually publish the response curves.

Is the spectral response of the A7R2 sensor a trade secret for Sony? Many sensor manufacturers publish this data.
« Last Edit: December 30, 2015, 11:17:40 am by BrianVS »
Logged

AlterEgo

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1995
Re: Is there a reason for the blue/cyan bias on the Sony A7RII images?
« Reply #10 on: December 30, 2015, 12:47:49 pm »

Looking at the spectral response of other Sony BSI sensors, the peaks are shifted towards shorter wavelengths as compared with their Front-Side illuminated sensors. Blue response is increased, red response is decreased. What I would expect given the technology. You can compensate using white balance, which would scale down the Blue channel compared with Red. This is probably where Sony decided to get as many photons as possible out of the sensor and let the user color balance the image later. Easier to decrease channel values after the light has been collected.

http://www.sony.net/Products/SC-HP/new_pro/april_2014/imx214_e.html


I am not sure what you are comparing here ?

the URL has the following graph :



for 2 sensors : "Solid line: IMX214  Dashed line: IMX135"

and both sensors are BSI : IMX135 = "Back-illuminated process" and IMX214 = "Back-illuminated process + reduced height structure"

so it is BSI vs BSI... not FSI against BSI

plus you forget CFA - whatever difference in silicone in terms of what photones do pass might be totally altered by a different CFA recipe...

plus as people know P1 for example simple reused A7R profile for A7R2 in C1 (P1 does that all the time for dSLRs/dSLMs) ... and profile can totally alter things.

Logged

Jack Hogan

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 798
    • Hikes -more than strolls- with my dog
Re: Is there a reason for the blue/cyan bias on the Sony A7RII images?
« Reply #11 on: December 30, 2015, 12:53:04 pm »

There certainly is a blue/cyan shift for the Sony BSI sensor that they actually publish the response curves.

Are you referring to Figure 2?  Where is the shift in that filterless diagram?
Logged

AlterEgo

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1995
Re: Is there a reason for the blue/cyan bias on the Sony A7RII images?
« Reply #12 on: December 30, 2015, 12:57:52 pm »

filterless

you mean sensor w/o IR/UV cut glass on top of it ?
Logged

AlterEgo

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1995
Re: Is there a reason for the blue/cyan bias on the Sony A7RII images?
« Reply #13 on: December 30, 2015, 01:04:27 pm »

http://www.vision-systems.com/articles/2011/11/backside-illuminated-image-sensors-optimizing-manufacturing-for-a-sensitivity-payoff.html



"Quantum efficiencies over the visible spectrum, from bottom to top: (1) FSI imager without microlenses, (2) FSI imager with microlenses, (3) state-of-the-art BSI, (4) BSI from imec with record efficiency."

so you can optimize both ends in visible (range from ~380... ~730 shall cover it) and further alter with CFA and IR/UV cut filters on top of that

Logged

Jack Hogan

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 798
    • Hikes -more than strolls- with my dog
Re: Is there a reason for the blue/cyan bias on the Sony A7RII images?
« Reply #14 on: December 30, 2015, 01:32:19 pm »

you mean sensor w/o IR/UV cut glass on top of it ?

Right, as used.  To indicate a shift one should show sensor1+filters 'unshifted' and sensor2+filters 'shifted'.  That does look like a very wide green channel.

Jack
Logged

BrianVS

  • Full Member
  • ***
  • Offline Offline
  • Posts: 164
Re: Is there a reason for the blue/cyan bias on the Sony A7RII images?
« Reply #15 on: December 30, 2015, 03:17:34 pm »

Compare the BSI spectral response curves with traditional CMOS- Blue response is higher, and in the newer sensor shown- QE numbers for Blue increased, Red decreased.

Seems they are going for increased QE across the spectrum. With the High-ISO performance of cameras being the latest metric for Hot Rodding, seems they will optimize for High ISO. That means maximize QE and minimize the filters in the CFA. True separation filters would cut out 2 stops. No CFA filters comes close to that.

Anyway- BSI technology means placing the photosensitive layers closer to the surface of the chip. Blue light is absorbed close to the surface. The photosensitive portion is now closer to the surface and is more efficient at absorbing blue light. Absorbing light in the in the CFA cuts into the QE of the completed sensor package, means lower ISO performance. You can correct the color balance in post, but leave the photon count in the RAW image. You can always use a color correction filter to warm the image if you don't like what the CFA is doing.
Logged

eronald

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6642
    • My gallery on Instagram
Re: Is there a reason for the blue/cyan bias on the Sony A7RII images?
« Reply #16 on: December 30, 2015, 04:56:38 pm »

Yes, with mirrorless there is no reason not to use some filter. Whether a single filter can improve separation  is a different question.

My old 1Ds3 is my main camera, trees and shrubs in woods render much more subtly than anything I have seen from the A7R2 but the rez and DR are not in the same class.

Edmund

Compare the BSI spectral response curves with traditional CMOS- Blue response is higher, and in the newer sensor shown- QE numbers for Blue increased, Red decreased.

Seems they are going for increased QE across the spectrum. With the High-ISO performap)nce of cameras being the latest metric for Hot Rodding, seems they will optimize for High ISO. That means maximize QE and minimize the filters in the CFA. True separation filters would cut out 2 stops. No CFA filters comes close to that.

Anyway- BSI technology means placing the photosensitive layers closer to the surface of the chip. Blue light is absorbed close to the surface. The photosensitive portion is now closer to the surface and is more efficient at absorbing blue light. Absorbing light in the in the CFA cuts into the QE of the completed sensor package, means lower ISO performance. You can correct the color balance in post, but leave the photon count in the RAW image. You can always use a color correction filter to warm the image if you don't like what the CFA is doing.
Logged
If you appreciate my blog posts help me by following on https://instagram.com/edmundronald

AlterEgo

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1995
Re: Is there a reason for the blue/cyan bias on the Sony A7RII images?
« Reply #17 on: December 30, 2015, 05:20:13 pm »

Compare the BSI spectral response curves with traditional CMOS

sure - please post a link where ?

- Blue response is higher, and in the newer sensor shown- QE numbers for Blue increased, Red decreased.

as other parts of the spectrum (see my link above) - the manufacturer can control what and where will be increased and you again forget to include the effect of 1) CFA 2) IR/UV cut filters 3) camera profiles...

True separation filters would cut out 2 stops.

and the proof link is where ? why 2 and not 1.5 or 2.5 ?

Anyway- BSI technology means placing the photosensitive layers closer to the surface of the chip. Blue light is absorbed close to the surface.

yes and you can design thing with even more increase in "red" part



so why do you think that Sony didn't increase QE across the spectrum ?
Logged

Jack Hogan

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 798
    • Hikes -more than strolls- with my dog
Re: Is there a reason for the blue/cyan bias on the Sony A7RII images?
« Reply #18 on: December 31, 2015, 05:53:15 am »

My old 1Ds3 is my main camera, trees and shrubs in woods render much more subtly than anything I have seen from the A7R2...

You may very well be right, but until someone posts comparable Spectral Response (absolute QE in Kodak lingo) diagrams of the sensors in question as used in those two cameras all we've got to go on are the SMIs, which say the two cameras are comparable.  I think there is a lot of room for confusion and misunderstanding (especially on my part:-) where color is concerned.  For instance, in order to minimize the green discrimination problem you seem to be having with the a7RII in the woods, which of these two CFAs would you prefer?



C'mon Edmund, buy you a beer if you play ;)

Jack
Logged

Zorki5

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 486
    • AOLib
Re: Is there a reason for the blue/cyan bias on the Sony A7RII images?
« Reply #19 on: December 31, 2015, 10:36:38 am »

Would that be the case, I would think WB and profiling would take care of that.

My guess is that it may be another myth, or it could be that auto WB on Sony's may be a bit on the cold side. As an old Sony user I have found that doing WB on a grey card generally makes the images "warmer" than Sony's WB.

Ditto. Whatever is the difference, I believe it's deliberate. I'm also a long time Sony user, my first digital camera from them was DSC F717, and my observations match yours.

Not that it's directly related, but their Trinitron (CRT) monitors were quite noticeably "cooler" than pretty much anything else. I remember I was once choosing between a Trinitron-based monitor from Sony, and an iiyama... I had both delivered, and was choosing between the two having them side-by-side. I was... "agonizing" until I displayed an explosion shot on both. WOW. On iiyama, it was an EXPLOSION, I could almost feel the heat! On Sony... well, I just sent it back.
Logged
Pages: [1] 2 3   Go Up