H3D39 as first step to MFDB/Technical (& vs DSLR)

[timparkin]

So, you have shown that most have a "similar" color rendering and one does not. We have no idea about the quality of the rendering of any of them. We also don't know why one is the outlier--you don't have enough data points to say the cause and so the outlier could be down to human error for all we know. We could do a little research and find out if there are an unusual number of complaints about the color from the p45+. If your "test" is actually showing it is an outlier, then there should be plenty of evidence. If not, that points back to a flaw in your test.

Actually if our test doesn't show it to be an outlier then Portra, Velvia, IQ180 and Sony A900 will be showing 'incorrect colour' ....

Plus, having spent a long time in the english countryside full of agricultural greens I'm fairly sure which looks correct and which looks wrong.

Tim

p.s. take a look at this image a part of a bunch of flowers. The images was clibrated with an IT8 target and the resultant target is a visual match bar slight variations in saturation. The colours are quite different though.

http://static.timparkin.co.uk/static/tmp/compare-crop-sat-sharp.jpg

 

[timparkin]

Can you post some evidence of illuminant metamerism? So far you have posted nothing that shows this. If it the image you used in your reply to me, that is not metamerism.

Actually the image I showed demonstrated observer metamerism.

Tim

 

[Shashin]

Actually the image I showed demonstrated observer metamerism.

Tim

Actually Tim, it does not.

In order for you to do that you would need a photograph of two objects, say a red wool sweater and a red silk tie, where the colors of those objects appear the same in the image. Then you would need to photograph the sweater and tie with another system and in the resulting image would show a mismatch between the reds, but not between the two photographs, but between the two objects. Metamerism has nothing to do with color accuracy.

You are not the first photographer not to understand metamerism.

 

[timparkin]

Actually the image I showed demonstrated observer metamerism.

Tim

Here's an example of illuminant metamerism

http://4.bp.blogspot.com/-x7rtJ6peu...AAAE0/U_vMZOz7Ylk/s1600/metamerism-gloves.jpg

 

[timparkin]

Actually Tim, it does not.

In order for you to do that you would need a photograph of two objects, say a red wool sweater and a red silk tie, where the colors of those objects appear the same in the image. Then you would need to photograph the sweater and tie with another system and in the resulting image would show a mismatch between the reds, but not between the two photographs, but between the two objects. Metamerism has nothing to do with color accuracy.

You are not the first photographer not to understand metamerism.

So I've shown you some moss and a fern that look the same under one system and then a mismatch under another system... er....

You're not the first person to make a mistake on a forum...

 

[timparkin]

...

You are not the first photographer not to understand metamerism.

p.p.s that comment was very presumptuous and patronising but you're not the first photographer on a forum to be so

 

[timparkin]

Here Tim, this is an article about metamerism:

Metamerism (color) - Wikipedia, the free encyclopedia)

It has to do with objects and human observers, not cameras. Cameras, unlike humans, have the same spectral response. And it still come down to observing two different objects that appear to have the same color. It is not that the observer sees them as the same, but the colors match. It has nothing to do with color accuracy or whether two observers see the same color. Your example is easy to explain--two different color profiles.

You really should familiarise yourself with the scientific literature, not just read wikipedia.

http://www.cis.rit.edu/mcsl/research/PDFs/Spec_Met.pdf

This is nothing to do with colour profiles, it's to do with the interaction of the emission spectra of the object being photographed and the transmission spectra of both the colour filter array and the silicon sensor beneath.

 

[timparkin]

Actually Tim, it does not.

In order for you to do that you would need a photograph of two objects, say a red wool sweater and a red silk tie, where the colors of those objects appear the same in the image. Then you would need to photograph the sweater and tie with another system and in the resulting image would show a mismatch between the reds, but not between the two photographs, but between the two objects. Metamerism has nothing to do with color accuracy.

You are not the first photographer not to understand metamerism.

Here Tim, this is an article about metamerism:

Metamerism (color) - Wikipedia, the free encyclopedia)

It has to do with objects and human observers, not cameras. Cameras, unlike humans, have the same spectral response. And it still come down to observing two different objects that appear to have the same color. It is not that the observer sees them as the same, but the colors match.

Hi Shashin,

I've just been rushing to catch a train but I'm now sitting with a bit of spare time so I thought I'd discuss a few things relating to metamerism.

The idea of metamerism doesn't have any reliance on a human observer, although the majority of literature talks about biological systems and you could be mistaken to think this was so.

The spectral energy of a system can be defined as the combination of the illuminant spectral power distribution, the surface reflectance function and the sensor spectral response (which for a bayer sensor is itself is a combination of the colour filter array tramission spectral response and the silicon spectral absorption).

So E is illuminant, S is the subject reflectance spectrum and C is the sensor response.

The "colour" of the object is then derived from three of these spectral energies using a tristimulus approximation.

Pr = E.S.Cr
Pg = E.S.Cg
Pb = E.S.Cb

However for a single colour in the tristimus you can have many different illuminants **or sensors** that will create the same colour

Pr = E.S1.C1r = E.S2.C2r

Pr = E1.S1.C1r = E2.S2.C1r

This is the scientific definition of metamerism

Now the situations I have shown you are drawn from the same mathematics.

The first example I showed with the ferns showed two substances that looked the same on one camera but different on another but on another camera they looked the same.

This would be because of the interaction of the substance spectral reflectance with the sensor absorption spectrum (this is a simplification as mentioned previously).

Now the main issue with this is that if my eye saw these two substances as different and I wanted to represent this in my photograph, I could not..

The second example showed a colour calibrated image using an IT8 target but where the cobalt blues and powder blues had shifted towards magenta to some extent on different sensors.

This isn't much different than the problems discussed in this paper about accurate painting reproduction.

http://www.rmimaging.com/information/color_accurate_photography.pdf

This exceptionally 'illuminating' paper explains the sorts of problems a photographer needs to be aware of when trying to reproduce images that match the typical human tristimulus response. It's well worth a read and I would recommend paying particular attention to the part about CCD spectral response and colour filters.

The paper also explains why very strong infra red filters are not installed in cameras by default..

Tim

 

[Shashin]

You really should familiarise yourself with the scientific literature, not just read wikipedia.

http://www.cis.rit.edu/mcsl/research/PDFs/Spec_Met.pdf

This is nothing to do with colour profiles, it's to do with the interaction of the emission spectra of the object being photographed and the transmission spectra of both the colour filter array and the silicon sensor beneath.

Tim, random internet searched are not going to help. But I thank you for pointing to a paper written from RIT--the place I studied color science. The paper essentially agrees with my position. Metamerism is still about color mismatch of two objects because of the change in illuminant or observer, not of why cameras produce different colors.

 

[timparkin]

Tim, random internet searched are not going to help. But I thank you for pointing to a paper written from RIT--the place I studied color science. The paper essentially agrees with my position. Metamerism is still about color mismatch of two objects because of the change in illuminant or observer, not of why cameras produce different colors.

Hardly random search as I'd already found this in the past - it's just one of the papers that demonstrates the case that you don't necessarily need a human observer (something you stated earlier). I don't see how the paper agrees with this position.

In any case, the camera is the observer and as such you can get a colour mismatch if the cameras total spectral response changes.

I don't know what is difficult to understand here?

Also at one point you say

1) "And since every camera can make a relatively accurate real color image"

regarding number one, I think you should take a look again at some early phone cameras...

2) "Cameras do not suffer from metamerism"

Nothing "suffers from metamerism" - that's a nonsensical statement. two different cameras behaving as two different observers can contribute to a system that exhibits metamerism.

3) "Metamerism is about two different materials that appear the same color under one light source, but no longer match under another."

No - that is only one part of metamerism. For someone with a degree (or higher?) in colour science I'm surprised that you ignored the observer. I'm also surprised that later in the thread you include the observer but then say they must be human. And then later agree with a paper that suggests that observer can be a digital sensor. At least commit to a single point of view!?

4) "It has to do with objects and human observers, not cameras. Cameras, unlike humans, have the same spectral response."

I don't know how you can possibly say you have a qualification in colour science and say this!!! really??!?

5) And it still come down to observing two different objects that appear to have the same color. It is not that the observer sees them as the same, but the colors match.

You're making massive mistakes about what colour is. Colour does not exist without an observer. It is primarily a human construct but we've made approximations that allow a tristimulus approximation to colour for machine systems.

You're also mixing up the role of a colour profile and didn't read the fact that the scenes I've been talking about that showed different colours were profiled. Hence it's not a profiling issue.

Out of all of these points though I think you should explain your statement 4

"Cameras, unlike humans, have the same spectral response"

How have you come to believe this?

Tim

 

[gazwas]

Tim, sorry but I'm with Marc and Shashin on this one and the fact that P45's and H3D39's have been used for years without user reports of the greens being off just goes to support this. If you said you preferred the green rendering of another camera then that is one thing but a blanket comment about the Kokak 39Mpix chip being "off" is wrong.

If you don't like how the P45 (and H3D39) captures green, then just set a custom colour profile that is applied on import in C1 and all the greens come in to your liking.

To me this debate is a little like people who judge the severity of a colour cast without applying an LCC. Sure some lenses look worse that others but what matters is the end result and if the lens cast, fall off or green rendering looks good in the final output who cares how you get there......

 

[Shashin]

Tim,

The observer can be anything. I am not doubting that camera will reproduce objects differently. But what I am saying, with an known illuminant, and unknown target, and an unknown result, simply saying it is down to metamerism (and nothing to do with a profile) is too much of a stretch. Your fern picture, at least going through my system, shows the two patches on the left do not match and nor do their values (the choice of sample point is a little dodgy too).

Clearly film and digital see the world differently, but that in and of itself is not metamerism. There is not a lot of choice in the filters used for a Bayer array. We never see RAW data, just what is transformed through the RAW processor. The Kodak sensor that is in my Pentax 645D is the same one that is in many other camera from other companies, but the color my camera produces is not the same. Convince me it is not the profile, but the sensor spectral response.

Except for some nice outdoor photographs, I don't see anything that points to a clear cause for color difference as simply metamerism. Especially since you simply discount profiles that would bias color information in the file.

Hi Shashin,

I've just been rushing to catch a train but I'm now sitting with a bit of spare time so I thought I'd discuss a few things relating to metamerism.

The idea of metamerism doesn't have any reliance on a human observer, although the majority of literature talks about biological systems and you could be mistaken to think this was so.

The spectral energy of a system can be defined as the combination of the illuminant spectral power distribution, the surface reflectance function and the sensor spectral response (which for a bayer sensor is itself is a combination of the colour filter array tramission spectral response and the silicon spectral absorption).

So E is illuminant, S is the subject reflectance spectrum and C is the sensor response.

The "colour" of the object is then derived from three of these spectral energies using a tristimulus approximation.

Pr = E.S.Cr
Pg = E.S.Cg
Pb = E.S.Cb

However for a single colour in the tristimus you can have many different illuminants **or sensors** that will create the same colour

Pr = E.S1.C1r = E.S2.C2r

Pr = E1.S1.C1r = E2.S2.C1r

This is the scientific definition of metamerism

Now the situations I have shown you are drawn from the same mathematics.

The first example I showed with the ferns showed two substances that looked the same on one camera but different on another but on another camera they looked the same.

This would be because of the interaction of the substance spectral reflectance with the sensor absorption spectrum (this is a simplification as mentioned previously).

Now the main issue with this is that if my eye saw these two substances as different and I wanted to represent this in my photograph, I could not..

The second example showed a colour calibrated image using an IT8 target but where the cobalt blues and powder blues had shifted towards magenta to some extent on different sensors.

This isn't much different than the problems discussed in this paper about accurate painting reproduction.

http://www.rmimaging.com/information/color_accurate_photography.pdf

This exceptionally 'illuminating' paper explains the sorts of problems a photographer needs to be aware of when trying to reproduce images that match the typical human tristimulus response. It's well worth a read and I would recommend paying particular attention to the part about CCD spectral response and colour filters.

The paper also explains why very strong infra red filters are not installed in cameras by default..

Tim

 

[timparkin]

Tim, sorry but I'm with Marc and Shashin on this one and the fact that P45's and H3D39's have been used for years without user reports of the greens being off just goes to support this. If you said you preferred the green rendering of another camera then that is one thing but a blanket comment about the Kokak 39Mpix chip being "off" is wrong.

If you don't like how the P45 (and H3D39) captures green, then just set a custom colour profile that is applied on import in C1 and all the greens come in to your liking.

To me this debate is a little like people who judge the severity of a colour cast without applying an LCC. Sure some lenses look worse that others but what matters is the end result and if the lens cast, fall off or green rendering looks good in the final output who cares how you get there......

I think you've missed the point of the discussion. It's not that the green is something I don't like. It's that some green substances are rendering differently.

e.g In my scene I might have a moss green and a grass green that are looking different. With a certain camera rendition the two greens can end up looking the same. No amount of colour profiling can fix this.

i.e. a photograph of green on a colour checker comes out correctly but the colour green of something chlorophyll based comes out differently

Here's a sample image taken from an iq180 and a p45 side by side.

Try to make the P45 look like the IQ180 through profiling is impossible

http://static.timparkin.co.uk/static/tmp/p45.jpg
http://static.timparkin.co.uk/static/tmp/iq180.jpg

You can see more comparisons here..

http://static.timparkin.co.uk/static/tmp/cameratest-2/800px.html

This has nothing to do with LCCs either (we used an LCC on both)

Tim

 

[gazwas]

Hard to see on such small images but the IQ180 looks more sensitive to red which can be seen visually and by looking at the histogram but I don't see anything wrong or "off" as you put it either.

 

[timparkin]

Tim,

The observer can be anything. I am not doubting that camera will reproduce objects differently. But what I am saying, with an known illuminant, and unknown target, and an unknown result, simply saying it is down to metamerism (and nothing to do with a profile) is too much of a stretch. Your fern picture, at least going through my system, shows the two patches on the left do not match and nor do their values (the choice of sample point is a little dodgy too).

Yes but in the comparison of the photographs taken with the P45 and IQ180 (amongst others) we have known illiminant, known subject and two different sensors which when profiled so that a colour target produces identical results fails for certain substances.

The only explanation for this is metamerism as far as I can tell - I'd be interested in what you think could explain it.

Clearly film and digital see the world differently, but that in and of itself is not metamerism. There is not a lot of choice in the filters used for a Bayer array.

Don't be silly! There is a huge range of choices for colour filter arrays and they come in varying densities too.

We never see RAW data, just what is transformed through the RAW processor. The Kodak sensor that is in my Pentax 645D is the same one that is in many other camera from other companies, but the color my camera produces is not the same. Convince me it is not the profile, but the sensor spectral response.

Well I suggest that if you take two pictures on two cameras and fully profile them but certain substances still don't match then this is down to spectral response. The Leica IR response is a classic example - nothing to do with profiling. Some blacks looked black, some blacks looked purple.

This problem is exhibited by the P25 and P45 too - take a look at this thread

PHASEONE color profiling - Open Photography Forums

This is mostly due to infrared issues but the complex spectra of chlorophyll interacts with these sorts of spectral response to shift colours. If a green doesn't have a lot of chlorophyll then the colours don't shift. You can't apply a profile that corrects one and not the other.

If you want a simple test whether cameras sensors can cause metameric shifts, try taking a photograph of a GATF/RHEM light indicator under a light source where the colours all look the same for your eye. I guarantee the colours will look different to the camera sensor - that is metameric failure.

Except for some nice outdoor photographs, I don't see anything that points to a clear cause for color difference as simply metamerism. Especially since you simply discount profiles that would bias color information in the file.

I don't discount profiles - I've used them based on an IT8 target in one example and a Hutch target in the other. the colours of the targets in the profiled pictures match but some objects in the scene don't.

Tim

 

[timparkin]

Hard to see on such small images but the IQ180 looks more sensitive to red which can be seen visually and by looking at the histogram but I don't see anything wrong or "off" as you put it either.

If you can't see the differences in the greens of the fields in the centre of the picture then it's no point continuing the conversation. Perhaps I've become more attuned to greens as a landscape photographer (who lives in britain).

This is primarily a chlorophyll issue and I have seen few problems with other colours. Unfortunately chlorophyll green is pretty common in the landscape (especially in the UK)

 

[Shashin]

If you throw those images in Photoshop and just hit auto color, you get two very similar, but butt ugly, images. The saturation is a hair higher for the p45, but depending how that went through the RAW processor, the difference in DR could account for that. When you map the difference in layers with the autocolor images, there is not a great deal to see. I even tried playing with making the image similar, I could get them closer, but clearly saturation/contrast is not the same in each that it makes it difficult to get a perfect match. Here again, just a difference in DR? I suspect, these images have been processed, at least LCC were applied, and how that impacts results, who can say.

There are so many variables that can affect the color output beyond the the simple spectral response of the sensor, that is is difficult to make any kind of judgement. And this is hardly a controlled test.

 

[timparkin]

If you throw those images in Photoshop and just hit auto color, you get two very similar, but butt ugly, images. The saturation is a hair higher for the p45, but depending how that went through the RAW processor, the difference in DR could account for that. When you map the difference in layers with the autocolor images, there is not a great deal to see. I even tried playing with making the image similar, I could get them closer, but clearly saturation/contrast is not the same in each that it makes it difficult to get a perfect match. Here again, just a difference in DR? I suspect, these images have been processed, at least LCC were applied, and how that impacts results, who can say.

There are so many variables that can affect the color output beyond the the simple spectral response of the sensor, that is is difficult to make any kind of judgement. And this is hardly a controlled test.

If you want a controlled test why don't you take a look at DxOs metameric index using the ISO standard methodology... I'll save you the trouble and show you the values for some of the cameras we've been discussing and a few more.. 100 is perfect - 50 the sort of metameric failure you get from an illuminant change from daylight to flourescents.

Sony A900 87
Canon 5D 84
Nikon D700 83
Nikon D90 82
Panasonic G3 81
Nikon V1 81
Phase IQ180 80
Phase P40 80
Canon 5dMkII 80
Nikon D3200 80
Panasonic GH1 79
Nikon D3S 79
Nikon D3X 79
Hassleblad H50 78
Canon 7D 78
Nikon D800 77
Panasonic GH2 77
Phase P65+ 76
Leica M9 76
Leica M240 75
Panasonic LX5 75
Hassleblad H39 75
Aptus Leaf 75
Canon 5Dmk3 74
Phase P45 72
Canon 6D 69

Of course if you don't believe in sensor metamerism or you don't think DxO can do measurements properly or that ISO can set an standard....

 

[gazwas]

If you can't see the differences in the greens of the fields in the centre of the picture then it's no point continuing the conversation. Perhaps I've become more attuned to greens as a landscape photographer (who lives in britain).

You live in Britain so are more sensitive to green is a classic and because I don't agree with you I shouldn't continue with this conversation...... BRILLIANT! :ROTFL:

 

[torger]

No need to get hard-edged.

This article I think was quite good introduction at metameric error in digital photography for anyone interested:
http://dougkerr.net/pumpkin/articles/Metameric_Error.pdf

It is clear that cameras vary in their ability to capture colors so they render in the same way the eye see them. In a camera the issue is that many spectrums can get registered with the same R, G and B values and thus be registered as the same color, despite that the eye may see different colors.

We can argue about if this actually affects green in KAF-39000 sensor, if its easy to correct for and if it matters. I'm no expert in that area so I'm open for that it could be any answer to these, but I don't really trust the general "noone has complained therefore its perfect", in fact I've seen people "complain" indirectly by saying that Dalsa sensors have more accurate color.

After all this hard-edged discussion it doesn't seem to matter much though. The H3D-39 is okay, buy it if you find a good deal