Color with the Fuji GFX as Compared to the Hasselblad X1D

[Bart_van_der_Wolf]

MT: “If they [Fuji] stick to the same color profile as their other cameras, honestly, no: the blues always run cyan, and the shadows block up quickly to extend highlights. This may be different on the GFX sensor, but I don’t think it’s going to solve the cyan shift since that seems to be part of the family ‘look’.”
...
I hope these comments are helpful. Please don’t bother to disparage Ming Thein’s comments just because he is loosely associated with Hasselblad. With some folks, bias may be true, but I don’t believe it is true in Thein’s case. I read what he says. Either Fuji has blues which tend to run cyan, etc. or not.

Hi Michael,

I think Ming Thein's observations are reasonable, but they do not take custom profiles into account. Camera manufacturers do add a specific 'look' to the images, but it doesn't have to stop there.

Cheers,
Bart

[Michael Erlewine]

Hi Michael,

I think Ming Thein's observations are reasonable, but they do not take custom profiles into account. Camera manufacturers do add a specific 'look' to the images, but it doesn't have to stop there.

Cheers,
Bart

Bart,

What you are suggesting is that the more "vivid" style of Fuji, if it exists, can be entirely removed by a custom profile? And that, with a profile, I could back off the saturation (or whatever) and attain the less vivid Hasselblad style?

[ErikKaffehr]

Hi,

Probably yes.

Keep in mind a raw file has no colour, it has normally four monochrome channels, these are either clipped or not. An image that contains clipped data is overexposed.

So, with a correctly exposed image we have four unclipped channels that are just numbers.

The colour is carried in the colour profile, at least as long as you use the Adobe processing pipeline. The profile contains the recipe for calculating colour. So, as long you work with raw files all the colour info is generated in the raw processor.

So talking about saturated or clipped colour in the raw file is pure nonsense. The colour profile can generate saturated colour channels.

A test I made illustrates this:

	Lightroom	Capture One
P45+
Sony Alpha 99

The samples below are based on spectrometer readings of the petals:


What I think the examples show in this case is my Phase One P45+ back renders similar to my Sony Alpha 99 when using Lightroom, but colours in Capture One are rendered differently.

Just to say, bluish purple is a difficult colour to render. In this case Adobe's pipeline handles it better than Capture Ones.

Best regards
Erik

Bart,

What you are suggesting is that the more "vivid" style of Fuji, if it exists, can be entirely removed by a custom profile? And that, with a profile, I could back off the saturation (or whatever) and attain the less vivid Hasselblad style?

[Michael Erlewine]

Hi,

Probably yes.

Keep in mind a raw file has no colour, it has normally four monochrome channels, these are either clipped or not. An image that contains clipped data is overexposed.

So, with a correctly exposed image we have four unclipped channels that are just numbers.

The colour is carried in the colour profile, at least as long as you use the Adobe processing pipeline. The profile contains the recipe for calculating colour. So, as long you work with raw files all the colour info is generated in the raw processor.

So talking about saturated or clipped colour in the raw file is pure nonsense. The colour profile can generate saturated colour channels.

OK, I understand. Given a carefully made profile I can ignore or not-use whatever penchant Fuji has for over-saturation and just do my own thing.

I always and only shoot raw, so that should good.

[ErikKaffehr]

Yes,

That is true. But, there are differences between colour sensors that cannot be handled by profiles.

You can read this:
http://www.fredmiranda.com/forum/topic/1234124/2&year=2013

But at that time most MFDs used Kodak CCD sensors, now they mostly use Sony CMOS sensors.

I started a long thread about Capture One and LR6 on GetDPI and a lot of knowledgeable persons chimed in:
http://www.getdpi.com/forum/medium-format-systems-and-digital-backs/59120-capture-one-lr6.html

Best regards
Erik Kaffehr

OK, I understand. Given a carefully made profile I can ignore or not-use whatever penchant Fuji has for over-saturation and just do my own thing.

I always and only shoot raw, so that should good.

[Michael Erlewine]

Yes,

That is true. But, there are differences between colour sensors that cannot be handled by profiles.

Is this true even through both the X1D and the GFX use the same sensor... and I am shooting raw?

[ErikKaffehr]

Hi,

It could be that Hasselblad and Fuji use different CFA designs, with different spectral responses. That would still be possible to handle by profiles, mostly.

The example I posted, I shot both with P45+, that has a quiet extreme CFA design from Kodak and with the Sony Alpha 99 which has a typical Sony CFA. The processing in Lightroom was very close, while Capture One produced very different results. It is an example that profiles are more important than sensors and CFAs.

Best regards
Erik

Is this true even through both the X1D and the GFX use the same sensor... and I am shooting raw?

[Michael Erlewine]

Hi,

It could be that Hasselblad and Fuji use different CFA designs, with different spectral responses. That would still be possible to handle by profiles, mostly.

The example I posted, I shot both with P45+, that has a quiet extreme CFA design from Kodak and with the Sony Alpha 99 which has a typical Sony CFA. The processing in Lightroom was very close, while Capture One produced very different results. It is an example that profiles are more important than sensors and CFAs.

Best regards
Erik

I believe it. I received camera profiles for LR and ACR from Ming Thein as part of his Series III workflow, and the one for the D810 was very helful. He has used so many cameras that he has profiles for most of the ones that would ever interest me.

[hubell]

Color science and camera profiling are not, to put it mildly, my areas of expertise. However, two things occur to me. First, I would think it is very difficult to get the color you get out of the GFX to look exactly like the color from files from the X1D. In order to make them "the same", I would think you would have to use camera profiling software to profile the two cameras in the same raw converter and change the profiles of both of them to look the same. That's very different from making the Fuji file look exactly like the Hasselblad file. Moreover, what if the beautiful color of the Hasselblad files you want to emulate is from files processed in Phocus, not Lightroom?
Second, the "look" of a file from a Hasselblad file and any other file is not just about color. Isn't there a lot of secret sauce (e.g., the tone curve) that goes into tuning a CAmera profile beyond adjustments to the "color"  by using an X Rite color checker to create a profile? I am sure that when Phase One develops the base profiles for its digital backs in Capture One, they are doing a lot more than just breaking out an X Rite color checker. Same for Hasselblad with Phocus.

[Michael Erlewine]

I am not so worried about emulating the Hasselblad, but rather the knowledge that I can create a profile, even if only in ACR, that does what I want. I am scared by the suggestion that I might get a camera with too much saturation or "vividness," which is not what I like. I am looking for what we do in video, which is have a neutral or "log" profile to start out with. And since this conversation here extends, I guess I am still looking for sure signs that the neutral format I like can be attained.

[ErikKaffehr]

Hi,

There is a lot of tuning options in colour profiles. The tone curve is the easy one.

In Adobe's DCP profiles there are "hue twists", see here: http://chromasoft.blogspot.se/2009/02/visualizing-dng-camera-profiles-part-3.html

Other raw converters would have similar tuning. They can be good or bad, it depends.

Hasselblad's Phocus is known to yield very good colour. So that is an advantage for Hasselblad if Phocus is what you want to use.

You are right that profiling will generate a good profile, but it will not have a lot of tuning.

But, it seems that having accurate profiles is going a long way.

Best regards
Erik

Color science and camera profiling are not, to put it mildly, my areas of expertise. However, two things occur to me. First, I would think it is very difficult to get the color you get out of the GFX to look exactly like the color from files from the X1D. In order to make them "the same", I would think you would have to use camera profiling software to profile the two cameras in the same raw converter and change the profiles of both of them to look the same. That's very different from making the Fuji file look exactly like the Hasselblad file. Moreover, what if the beautiful color of the Hasselblad files you want to emulate is from files processed in Phocus, not Lightroom?
Second, the "look" of a file from a Hasselblad file and any other file is not just about color. Isn't there a lot of secret sauce (e.g., the tone curve) that goes into tuning a CAmera profile beyond adjustments to the "color"  by using an X Rite color checker to create a profile? I am sure that when Phase One develops the base profiles for its digital backs in Capture One, they are doing a lot more than just breaking out an X Rite color checker. Same for Hasselblad with Phocus.

[razrblck]

I hope they ship a neutral profile, much like the Camera Neutral for Nikon bodies. I like starting from a less saturated, less contrasty base to work towards a more specific look.

I have often found the Adobe Standard profile for my cameras to be too contrasty and to saturate color channels too easily, making editing a bit more difficult and colors less natural (especially for skin tones). So I really hope Fuji doesn't just limit themselves in shipping a bunch of film profiles along with the Adobe Standard.

[Michael Erlewine]

More exchange with Ming Thein on his site:

MIchael: Since I only and always shoot raw, then with either camera I should be able to build a profile that allows me to get the “Hasselblad look” for the GFX. What am I missing in this approach?"

Ming Thein: "It depends very much on the sensor implementation used in the GFX: I have a lot of trouble making the X trans cameras look ‘neutral-flat’ like other Bayer cameras – there’s simply not enough shadow latitude left; the extended highlights are at the expense of the shadows. It should be possible…"

The discussion:

https://blog.mingthein.com/2017/01/21/cropping-sufficiency-resolution-take-three-or-thoughts-after-shooting-with-the-h6d-100c/#more-13953

[Paul2660]

Remember the GFX will not be x-trans.   Profiling should similar if not the same for any Bayer pattern sensor.

Paul Caldwell

[douglevy]

This may be true...and it may be a CCD vs. CMOS discussion, but I shoot D4/5/810 Credo. Recently my Credo 40 broke, and was replaced with a 60. In the window where I was without a back, I had to shoot a job to match a previous job shot with the Credo. Same lighting, same room, same raw converter - drastically different look. I'm certainly not an expert, but in trying to match using both C1 and LR, I just couldn't do it.

[razrblck]

This may be true...and it may be a CCD vs. CMOS discussion, but I shoot D4/5/810 Credo. Recently my Credo 40 broke, and was replaced with a 60. In the window where I was without a back, I had to shoot a job to match a previous job shot with the Credo. Same lighting, same room, same raw converter - drastically different look. I'm certainly not an expert, but in trying to match using both C1 and LR, I just couldn't do it.

Not long ago someone did a test on a back and a high resolution 35mm camera. While both images looked about the same, the digital back managed to better separate colors (greens from yellows, reds from oranges and so on) making color fine-tuning easier.

[Michael Erlewine]

Remember the GFX will not be x-trans.   Profiling should similar if not the same for any Bayer pattern sensor.

Paul Caldwell

I hear you, but I don't know what that means. I am trying to pull together all the various feedback from this topic. Here is where I am folks.

Both the X1D and the GFX share the same sensor, HOWEVER each may (and probably does) put their own spin on it. How is that done if it is raw? I originally thought that if I shot raw, that I would be safe and could set my own profile and get exactly what I want. But now I understand that, even with raw, the companies can push it this way or that. What I don't want is to have it pushed to be anything but neutral, meaning: not unnaturally vivid.

Here are some raw images, etc. from the GFX. They look lovely, but it is clear that the intent here is to show a dramatic image. I would like to see some plain old images. Suggestions are welcome.


http://www.photographyblog.com/previews/fujifilm_gfx_50s_photos/

[Bart_van_der_Wolf]

This may be true...and it may be a CCD vs. CMOS discussion, but I shoot D4/5/810 Credo. Recently my Credo 40 broke, and was replaced with a 60. In the window where I was without a back, I had to shoot a job to match a previous job shot with the Credo. Same lighting, same room, same raw converter - drastically different look. I'm certainly not an expert, but in trying to match using both C1 and LR, I just couldn't do it.

Hi Doug,

It will most likely not be 'easy', but it should be doable with the right tools. The 'DCamProf' tool by Anders Torger is probably the closest to a perfect tool for the job, but it's not simple to use if one attempts to get 100% identical looks. But with a lot of trial and error it is probably doable (we'll have to see the actual products first, shoot targets under identical lighting conditions, and tweak the settings of DCamProf).

But as Anders says, even with his coming GUI version, a manual for consultation is still needed.

Cheers,
Bart

[Bart_van_der_Wolf]

I hear you, but I don't know what that means. I am trying to pull together all the various feedback from this topic. Here is where I am folks.

Both the X1D and the GFX share the same sensor, HOWEVER each may (and probably does) put their own spin on it. How is that done if it is raw?

Assuming the supporting electronics will be similar, the only possibilities to make a change are by the strengyh of the IR filtration, and with a different Bayer CFA. Both will probably very similar, so the Raw data should be similar.

I originally thought that if I shot raw, that I would be safe and could set my own profile and get exactly what I want. But now I understand that, even with raw, the companies can push it this way or that. What I don't want is to have it pushed to be anything but neutral, meaning: not unnaturally vivid.

They do that (bake in a 'look') more likely than not with profiling, so it can also be changed by profiling.

Cheers,
Bart

[eronald]

If Fuji has tweaked the CFA and the cover glass then the two cameras are different, fullstop. Of course there is still the matter of lens transmittivity. I can back that up with maths if you insist. If the chip isn't tuned then I fail to see why the raws would differ, as these chois have a digital output.
Edmund

Basically, only the Bayer CFA could be a bit different, but they are usually quite similar in their transmission characteristics. The Out of Camera (OOC) color is a choice but with a manually made profile they can be made to look just like you prefer (within limits, because color based on just 3 channels is a huge problem to solve for human observers). A tool like DCAMprof, see here, can assist in synchronizing between different cameras, or create a user preferred color response.

Yes, each other, or both the same, or both to a personal flavor.

I expect the image quality of both to be good (after profiling), but do also look at supporting software options and lens options, and features that may be unique for a model. The articulating LCD is e.g. very useful for some of the macro work you do.

Cheers,
Bart