a7r II .... improving color fidelity?

[Wayne Fox]

I'm not wanting to open a can of worms here (or plunge down a rabbit hole), but after reading several posts from several of you, as well as some preliminary research from Iliah, I"m curious what it might take to find a correct filter to use to perhaps improve the color fidelity of the a7rII when shooting landscapes. I'm not saying the files are bad, they are great, but even though the dynamic range is great, I struggle getting things to look like I want them to when it comes to subtle color gradations as compared to my IQ180 files.

Maybe it isn't a simple thing, and not worth the effort.  but color me curious on this one ...

[eronald]

I'm not wanting to open a can of worms here (or plunge down a rabbit hole), but after reading several posts from several of you, as well as some preliminary research from Iliah, I"m curious what it might take to find a correct filter to use to perhaps improve the color fidelity of the a7rII when shooting landscapes. I'm not saying the files are bad, they are great, but even though the dynamic range is great, I struggle getting things to look like I want them to when it comes to subtle color gradations as compared to my IQ180 files.

Maybe it isn't a simple thing, and not worth the effort.  but color me curious on this one ...

Can you link to the relevant Borg (out)posts, please?
And could you post some sort of comparison image? I'm interested in seeing where the differences are apparent to *you*.
I and others would appreciate if you could put the raw files on something like dropbox and send us the link on request.

Edmund

[AlterEgo]

I"m curious what it might take to find a correct filter

put something like CC30M, make a shot, see if raw channels are saturated equally (in rawdigger for example)... if not make it CC40M

[Tony Jay]

Can you link to the relevant Borg (out)posts, please?
And could you post some sort of comparison image? I'm interested in seeing where the differences are apparent to *you*.
I and others would appreciate if you could put the raw files on something like dropbox and send us the link on request.

Edmund

I am curious too!
I have an A7R mark II but as yet have done no meaningful shooting with it and so have not been able to form an opinion on colour fidelity.
Perhaps it could be sorted with a simple preset on import but maybe a natural light DNG camera profile might be the way to improve colour rendition.

Tony Jay

[adrian tyler]

i too only have it set up for raw. and am interested in this comparison, i get the files from my a7r ii about 80-90% right (for my taste) before post.

have you tried the "creative style" menu? it's number 5 in the camera icon menu, there is a LOT of tweaking you can do there, imo i can't see that with the options available in this menu you couldn't nail your particular necesities, then again, each to his own!

[Bart_van_der_Wolf]

I'm not wanting to open a can of worms here (or plunge down a rabbit hole), but after reading several posts from several of you, as well as some preliminary research from Iliah, I"m curious what it might take to find a correct filter to use to perhaps improve the color fidelity of the a7rII when shooting landscapes. I'm not saying the files are bad, they are great, but even though the dynamic range is great, I struggle getting things to look like I want them to when it comes to subtle color gradations as compared to my IQ180 files.

Maybe it isn't a simple thing, and not worth the effort.  but color me curious on this one ...

Hi Wayne,

There are two sides to that coin, and one of them is subtle while the other is the elephant in the room. Let's start with the elephant, the color profile that is being used. Profiles have a significant impact, and since they are all a compromise for general use, it also means that they can be tweaked/optimized for specific tasks. As an example, a specific landscape profile would be less well suited for portrait photography. There is a chance that the generic profiles do consider portraits as an important subset of the images they will be used on.

The profile can only base its adjustments on the quality of the input data. So anything that improves the input data's suitability for a certain task, will also help (but also require a new profile to exploit the new opportunities). To improve the data quality, one needs to find out where it is lacking, again, for the specific task.

Maybe, but I have not investigated that myself yet, the Magenta filter suggestion will make enough of a difference for the generic profile to perform a little better, because the Raw data gets better separation for the color nuances that matter. But the real question is whether it improves the situation when compared to an already landscape tweaked profile enough to offset the drawbacks of additional optical filters in general.

Then there is also the issue of which type of landscape we are considering (which greens/yellows/reds/browns), and how to improve the processing of it (also in post-processing).

There is ultimately also the question whether we cannot already achieve what we want by better post-processing. Of course the combination of improved capture+improved profile+improved postprocessing will offer the best possibilities, but maybe the differences with only better post-processing will be subtle at best?

Tools like Topaz Labs Clarity and Detail, but also B&W effects when used as a luminosity mask, are very well suited for such interactive and updating in realtime post-processing endeavors. The Topaz "Intellicolor" technology allows for much better saturation and brightness control than Photoshop or Lightroom offer.

Also, a raw converter like Capture One Pro offers a Color editor which can save a tweaked ICC profile for repeated use, so one can tweak one for a specific type of landscape and vegetation.

Cheers,
Bart

[eronald]

Bart, here is a Canon 1Ds3 snapshot I took a couple of weeks ago. Every shrub is a different color, even the grass shows green/yellow shadings.  This camera can be improved by a profile, because it is already pretty discriminating.

From what I've seen of the A7II the colors are good, but they kind off run together, especially in the greens. I really think the OP should post some images so we look at the issues he has.

I may not understand much of this whole color stuff, as my own profile editor is simply built on top of Photoshop. I can basically bake into a profile anything I can do with Photoshop. There are other approaches, especially the DNG proflle editor from Adobe, but when it comes to my own work, I shoot a colorchecker DC, make a profile and then edit it. The edits are usually only as good as the base profile. In this sense, I can provide a "look" for someone who doesn't have the time or experience to create his own "digital emulsion", but it's hard to fix a camera that is color blind.

Edmund

Hi Wayne,

There are two sides to that coin, and one of them is subtle while the other is the elephant in the room. Let's start with the elephant, the color profile that is being used. Profiles have a significant impact, and since they are all a compromise for general use, it also means that they can be tweaked/optimized for specific tasks. As an example, a specific landscape profile would be less well suited for portrait photography. There is a chance that the generic profiles do consider portraits as an important subset of the images they will be used on.

The profile can only base its adjustments on the quality of the input data. So anything that improves the input data's suitability for a certain task, will also help (but also require a new profile to exploit the new opportunities). To improve the data quality, one needs to find out where it is lacking, again, for the specific task.

[Bart_van_der_Wolf]

Bart, here is a Canon 1Ds3 snapshot I took a couple of weeks ago. Every shrub is a different color, even the grass shows shadings.  This camera can be improved by a profile, because it is already pretty discriminating.

Edmund, as you may know, I also use the 1DS3 as my main camera and after almost 8 years I am familiar with its capabilities and its shortcomings. I also do know, that occasionally the rendering by Adobe applications (Photoshop and Lightroom) with the default profiles, can differ quite a bit from what Capture One Pro starts of with.  It's always interesting to compare that with e.g. a Canon DPP conversion. But even that is only the beginning. Subsequent post-processing is an inevitable step towards much better image quality.

From what I've seen of the A7II the colors are good, but they kind off run together, especially in the greens and maybe green/blues.

Again, that may be due to the profile one starts with, and the subsequent post-processing to bring the file data to life.

It would take a properly exposed set of almost identical images on e.g. 2 similar resolution cameras of such colors (preferably real vegetation and sky, shot at the same time/light) to allow a more meaningful judgement. In a non-direct comparison, there will be too many variables to get a good/objective view. Most web based JPEG samples are not a good basis for developing an opinion that as (some) merit, too much (or too little) has been done to the image data.

Cheers,
Bart

[eronald]

Bart,

 We seem to agree that the OP should provide Raw sample files.

Edmund

Edmund, as you may know, I also use the 1DS3 as my main camera and after almost 8 years I am familiar with its capabilities and its shortcomings. I also do know, that occasionally the rendering by Adobe applications (Photoshop and Lightroom) with the default profiles, can differ quite a bit from what Capture One Pro starts of with.  It's always interesting to compare that with e.g. a Canon DPP conversion. But even that is only the beginning. Subsequent post-processing is an inevitable step towards much better image quality.

Again, that may be due to the profile one starts with, and the subsequent post-processing to bring the file data to life.

It would take a properly exposed set of almost identical images on e.g. 2 similar resolution cameras of such colors (preferably real vegetation and sky, shot at the same time/light) to allow a more meaningful judgement. In a non-direct comparison, there will be too many variables to get a good/objective view. Most web based JPEG samples are not a good basis for developing an opinion that as (some) merit, too much (or too little) has been done to the image data.

Cheers,
Bart

[synn]

I'm not wanting to open a can of worms here (or plunge down a rabbit hole), but after reading several posts from several of you, as well as some preliminary research from Iliah, I"m curious what it might take to find a correct filter to use to perhaps improve the color fidelity of the a7rII when shooting landscapes. I'm not saying the files are bad, they are great, but even though the dynamic range is great, I struggle getting things to look like I want them to when it comes to subtle color gradations as compared to my IQ180 files.

Maybe it isn't a simple thing, and not worth the effort.  but color me curious on this one ...

In my experience, trying to get my Sony sensored D800 to be as discriminatory about colors as my Credo 40 was a lost cause. That said, please try using C1P and see if the Phase One IQ 250 color profile makes the Sony files render better. It certainly did so for my D800 files.

[hjulenissen]

...This camera can be improved by a profile, because it is already pretty discriminating....

You seem to be suggesting that color discrimination is (always) a good thing.

My understanding is that this is not true (if your goal is one of the sensible interpretations of the term "accurate color"). A camera sensitive to extremely narrow bands centered at "L", "M" and "S" would generally not be able to capture realistic recordings of general scenes?

-h

[Bart_van_der_Wolf]

You seem to be suggesting that color discrimination is (always) a good thing.

Hi,

It may, or it may not, depending on one's creative intent. As Edmund was focusing on the green rendition, I'll show (attached ) a small example of a Profile tweak in Capture One Pro for the same camera Raw file. I increased the discrimination between greens, yellow greens, and orange yellows, by altering the Camera profile with the C1P Color Editor. The difference is small (I like to keep it realistic) but may be worthwhile. The effect should be that there is a bit more difference in the various nuances of the landscapes I have to deal with most.

This does not necessarily improve the image, but it demonstrates that a profile makes a difference. It gets a bit too saturated for this image demonstration, IMHO, if nothing else is done (like changing overall saturation as well), but that can be changed to whatever is needed. Again, profiles make a difference.

Cheers,
Bart

[Wayne Fox]

To clarify things a little, I have no side by side evidence of any of this so "files" to demonstrate it, but the discussion by many of you in other threads and observations regarding color and challenges with some colors has made me wonder if it is part of the reason I struggle with files from my cmos based 35mm format gear as compared to my MFDB gear, using both c1 and LR.  I have done many of the things suggested, and the results are fine, and it could simply be a matter of experience and comfort level with the IQ180 files, since 90% of my work is shot with that gear.

But the "logic" of my non scientific mind and the reason for the original post goes something like this .... if there is indeed color pollution of one of the channels due to the CFA's (similar to what was shown with the RED camera), is there a reliable way to identify it, and if so is there a practical way to counter it.

Of course after reading the replies this raises the question as to  whether it's worth while.

I glanced at this article and a follow up article but haven't had time to study them thoroughly.

[eronald]

Bart,

 Try it with an A7R2 file, and you'll make us all happy
 I do notice that the OP wants help or advice but is stingy with files. Photographers are not worth the trouble

Edmund

Hi,

It may, or it may not, depending on one's creative intent. As Edmund was focusing on the green rendition, I'll show (attached ) a small example of a Profile tweak in Capture One Pro for the same camera Raw file. I increased the discrimination between greens, yellow greens, and orange yellows, by altering the Camera profile with the C1P Color Editor. The difference is small (I like to keep it realistic) but may be worthwhile. The effect should be that there is a bit more difference in the various nuances of the landscapes I have to deal with most.

This does not necessarily improve the image, but it demonstrates that a profile makes a difference. It gets a bit too saturated for this image demonstration, IMHO, if nothing else is done (like changing overall saturation as well), but that can be changed to whatever is needed. Again, profiles make a difference.

Cheers,
Bart

[ErikKaffehr]

Hi Wayne,

The articles you quote go back to 2008.

I think that we have a lot of more up to date stuff. Anders Torger has made a significant effort developing a profiling tool called DCamProf and published a lot of knowledge worth reading. I would be happy to link to those articles but they are just to many.

Best regards
Erik

To clarify things a little, I have no side by side evidence of any of this so "files" to demonstrate it, but the discussion by many of you in other threads and observations regarding color and challenges with some colors has made me wonder if it is part of the reason I struggle with files from my cmos based 35mm format gear as compared to my MFDB gear, using both c1 and LR.  I have done many of the things suggested, and the results are fine, and it could simply be a matter of experience and comfort level with the IQ180 files, since 90% of my work is shot with that gear.

But the "logic" of my non scientific mind and the reason for the original post goes something like this .... if there is indeed color pollution of one of the channels due to the CFA's (similar to what was shown with the RED camera), is there a reliable way to identify it, and if so is there a practical way to counter it.

Of course after reading the replies this raises the question as to  whether it's worth while.

I glanced at this article and a follow up article but haven't had time to study them thoroughly.

[hjulenissen]

It may, or it may not, depending on one's creative intent. As Edmund was focusing on the green rendition, I'll show (attached ) a small example of a Profile tweak in Capture One Pro for the same camera Raw file. I increased the discrimination between greens, yellow greens, and orange yellows, by altering the Camera profile with the C1P Color Editor.

The post that I quoted seemed to talk about "color discrimination" of the camera color filter. My position is that having a lot of color discrimination may not be desirable (from a "neutral reproduction" perspective, it may well be exactly what the photographer ordered for a very creative shot).

In this context, it is analog to having too little spatial prefiltering in the OLPF: you get lots of apparent detail, but also lots of artifacts. If you are sampling a continous spectra using 3 primaries, making the filters too narrow is bound to cause problems that no profile can fix.

-h

[eronald]

The post that I quoted seemed to talk about "color discrimination" of the camera color filter. My position is that having a lot of color discrimination may not be desirable (from a "neutral reproduction" perspective, it may well be exactly what the photographer ordered for a very creative shot).

In this context, it is analog to having too little spatial prefiltering in the OLPF: you get lots of apparent detail, but also lots of artifacts. If you are sampling a continous spectra using 3 primaries, making the filters too narrow is bound to cause problems that no profile can fix.

-h

Luther-Ives.
The filter shapes are determined by ...
Maybe we should go back to an informed discussion of capitalism vs. socialism.
E.

[ErikKaffehr]

Hi Wayne,

I don't think Iliah's suggestion using a magenta filter is mostly intended to improve colour, but to improve image quality by balancing out the channels of the sensor so "red", "blue" and "green" have same saturation levels. That way we can easily expose ETTR and all channels get similar exposure, thus reducing noise. Let's say, a blue sky may give little signal in the red channel, so red channel may have extensive noise. Rebalancing the light to match the sensor will give similar exposures on all channels.

So, I don't think a magenta filter will improve colour reproduction, rather on the contrary, if I may guess.

As I said, Anders Torger has developed DCamProf and he freely shared his learning experience. I have not read all several hundreds posting on the related thread, but I have followed it with some interest. My take from his postings is that:

• Sensors and CFA designs may not have the importance often claimed
• Almost all colour reproduction comes from colour profiles
• Accurate rendition is needed for reproduction, like financing your spendings by making your own banknotes, but for real life tonal curves are needed.
• Tonal curves could be applied on the luminosity component in a linear space, doing so would not shift colour.
• But, this is not the way either ICC or DCP curves are made or used.
• It seems that colour profiles contain a lot of black art
• Colour profiles can cause artefacts of their own, as illustrated on this page: http://echophoto.dnsalias.net/ekr/Articles/MFDLook/
• There is a measure called SMI that measures how well a CFA design can reproduce a given set of colours
• I have a low SMI camera (P45+) and a high SMI camera (Sony Alpha 99) both can produce decently accurate colours

I have an unpublished article originally intended from Tim Parkins "OnLandscape" that touches on the issues at hand, may be an interesting read: http://echophoto.dnsalias.net/ekr/Articles/OLS_OnColor/OnColor.pdf (please note it is a draft)

So, my best take is to spend some time working with profiles. Adobe DNG Profile Editor can create a calibrated profile and has tools to manipulate that profile or any other profile.
ColorChecker Passport is a "single klick" solution. There is also QPCard that has a different set of patches.

DCamProf that Anders has built is pretty advanced and can use any target. I have used it with one of Wolfgang Faust's IT8-targets.

Shooting a profiling target is not that simple, flare and reflections need to be kept low.

Best regards
Erik

Ps. Just want to say that I always appreciated Wayne's postings, they are always helpful and written in a nice and pleasant tone!

[AlterEgo]

So, I don't think a magenta filter will improve colour reproduction, rather on the contrary, if I may guess.

so your lesser exposed images tend to have a better color than properly exposed (not talking about ill effects that some cameras have close to sensel's saturation point) with a proper profile/converter ? now you certainly need to consider building a different profile when you use filters.

[ErikKaffehr]

Hi,

Yes exactly, pretty sure that a new profile is needed. The image will give you cleaner shadows in one of the channels. You of course loose some ISO.

I have never tested, because shadow noise has never been an issue for me. You know things are getting better over time. Going back over old images I may find the newer are better.

Best regards
Erik

so your lesser exposed images tend to have a better color than properly exposed (not talking about ill effects that some cameras have close to sensel's saturation point) with a proper profile/converter ? now you certainly need to consider building a different profile when you use filters.