Got my hands on the new Pentax 654Z today...

[Vladimirovich]

For my part, I am not sure DSLR CFAs are optimized for high ISO. The best optimisation for high ISO is low readout noise and fat pixels.

that was just an unfortunate wording - CFA are "optimized" not for "high ISO", but to let more light in through (naturally "blue"/"red" filters are primary targets to be less selective) - as those cameras had to be marketing-competetive in available light situations (or really competetive for actions shooters who do not care that much about proper color separation)... using higher gains just follows the available light situation (specifically for JPG shooters with raw conversion done in camera).

[ErikKaffehr]

Hi,

I am aware of that, but I have not seen any real evidence pointing in that direction for DSLRs.

DxO mark had an article discussing comparing two cameras, a Canon and a Nikon where they published spectral data for the CFA. DxO also shows something called SMI (Sensivity Metamerism Index) which is said to measure how far a sensor is from Luther-Ives condition for a given illuminant. The SMI values that DxO has measured for DSLRs is mostly higher than for for MFDBs. Some examples (taken from Tim Parkins article: https://www.onlandscape.co.uk/2012/02/the-myth-of-universal-colour/ )

Sony Alpha 87
Canon 5D   84
Canon 5DII 80
Phase One P40 80
Phase One IQ180 80
Nikon D3X  79
Leica M9 76
Phase One P45 72

These data were taken from Tim's list who argues that the SMI values measured by DxO coincide with his perception of good/accurate colour.

Best regards
Erik

that was just an unfortunate wording - CFA are "optimized" not for "high ISO", but to let more light in through (naturally "blue"/"red" filters are primary targets to be less selective)

[eronald]

Erik,


 Spectrally speaking, a lens is like a set of stacked filters. Indeed it may even contain a filter as frontal or back element, by design. So you can "assume" only if you "assume" that any filter can be taken care of white balance.

 As for the "fixes" you assume the same person did everything, and is still working there. Do you also assume every photographer who worked for a magazine has the same taste and abilities?
Edmund

Hi,

I see your reasoning.

Just adding a few points:

• I am not sure lenses affect color rendition that much because I would guess lens colour shift will be mostly well taken care of by choosing white balance.
• Regarding tuned profiles, I would presume that it takes a lot of experience to fix a pleasant profile and I guess the 'image professor' at Phase One is very good at it.

I compared the normal studio flash profile for my P45+ with the portrait profile, and they are very, very close. As far as I can recall the DeltaE just differed about 1.0 on an IT8 test target. So those profiles are very subtle work.

Best regards
Erik

[torger]

I think you're talking about different live views. MF CCD live views will not work well with NDs, so one would have to remove it while using live view and then put it on when exposing, which makes the workflow a bit more cumbersome. On a modern CMOS sensor like the live view is thanks to the great ISO performance almost like night vision, you see better with live view than with the naked eye.

I use either 8X or a Kenko variable NDX, never had problem with live view on the Sony, but I don't use ND filters in really low light as they are not needed for the exposures I want to achieve.

[torger]

skin tone are a mix of color and texture. color depends on the CFAs except in extreme shadows and highlights where camera firmware affects antiblooming settings (top shoulder) and channel noise removal (bottom clipping). Texture depends on firmware filtering and the raw converter.

That "texture" thing I find very interesting, haven't really seen an example of it so I don't know what to look for. I assume it's something you see when you zoom in to 100% rather than looking at the picture from a distance?

One thing I have noticed though is that when I zoom in my Leaf Aptus landscape images the texture I get in Capture One is more artificial and pastel-like compared to the more photographic look I get in RawTherapee. On the other hand Capture One is better at suppressing aliasing artifacts.

Attached an example, RawTherapee to the left, Capture One to the right. As always(?) when doing side-by-side testing the difference is smaller than what I had from memory, but I still think it's there, and as some would say here "it's a galaxy apart" . The rock with the lichens, the narrow branches behind it and all the tiny leaves, especially the red ones on the ground (be sure to zoom in to 100%). I think the texture is flatter and more pastel-like with Capture One, while it looks more real and photographic with RawTherapee. I haven't really compared if it's visible in a real print though.

[eronald]

That "texture" thing I find very interesting, haven't really seen an example of it so I don't know what to look for. I assume it's something you see when you zoom in to 100% rather than looking at the picture from a distance?

One thing I have noticed though is that when I zoom in my Leaf Aptus landscape images the texture I get in Capture One is more artificial and pastel-like compared to the more photographic look I get in RawTherapee. On the other hand Capture One is better at suppressing aliasing artifacts.

Attached an example, RawTherapee to the left, Capture One to the right. As always(?) when doing side-by-side testing the difference is smaller than what I had from memory, but I still think it's there, and as some would say here "it's a galaxy apart" . The rock with the lichens, the narrow branches behind it and all the tiny leaves, especially the red ones on the ground (be sure to zoom in to 100%). I think the texture is flatter and more pastel-like with Capture One, while it looks more real and photographic with RawTherapee. I haven't really compared if it's visible in a real print though.

CMOS is more intensively filtered by firmware than CCD, it appears, and so texture loss is going to hit people in MF around now, I think.

Edmund

[ErikKaffehr]

Hi,

Default noise reduction in C1 is pretty harsh. Try setting noise reduction to zero.

With LR I sometimes use median filter on a and b channels on the image converted to Lab in Photoshop to get rid of colour artifacts. Raw therapy has an option for that, too.

Best regards
Erik

That "texture" thing I find very interesting, haven't really seen an example of it so I don't know what to look for. I assume it's something you see when you zoom in to 100% rather than looking at the picture from a distance?

One thing I have noticed though is that when I zoom in my Leaf Aptus landscape images the texture I get in Capture One is more artificial and pastel-like compared to the more photographic look I get in RawTherapee. On the other hand Capture One is better at suppressing aliasing artifacts.

Attached an example, RawTherapee to the left, Capture One to the right. As always(?) when doing side-by-side testing the difference is smaller than what I had from memory, but I still think it's there, and as some would say here "it's a galaxy apart" . The rock with the lichens, the narrow branches behind it and all the tiny leaves, especially the red ones on the ground (be sure to zoom in to 100%). I think the texture is flatter and more pastel-like with Capture One, while it looks more real and photographic with RawTherapee. I haven't really compared if it's visible in a real print though.

[bjanes]

Hi,

Default noise reduction in C1 is pretty harsh. Try setting noise reduction to zero.

With LR I sometimes use median filter on a and b channels on the image converted to Lab to get rid of colour artifacts. Raw therapy has an option for that, too.

Erik,

Is that a typo? I was not aware one could do that in Lighttroom.

Regards,

Bill

[ErikKaffehr]

Bill,

Not a typo, bad writing. I open the image in Photoshop, convert to Lab, do the median filter and save the image.

There is less need to do it in C1.

Best regards
Erik

Erik,

Is that a typo? I was not aware one could do that in Lighttroom.

Regards,

Bill

[torger]

Default noise reduction in C1 is pretty harsh. Try setting noise reduction to zero.

I tested that and for that particular crop the difference is practically zero. Instead I think the in comparison lack of texture and pastel-like look in C1 is a result of the demosaicer, with RawTherapee one can cycle through a number of  demosaicers and the result is very different for this type of subject. For example the classic VNG4 does not succeed as good to keep the photographic look at 100%, unlike the default Amaze algorithm (written by Emil Martinec), which I think also does a bit better job than C1 as seen in my example. However C1's demosaicer is more robust against phenomena like crosstalk (relevant for certain tech wide + sensor combinations), while the Amaze demosaicer can start producing artifacts in those situations.

I would imagine that the demosaicer algorithm would have a considerably larger effect on the micro textures than any CMOS hardware filtering... but I don't know.

[ErikKaffehr]

Hi,

Thanks for sharing. I see your point. I have seen that kind of softness in C1 before but always attributed it to noise reduction, it seems that may not be the case.

Best regards
Erik

I tested that and for that particular crop the difference is practically zero. Instead I think the in comparison lack of texture and pastel-like look in C1 is a result of the demosaicer, with RawTherapee one can cycle through a number of  demosaicers and the result is very different for this type of subject. For example the classic VNG4 does not succeed as good to keep the photographic look at 100%, unlike the default Amaze algorithm (written by Emil Martinec), which I think also does a bit better job than C1 as seen in my example. However C1's demosaicer is more robust against phenomena like crosstalk (relevant for certain tech wide + sensor combinations), while the Amaze demosaicer can start producing artifacts in those situations.

I would imagine that the demosaicer algorithm would have a considerably larger effect on the micro textures than any CMOS hardware filtering... but I don't know.

[eronald]

I would imagine that the demosaicer algorithm would have a considerably larger effect on the micro textures than any CMOS hardware filtering... but I don't know.

Nah.

Edmund

[torger]

Nah.

Edmund

Try a few different demosaicers and watch the difference. It's easily seen. I have yet to see a demonstration of the hardware filtering effects, and undemonstrated effects makes me remain skeptical. MFD is full of often claimed and never demonstrated advantages, I'm probably guilty of repeating some of them myself... :-)

I'm not saying there can be a difference, but I find it just as likely that there is no visible impact. If you believe in a difference you will see it even if it's not there, until side by side testing shows that there is none (if that's the case). Sensors produce a pretty linear signal up to clipping, and for well exposed areas noise is low from any modern sensor.

[eronald]

Sensors produce a pretty linear signal up to clipping

afaik CMOS sensors have (firmware) adjustable shoulders thx to the antiblooming transistors. Aptina confirmed this to me at last PK.
 
Edmund

[ErikKaffehr]

Hi,

From some of my reading I got the impression that digital sensors essentially have a shoulder but the ADCs just use the linear part of the signal. I don't know if this is correct.

Best regards
Erik

Try a few different demosaicers and watch the difference. It's easily seen. I have yet to see a demonstration of the hardware filtering effects, and undemonstrated effects makes me remain skeptical. MFD is full of often claimed and never demonstrated advantages, I'm probably guilty of repeating some of them myself... :-)

I'm not saying there can be a difference, but I find it just as likely that there is no visible impact. If you believe in a difference you will see it even if it's not there, until side by side testing shows that there is none (if that's the case). Sensors produce a pretty linear signal up to clipping, and for well exposed areas noise is low from any modern sensor.

[Bart_van_der_Wolf]

afaik CMOS sensors have (firmware) adjustable shoulders thx to the antiblooming transistors. Aptina confirmed this to me at last PK

Hi Edmund,

I would not generalize that to all CMOS sensors having such circuits and settings. Most photographic sensor related ADCs output very linear quantized values. It's measurable with very simple means, and I've done it for a number of cameras, all had a linear response curve for most of the tonal range (if we eliminate noise).

Cheers,
Bart

[eronald]

Hi Edmund,

I would not generalize that to all CMOS sensors having such circuits and settings. Most photographic sensor related ADCs output very linear quantized values. It's measurable with very simple means, and I've done it for a number of cameras, all had a linear response curve for most of the tonal range (if we eliminate noise).

Cheers,
Bart

Bart,

 Maybe we should separate circuits and settings ... the fact that the circuitry is there does not mean that the firmware allows the top counts to go all the way through the shoulder into the raw files.

 My take on digital SLR cameras, after issues with highlights on my D4, is that the top stop of a saturating channel is problematic due to non linearity, the bottom 2 stops also due to noise and amp mismatch, 1 stop is often lost to channel mismatch  eg. incandescent or crepuscular illuminant, so a 14 stop camera is a 10 stop camera under less than ideal conditions, even at low ISO. Exposure lattitude is thus 1 stop each way from the center, move more and you will see damage effects at one end of the histogram. I'm sure my reasoning is wrong somewhere, but  this corresponds to what I see in practice

 I may be wrong, but my impression is that better calibration (or circuitry) and harder clipping add a stop or so to CCD digital backs.

 Interestingly, I (most people?) seem to be very sensitive to near-white highlight hues, so non-linearity of a channel at the top means bad color discrimination in the clouds and highlights and me unhappy; on the other hand, a shoulder does mean no hard burnout. I don't think ETTR really makes sense because of this, as eg. sky and cloud nuances will look very different from reality if a channel goes non-linear at the top.


Edmund

[ErikKaffehr]

Hi,

I don't see those effects on my Sonys. Actually, I think they would show up in my testing as I always test ETTR. Nikon uses sensors from Sony or Aptina mostly, but the cameras like D4 and DF use Nikon designed sensors.

Best regards
Erik

Bart,

 Maybe we should separate circuits and settings ... the fact that the circuitry is there does not mean that the firmware allows the top counts to go all the way through the shoulder into the raw files.

 My take on digital SLR cameras, after issues with highlights on my D4, is that the top stop of a saturating channel is problematic due to non linearity, the bottom 2 stops also due to noise and amp mismatch, 1 stop is often lost to channel mismatch  eg. incandescent or crepuscular illuminant, so a 14 stop camera is a 10 stop camera under less than ideal conditions, even at low ISO. Exposure lattitude is thus 1 stop each way from the center, move more and you will see damage effects at one end of the histogram. I'm sure my reasoning is wrong somewhere, but  this corresponds to what I see in practice

 I may be wrong, but my impression is that better calibration (or circuitry) and harder clipping add a stop or so to CCD digital backs.

 Interestingly, I (most people?) seem to be very sensitive to near-white highlight hues, so non-linearity of a channel at the top means bad color discrimination in the clouds and highlights and me unhappy; on the other hand, a shoulder does mean no hard burnout. I don't think ETTR really makes sense because of this, as eg. sky and cloud nuances will look very different from reality if a channel goes non-linear at the top.


Edmund

[Bart_van_der_Wolf]

Maybe we should separate circuits and settings ... the fact that the circuitry is there does not mean that the firmware allows the top counts to go all the way through the shoulder into the raw files.

Hi Edmund,

Indeed, circuitry (some of which is not present in all designs) is different from firmware settings. The settings used in the ADC process will determine how much of the total captured signal is transferred to discrete values. Noise is added to the already shotnoise riddled signal during that process. Some conversions are therefore cleaner than others.

In addition, e.g. Sony have started using a non-linear encoding of the in essence linear ADC output in their Raw format. So now it seems like the camera's response curve is vastly non-linear if one were to look at the Raw data in isolation. However, during Raw conversion this added non-linearity is removed again, because demosaicing requires a linear gamma data space.

My take on digital SLR cameras, after issues with highlights on my D4, is that the top stop of a saturating channel is problematic due to non linearity, the bottom 2 stops also due to noise and amp mismatch, 1 stop is often lost to channel mismatch  eg. incandescent or crepuscular illuminant, so a 14 stop camera is a 10 stop camera under less than ideal conditions, even at low ISO. Exposure lattitude is thus 1 stop each way from the center, move more and you will see damage effects at one end of the histogram. I'm sure my reasoning is wrong somewhere, but  this corresponds to what I see in practice

There is probably also a Raw conversion effect in play here, which is not a sensor characteristic. I rarely look at the Raw data, unless I need to analyze it before the demosaicing and conversion process turns the data into an image. During that process a lot can happen to the tonecurve, especially in the LR and ACR Process 2012 (which significantly compresses highlights), or in Capture One when a filmcurve is used instead of a linear tonecurve. But that is all postprocessing, and is not a sensor characteristic.

Interestingly, I (most people?) seem to be very sensitive to near-white highlight hues, so non-linearity of a channel at the top means bad color discrimination in the clouds and highlights and me unhappy; on the other hand, a shoulder does mean no hard burnout. I don't think ETTR really makes sense because of this, as eg. sky and cloud nuances will look very different from reality if a channel goes non-linear at the top

I'm also sensitive to highlight rendering. Using a linear tone curve instead of compressing highlights helps a lot, no, it's essential. It's easier to add a shoulder roll-off, only if and when/where it is needed, than to create linearity where the detail has already been lost. A fantastic tool like Topaz Labs Clarity helps a lot to regain some sparkle in the image by altering local contrast in a very intelligent (and halo free) way, if necessary to be applied locally by clever built-in masking tools, but it won't invent detail where there was none to begin with.

Cheers,
Bart

[Pics2]

.