GFX 100sii vs X2D: Color Comparison

[SrMi]

What are the differences between the micro lenses of the X1D and GFX50? Is there an article on this?

I'd be interested in hearing about this, too. When I tested the GFX 50x cameras, I found the effective aperture of the microlens array to be well below what would be required for 100% fill factor. I haven't seen measurements on the X1D.

I assumed that the X1D's microlenses had a more regular fill factor and that the source was your articles. It seems I am wrong, as I do not remember any other source discussing microlenses.

Comparing images in the DPR studio scene, I notice more aliasing with the X1D than with the GFX50S. (link).

 

[JimKasson]

What are the differences between the micro lenses of the X1D and GFX50? Is there an article on this?

I'd be interested in hearing about this, too. When I tested the GFX 50x cameras, I found the effective aperture of the microlens array to be well below what would be required for 100% fill factor. I haven't seen measurements on the X1D.

I assumed that the X1D's microlenses had a more regular fill factor and that the source was your articles. It seems I am wrong, as I do not remember any other source discussing microlenses.

I never tested the X1D.

Comparing images in the DPR studio scene, I notice more aliasing with the X1D than with the GFX50S. (link).

Hmm...

 

[Oren Grad]

My experience editing and printing for folk using many different cameras for capture confirms what I’m seeing here. There’s not a nickel’s worth of difference between the two. Color perception and appearance is DRAMATICALLY impacted by the photographer and his/her workflow and choice of editing software, FAR beyond any native differences - especially with cameras using the same sensor.

Rand

X2D and GFX 100 s II do not have the same sensor (different toppings), and raw massaging is different..

Can you be specific about the differences between the toppings?

We know that the number of PDAF rows is different. We know that CFA can be customized for Sony sensors. We know that microlenses can be different (M11 vs a7rV). Other than that, we do not know whether they are using the same or different CFA.

It sounds like you are saying that we know that one thing is different.

If the color matrices used for demosaicing are substantially different, one should be able to conclude that the CFAs must be different, no?

 

[JimKasson]

The Fuji colors are more pleasing to my eye but I can see they are biased towards magenta. The X2D colors might be more accurate. You have to use Phocus to get the best out of HNCS.

Well, at least we have that settled

Most certainly. I always maintained Hasselblad colors were discernably more accurate.

Is this a marketing thing or real?

X2D and GFX 100S color accuracy with various profiles - the last word

This is the 27th in a series of posts on the Hasselblad X2D 100C camera and the XCD lenses. You will be able to find all the posts in this series by looking at the righthand column on this page and finding the Category "X2D". News flash: You can get even greater accuracy if you

blog.kasson.com

 

[JimKasson]

My experience editing and printing for folk using many different cameras for capture confirms what I’m seeing here. There’s not a nickel’s worth of difference between the two. Color perception and appearance is DRAMATICALLY impacted by the photographer and his/her workflow and choice of editing software, FAR beyond any native differences - especially with cameras using the same sensor.

Rand

X2D and GFX 100 s II do not have the same sensor (different toppings), and raw massaging is different..

Can you be specific about the differences between the toppings?

We know that the number of PDAF rows is different. We know that CFA can be customized for Sony sensors. We know that microlenses can be different (M11 vs a7rV). Other than that, we do not know whether they are using the same or different CFA.

It sounds like you are saying that we know that one thing is different.

If the color matrices used for demosaicing are substantially different, one should be able to conclude that the CFAs must be different, no?

You mean the ones in the metadata? That could relate to the optimization methods used to generate the compromise matrices.

 

[kciuk aka Ken60]

I am not a techno fiend like some, but I noticed that , to me , it was the choice of AWB that influenced the images most. There were images that were quite similar, when the ambient colour mix allowed. When we look at for example the image of the purple flowers in the foreground the difference in colour is marked by comparison.

I asked the OP if it was possible to reshoot with both 5500K and daylight colour balance, to see if this might unify the results..... sadly this was not possible.

Perhaps the other take from this is that if the images were printed and framed neither brand identity would show, and the abilities of the photographer to create a captivating picture would outweigh technical variance.

 

[JimKasson]

I am not a techno fiend like some, but I noticed that , to me , it was the choice of AWB that influenced the images most.

I hate AWB in all its forms. It sometimes gets close to what I want, but it almost never gets it the way I want it. This is true of every camera I've used.

 

[SrMi]

I am not a techno fiend like some, but I noticed that , to me , it was the choice of AWB that influenced the images most.

I hate AWB in all its forms. It sometimes gets close to what I want, but it almost never gets it the way I want it. This is true of every camera I've used.

Yes, yes, and yes!

One of the biggest steps forward in photography I took recently was to turn off AWB on all my cameras. Otherwise, it's just the camera's guesswork.

 

[Oren Grad]

My experience editing and printing for folk using many different cameras for capture confirms what I’m seeing here. There’s not a nickel’s worth of difference between the two. Color perception and appearance is DRAMATICALLY impacted by the photographer and his/her workflow and choice of editing software, FAR beyond any native differences - especially with cameras using the same sensor.

Rand

X2D and GFX 100 s II do not have the same sensor (different toppings), and raw massaging is different..

Can you be specific about the differences between the toppings?

We know that the number of PDAF rows is different. We know that CFA can be customized for Sony sensors. We know that microlenses can be different (M11 vs a7rV). Other than that, we do not know whether they are using the same or different CFA.

It sounds like you are saying that we know that one thing is different.

If the color matrices used for demosaicing are substantially different, one should be able to conclude that the CFAs must be different, no?

You mean the ones in the metadata? That could relate to the optimization methods used to generate the compromise matrices.

Thanks - I'm a bit out of my depth here, needed to go back and review your posts from 2015 on color reproduction, also some of Jack Hogan's writings. I did have the compromise matrices in mind, though I was of the impression (perhaps incorrectly?) that the compromise matrices and/or LUTs used by raw converters aren't always or necessarily drawn from the metadata. For cases where the matrices/LUTs are provided by the raw converter, one might presume that any optimization methods are applied consistently, so that differences in the matrices/LUTs should be at least prima facie evidence of differences in the CFAs.

 

[JimKasson]

My experience editing and printing for folk using many different cameras for capture confirms what I’m seeing here. There’s not a nickel’s worth of difference between the two. Color perception and appearance is DRAMATICALLY impacted by the photographer and his/her workflow and choice of editing software, FAR beyond any native differences - especially with cameras using the same sensor.

Rand

X2D and GFX 100 s II do not have the same sensor (different toppings), and raw massaging is different..

Can you be specific about the differences between the toppings?

We know that the number of PDAF rows is different. We know that CFA can be customized for Sony sensors. We know that microlenses can be different (M11 vs a7rV). Other than that, we do not know whether they are using the same or different CFA.

It sounds like you are saying that we know that one thing is different.

If the color matrices used for demosaicing are substantially different, one should be able to conclude that the CFAs must be different, no?

You mean the ones in the metadata? That could relate to the optimization methods used to generate the compromise matrices.

Thanks - I'm a bit out of my depth here, needed to go back and review your posts from 2015 on color reproduction, also some of Jack Hogan's writings. I did have the compromise matrices in mind, though I was of the impression (perhaps incorrectly?) that the compromise matrices and/or LUTs used by raw converters aren't always or necessarily drawn from the metadata.

That's true. And that's yet another confounding factor.

For cases where the matrices/LUTs are provided by the raw converter, one might presume that any optimization methods are applied consistently, so that differences in the matrices/LUTs should be at least prima facie evidence of differences in the CFAs.

I think all the Adobe Color Profiles are tweaked by a person, but I don't really know.

 

[kciuk aka Ken60]

I can't speak highly enough of Capture One's rendition and colour, its come a long way in the last few years. Yes there is a whole new way of playing without Adobe thinking , but sometimes that a very good thing.

 

[Oren Grad]

My experience editing and printing for folk using many different cameras for capture confirms what I’m seeing here. There’s not a nickel’s worth of difference between the two. Color perception and appearance is DRAMATICALLY impacted by the photographer and his/her workflow and choice of editing software, FAR beyond any native differences - especially with cameras using the same sensor.

Rand

X2D and GFX 100 s II do not have the same sensor (different toppings), and raw massaging is different..

Can you be specific about the differences between the toppings?

We know that the number of PDAF rows is different. We know that CFA can be customized for Sony sensors. We know that microlenses can be different (M11 vs a7rV). Other than that, we do not know whether they are using the same or different CFA.

It sounds like you are saying that we know that one thing is different.

If the color matrices used for demosaicing are substantially different, one should be able to conclude that the CFAs must be different, no?

You mean the ones in the metadata? That could relate to the optimization methods used to generate the compromise matrices.

Thanks - I'm a bit out of my depth here, needed to go back and review your posts from 2015 on color reproduction, also some of Jack Hogan's writings. I did have the compromise matrices in mind, though I was of the impression (perhaps incorrectly?) that the compromise matrices and/or LUTs used by raw converters aren't always or necessarily drawn from the metadata.

That's true. And that's yet another confounding factor.

For cases where the matrices/LUTs are provided by the raw converter, one might presume that any optimization methods are applied consistently, so that differences in the matrices/LUTs should be at least prima facie evidence of differences in the CFAs.

I think all the Adobe Color Profiles are tweaked by a person, but I don't really know.

Jim, how do you interpret the Color Response data that are posted by DXO? (No GFX camera available in their data set, alas.) A couple of examples - click on the "Color Response" button in each:

https://www.dxomark.com/Cameras/Hasselblad/X1D-50c---Measurements

https://www.dxomark.com/Cameras/Pentax/645Z---Measurements

 

[JimKasson]

My experience editing and printing for folk using many different cameras for capture confirms what I’m seeing here. There’s not a nickel’s worth of difference between the two. Color perception and appearance is DRAMATICALLY impacted by the photographer and his/her workflow and choice of editing software, FAR beyond any native differences - especially with cameras using the same sensor.

Rand

X2D and GFX 100 s II do not have the same sensor (different toppings), and raw massaging is different..

Can you be specific about the differences between the toppings?

We know that the number of PDAF rows is different. We know that CFA can be customized for Sony sensors. We know that microlenses can be different (M11 vs a7rV). Other than that, we do not know whether they are using the same or different CFA.

It sounds like you are saying that we know that one thing is different.

If the color matrices used for demosaicing are substantially different, one should be able to conclude that the CFAs must be different, no?

You mean the ones in the metadata? That could relate to the optimization methods used to generate the compromise matrices.

Thanks - I'm a bit out of my depth here, needed to go back and review your posts from 2015 on color reproduction, also some of Jack Hogan's writings. I did have the compromise matrices in mind, though I was of the impression (perhaps incorrectly?) that the compromise matrices and/or LUTs used by raw converters aren't always or necessarily drawn from the metadata.

That's true. And that's yet another confounding factor.

For cases where the matrices/LUTs are provided by the raw converter, one might presume that any optimization methods are applied consistently, so that differences in the matrices/LUTs should be at least prima facie evidence of differences in the CFAs.

I think all the Adobe Color Profiles are tweaked by a person, but I don't really know.

Jim, how do you interpret the Color Response data that are posted by DXO? (No GFX camera available in their data set, alas.) A couple of examples - click on the "Color Response" button in each:

https://www.dxomark.com/Cameras/Hasselblad/X1D-50c---Measurements

https://www.dxomark.com/Cameras/Pentax/645Z---Measurements

I have no idea how to convert that into the spectral response of the four Bayer channels.

 

[Oren Grad]

If the color matrices used for demosaicing are substantially different, one should be able to conclude that the CFAs must be different, no?

You mean the ones in the metadata? That could relate to the optimization methods used to generate the compromise matrices.

Thanks - I'm a bit out of my depth here, needed to go back and review your posts from 2015 on color reproduction, also some of Jack Hogan's writings. I did have the compromise matrices in mind, though I was of the impression (perhaps incorrectly?) that the compromise matrices and/or LUTs used by raw converters aren't always or necessarily drawn from the metadata.

That's true. And that's yet another confounding factor.

For cases where the matrices/LUTs are provided by the raw converter, one might presume that any optimization methods are applied consistently, so that differences in the matrices/LUTs should be at least prima facie evidence of differences in the CFAs.

I think all the Adobe Color Profiles are tweaked by a person, but I don't really know.

Jim, how do you interpret the Color Response data that are posted by DXO? (No GFX camera available in their data set, alas.) A couple of examples - click on the "Color Response" button in each:

https://www.dxomark.com/Cameras/Hasselblad/X1D-50c---Measurements

https://www.dxomark.com/Cameras/Pentax/645Z---Measurements

I have no idea how to convert that into the spectral response of the four Bayer channels.

Sure, but I'm not that ambitious here - I'm simply looking for a source of plausible empirical evidence, even if indirect, that the CFAs are different. DXO generates its metrics based on raw data collected under controlled illumination, without demosaic and so bypassing any corrections smuggled in via a "tuned" compromise matrix created by someone else. Yet they still end up with numbers that are different for the same underlying sensor deployed in different camera models from different companies.

 

[JimKasson]

If the color matrices used for demosaicing are substantially different, one should be able to conclude that the CFAs must be different, no?

You mean the ones in the metadata? That could relate to the optimization methods used to generate the compromise matrices.

Thanks - I'm a bit out of my depth here, needed to go back and review your posts from 2015 on color reproduction, also some of Jack Hogan's writings. I did have the compromise matrices in mind, though I was of the impression (perhaps incorrectly?) that the compromise matrices and/or LUTs used by raw converters aren't always or necessarily drawn from the metadata.

That's true. And that's yet another confounding factor.

For cases where the matrices/LUTs are provided by the raw converter, one might presume that any optimization methods are applied consistently, so that differences in the matrices/LUTs should be at least prima facie evidence of differences in the CFAs.

I think all the Adobe Color Profiles are tweaked by a person, but I don't really know.

Jim, how do you interpret the Color Response data that are posted by DXO? (No GFX camera available in their data set, alas.) A couple of examples - click on the "Color Response" button in each:

https://www.dxomark.com/Cameras/Hasselblad/X1D-50c---Measurements

https://www.dxomark.com/Cameras/Pentax/645Z---Measurements

I have no idea how to convert that into the spectral response of the four Bayer channels.

Sure, but I'm not that ambitious here - I'm simply looking for a source of plausible empirical evidence, even if indirect, that the CFAs are different. DXO generates its metrics based on raw data collected under controlled illumination, without demosaic and so bypassing any corrections smuggled in via a "tuned" compromise matrix created by someone else. Yet they still end up with numbers that are different for the same underlying sensor deployed in different camera models from different companies.

I’m not saying it’s impossible. I am not even saying that it’s unlikely. I’m just looking for actual evidence in a reproducible experiment.

 

[highdesertmesa]

What are the differences between the micro lenses of the X1D and GFX50? Is there an article on this?

I'd be interested in hearing about this, too. When I tested the GFX 50x cameras, I found the effective aperture of the microlens array to be well below what would be required for 100% fill factor. I haven't seen measurements on the X1D.

I assumed that the X1D's microlenses had a more regular fill factor and that the source was your articles. It seems I am wrong, as I do not remember any other source discussing microlenses.

Comparing images in the DPR studio scene, I notice more aliasing with the X1D than with the GFX50S. (link).

Wouldn't the most likely reason be that the lens used on the X1D was higher resolving than the one used the 50S? One would need to adapt the same lens to both cameras to really know what the sensors were doing relative to each other.

 

[SrMi]

What are the differences between the micro lenses of the X1D and GFX50? Is there an article on this?

I'd be interested in hearing about this, too. When I tested the GFX 50x cameras, I found the effective aperture of the microlens array to be well below what would be required for 100% fill factor. I haven't seen measurements on the X1D.

I assumed that the X1D's microlenses had a more regular fill factor and that the source was your articles. It seems I am wrong, as I do not remember any other source discussing microlenses.

Comparing images in the DPR studio scene, I notice more aliasing with the X1D than with the GFX50S. (link).

Wouldn't the most likely reason be that the lens used on the X1D was higher resolving than the one used the 50S? One would need to adapt the same lens to both cameras to really know what the sensors were doing relative to each other.

This could be, though I would not expect such a difference. The lenses used were GF63/2.8 and XCD 90/3.2.

 

[Last Mango]

I am not a techno fiend like some, but I noticed that , to me , it was the choice of AWB that influenced the images most.

I hate AWB in all its forms. It sometimes gets close to what I want, but it almost never gets it the way I want it. This is true of every camera I've used.

Yes, yes, and yes!

One of the biggest steps forward in photography I took recently was to turn off AWB on all my cameras. Otherwise, it's just the camera's guesswork.

I am learning this very fact, which I have previously blamed on the camera, but seems like a universal issue....what is the best method to set WB, grey card?

 

[SrMi]

I am not a techno fiend like some, but I noticed that , to me , it was the choice of AWB that influenced the images most.

I hate AWB in all its forms. It sometimes gets close to what I want, but it almost never gets it the way I want it. This is true of every camera I've used.

Yes, yes, and yes!

One of the biggest steps forward in photography I took recently was to turn off AWB on all my cameras. Otherwise, it's just the camera's guesswork.

I am learning this very fact, which I have previously blamed on the camera, but seems like a universal issue....what is the best method to set WB, grey card?

I set it to an appropriate preset (sunny, cloudy) as I shoot in raw and adjust the WB in post. The WB may not be correct initially but is consistent.

 

[kciuk aka Ken60]

I think it depends on the place and the camera.

With the GFX 100S II there are 3 auto white balances , but daylight seems a good choice outdoors. Cloudy is always way too warm. There is another outdoor alternative and that is to set 5500 to 5800 Kelvin. Of course if you are shooting next to a white car or true white surface is handy you can do a custom white. I think have a play with those options and see what you think works for you .

------------------------- Quick edit

If you use custom white then be aware of your ambiance. For example a white card held out to set a balance under trees will have a lot of green biased light on it, and if you then walk over away from the trees and shoot the magenta induced will spoil your balance. So I think the K option and adjust to pref in post.