Luminous Landscape Forum
Equipment & Techniques => Medium Format / Film / Digital Backs – and Large Sensor Photography => Topic started by: ErikKaffehr on March 22, 2018, 10:42:17 pm
-
Update 2018-03-26:
----------------------------------------------------------------------------------------------------------------------------------------------------------------
You may have noticed the numbers on the fruits. Those correspond to samples I have read with my ColorMunki Photo spectrometer.
I have added most of those samples as a layer on top. Choosing mixing mode "color" overlays the color ignoring the luminousity, I think.
Here is that link: http://echophoto.dnsalias.net/ekr/Articles/Temp/Tricolore/Stacked_with_samples.tif
The fruit that some readers interpret to be a tangerine was called an orange where I bought it.
What I can see the Habeneros 3 and 4 show the largest deviations.
Please note that some colours are outside sRGB.
---------------------------------------------------------------------------------------------------------------
Note! Below is some new information...
This is a colour test. Please don't try to mix focus, DoF etc in the judgement, it will just be missleading.
Just to explain, the original ideal was to see how a set of reasonably well defined colours would map to Lab colour space using different profiles. The results were "all over the place", so I gave up on interpretation.
Here are the details how the images were made:
- Obviously all images were shot with three cameras from similar angle of view.
- Focus is different between the images. Focus doesn't affect color, so I strived for decent focus but it differs between the images. Forget about focus. This is a colour test!
- All images were processed in Lightroom, with the DCP profiles I use with each camera. Those DCP profiles are all created with LumaRiver Profile Designer.
- White balance was set on second brightest grey patch on each image.
- Luminance channel was set to around 51 on fourth brightest grey patch. That is the reference value for the Xrite ColorChecker.
- The images were exported at 1500 pixels width and opened in Photoshop.
- Each image was resized to 1500x1050 size. copied and pasted as a new layer in the image.
- All the layers have been aligned. That adds distortion in the mix, but makes it easier to compare color.
- The layers were copied as individual pictures from each layer.
Here is a layered image: http://echophoto.dnsalias.net/ekr/Articles/Temp/Tricolore/Stacked.tif
Hi
Mucho discussions about Colour on X1D or GFX. I have neither...
But I have shot three images under controlled conditions, generated DCP Colour Profiles for each camera and processed the images the same way.
Two of the images were shot on Sony cameras (A900 and A7rII) and one P45+.
So, can you tell which one is the P45+
(http://echophoto.dnsalias.net/ekr/Articles/Temp/Tricolore/Tricolore2.jpg)
Notes:
I have reworked the samples, to make sure I made no mistakes.
A tiff version of the file can be downloaded here: http://echophoto.dnsalias.net/ekr/Articles/Temp/Tricolore/Tricolore2.tif
Best regards
Erik
-
Not sure, but can you tell me which one in Johny Walker Blue?
-
Hi Joe,
No, I cannot. But I neither drink whisky nor take pictures of it. And of course I never claimed I would be able I would be able I would be able to tell two whiskies apart.
On the other hand, a lot guys say that there is a big difference between MFDs and DSLRs with regard to colour. So as I have happened to have those images I felt it was a good idea to see what folks think.
Best regards
Erik
Not sure, but can you tell me which one in Johny Walker Blue?
-
Not sure, but can you tell me which one in Johny Walker Blue?
I hope none of them. The idea of that much ice in a glass of Blue makes me ill to my stomach. That horrible things happen in the world is something I was vaguely aware of. But to have such travesty so directly in my face. It's awful. Simply awful.*
Mods: can we get a dedicated whiskey subforum? We could even have a large format (https://www.acespirits.com/buy-large-format-wine-and-spirits-big-liquor-bottles.html) and small-format (https://www.masterofmalt.com/tasting-set/) section of it
*Joking. Drinking should never be taken too seriously. Blue is a nice whiskey and you should enjoy it however you like. Plus one could argue that pictures of whiskey look weird without the visual interest of a nice ice cube.
-
I'll say C, as the plants looks a bit more dimensional there.
Bill
-
Hi,
Thanks for your comment. nice to see some votes coming in.
Best regards
Erik
I'll say C, as the plants looks a bit more dimensional there.
Bill
-
Not sure, but can you tell me which one in Johny Walker Blue?
Of course I can't. You used different glass for each. Don't you know you have to use the same glass when doing these tests? So many people do these critical tests and get the basic stuff wrong.
Jeeze.
8)
Dave
-
The way you present the pictures makes a direct comparison difficult, so I had to extract the 3 images to be able to flip between them. When I flip between the images, I notice the following:
-the colours on the colorchecker chart are identical almost identical (*). Presumably, you used that chart to calibrate the images. Please note that, by calibrating the images, you may have destroyed some of the original colour differences.
-B and C look almost identical on all subjects.
-A looks markedly different on the tangerine (one of the two items 8 ) and a bit different on the chilis marked 4 and 3. Therefore, I vote for A.
-I see almost no difference on the other items. While your images have a range of colours from natural objects (and that is a good thing), I miss skin tones. Interestingly, the tangerine is the object that comes closest to the area where "caucasian" skin tones would be, no other object is pink-orange. This is also the object where the difference is more important.
I attach a smaller version of the separated images, if somebody else wants to do a quick comparison flip.
(*) Edit: with the later posted animation, I see some differences in the blue and violet squares of the chart on image C. I suppose that your calibration was not perfectly correct with that camera on that camera, which may also explain the darker flower petals. Still: the orange patches are matched between all cameras, while the tangerine is not, a proof that cameras do not match the Luther–Ives conditions...
-
The C
-
Note that the plant on C looks a bit different in the montage of 3 pictures, because its surroundings are different. It is the effect demonstrated on this optical illusion, where all centre squares are identical:
(http://www.coolopticalillusions.com/optical_illusions_images_2/images/shadesofgray.gif)
If the pictures are isolated, the plant looks identical on all pictures to my eye.
-
At first I thought the plant in C looked more dimensional, but now that I isolate the plant I can't tell. But the plant in B and C look about the same and A with the most difference. At this point I don't know which one might be the Phase.
Bill
-
I'm away so just can see the images on the smartphone where the C looked more 3d at first so to say. But seeing it again a bit closer I'd choose the A.
-
Hi,
Great observation! I did not think about that!
Any suggestions for better presentation? I have noticed that C got most votes this far, but that may be attributed to the point you make!
Thanks a lot!
Erik
Note that the plant on C looks a bit different in the montage of 3 pictures, because its surroundings are different. It is the effect demonstrated on this optical illusion, where all centre squares are identical:
(http://www.coolopticalillusions.com/optical_illusions_images_2/images/shadesofgray.gif)
If the pictures are isolated, the plant looks identical on all pictures to my eye.
-
Hi Fred,
Thanks for your input!
:-) Erik :-)
I'm away so just can see the images on the smartphone where the C looked more 3d at first so to say. But seeing it again a bit closer I'd choose the A.
-
Thanks for your efforts, much appreciated!
As far as I recall, I have used profiles generated from a different set of shoots using a Solux lamp at overvoltage as D50 illuminant.
At this stage I don't want to do any more disclosure, I hope you have understanding for that!
Best regards
Erik
The way you present the pictures makes a direct comparison difficult, so I had to extract the 3 images to be able to flip between them. When I flip between the images, I notice the following:
-the colours on the colorchecker chart are identical. Presumably, you used that chart to calibrate the images. Please note that, by calibrating the images, you may have destroyed some of the original colour differences.
-B and C look almost identical on all subjects.
-A looks markedly different on the tangerine (one of the two items 8 ) and a bit different on the chilis marked 4 and 3. Therefore, I vote for A.
-I see almost no difference on the other items. While your images have a range of colours from natural objects (and that is a good thing), I miss skin tones. Interestingly, the tangerine is the object that comes closest to the area where "caucasian" skin tones would be, no other object is pink-orange. This is also the object where the difference is more important.
I attach a smaller version of the separated images, if somebody else wants to do a quick comparison flip.
-
Any suggestions for better presentation?
The best way to see the differences is to load all pictures into different layers in photoshop and animate them, but I am not sure how to present that data on the web. I made an animated gif to give you the idea, but of course the gif format considerably reduces the color palette, so it only gives a rough idea. Even then the differences in the tangerine are obvious.
You need to click on the picture below for it to animate. With the animation, it seems that B and C have the same lens distortions, which would also indicate A is the P45+.
-
it seems that B and C have the same lens distortions, which would also indicate A is the P45+.
Also in my smartphone, B & C have similar red and orange while A is a bit less "overdone" or slightly more refined.
But they are very very close. I stick with A.
-
I picked C. The depth of field seems less.
Greg
-
Hi,
This is about colour, not focus or depth of field. So don't take DoF as a clue.
Best regards
Erik
I picked C. The depth of field seems less.
Greg
-
Hi,
This is a colour test. Please don't try to mix focus, DoF etc in the judgement, it will just be missleading.
Just to explain, the original ideal was to see how a set of reasonably well defined colours would map to Lab colour space using different profiles. The results were "all over the place", so I gave up on interpretation.
Here are the details how the images were made:
- Obviously all images were shot with three cameras from similar angle of view.
- Focus is different between the images. Focus doesn't affect color, so I strived for decent focus but it differs between the images. Forget about focus. This is a colour test!
- All images were processed in Lightroom, with the DCP profiles I use with each camera. Those DCP profiles are all created with LumaRiver Profile Designer.
- White balance was set on second brightest grey patch on each image.
- Luminance channel was set to around 51 on fourth brightest grey patch. That is the reference value for the Xrite ColorChecker.
- The images were exported at 1500 pixels width and opened in Photoshop.
- Each image was resized to 1500x1050 size. copied and pasted as a new layer in the image.
- All the layers have been aligned. That adds distortion in the mix, but makes it easier to compare color.
- The layers were copied as individual pictures from each layer.
Here is a layered image: http://echophoto.dnsalias.net/ekr/Articles/Temp/Tricolore/Stacked.tif
Best regards
Erik
-
By making profiles for each camera and then processing them in Adobe software you've robbed the Phase One of one of its core assets: that the same company makes the lenses, camera, software, and color profiles. All components are designed together for the best end results.
It's like you've taken a ready-to-race dragster and stripped it down, replaced the tires with generic tires, replaced the chassis with a square block of metal, and put in low-octane fuel. And now you're running a test on the engine versus other engines. But the engine wasn't designed to run on its own; it was designed as part of the overall car.
Also notably this back is from 2007. The improvements Phase One has made in the 11 years since are also pretty profound. The Trichromatic (https://digitaltransitions.com/phase-one-trichromatic-part-2-results/) has the best color of any camera I've worked with (in my highly biased and subjective opinion).
-
Hi Doug,
If I will use any camera, ever, I will be in control of my color profiles.
Best regards
Erik
By making profiles for each camera and then processing them in Adobe software you've robbed the Phase One of one of its core assets: that the same company makes the lenses, camera, software, and color profiles. All components are designed together for the best end results.
It's like you've taken a ready-to-race dragster and stripped it down, replaced the tires with generic tires, replaced the chassis with a square block of metal, and put in low-octane fuel. And now you're running a test on the engine versus other engines. But the engine wasn't designed to run on its own; it was designed as part of the overall car.
Also notably this back is from 2007. The improvements Phase One has made in the 11 years since are also pretty profound. The Trichromatic (https://digitaltransitions.com/phase-one-trichromatic-part-2-results/) has the best color of any camera I've worked with (in my highly biased and subjective opinion).
-
By making profiles for each camera and then processing them in Adobe software you've robbed the Phase One of one of its core assets: that the same company makes the lenses, camera, software, and color profiles. All components are designed together for the best end results.
It's like you've taken a ready-to-race dragster and stripped it down, replaced the tires with generic tires, replaced the chassis with a square block of metal, and put in low-octane fuel. And now you're running a test on the engine versus other engines. But the engine wasn't designed to run on its own; it was designed as part of the overall car.
Also notably this back is from 2007. The improvements Phase One has made in the 11 years since are also pretty profound. The Trichromatic (https://digitaltransitions.com/phase-one-trichromatic-part-2-results/) has the best color of any camera I've worked with (in my highly biased and subjective opinion).
Yes and no.
Erik has repeatedly expressed that he wants exact colours. Erik, if I understand his ideas correctly, wants to use a colour chart and believes that with the correction all colours will be the same and, in turn, will be the same as the colours of the subject. Even if that is not really possible, Erik wants the best approximation. In his ideas, the camera with the colours closest to the subject will be the best camera. Note that this is a perfectly valid workflow if one reproduces paintings, for example.
Exact colour is a different thing than best colour. It seems to me that you are advertising the latter: Phase camera and software would be tuned to give the most pleasing colours. Pleasing may not be exact. For example: skin tones that make the model look younger are considered to be better than accurate skin tones.
Obviously, "more pleasing" is subjective, so what one considers "best colour" depends on the person, even if there are some general lines most people agree on. OTOH, "accurate colour" is objective and there even is an ISO procedure to note it. That procedure is independent of the software.
I should probably add a disclaimer, knowing that I am talking to an official representative from Phase One: a camera - software combination can have both "best" and "accurate" colour, when the software can be told to switch between one output and the other. I am not saying that Phase One cameras cannot have "accurate" colour. In truth, I don't know but I suppose they can since the reproduction of heritage documents is a part of Phase One business.
-
2018-03-31, changed wording, replacing Phase One with Capture One.
Hi,
I guess I may be a bit biased, influenced by landscape photographers Tim Parkin and Joe Cornish who used to be very critical of P45+ colour. I have some private communication with Tim on that issue.
The other point I may make is that it makes perfect sense to achieve accurate colour and apply a "look" to that. Unfortunately, there is a tendency to use colour profiles for that.
My experience with C1 is limited. I have paid for version of three editions, I think, and it supports my P45+ back.
But, I prefer to use Lightroom. Why? Many reasons, I have used it since it's inception 2006. As a matter of fact, I considered writing software like Lightroom back in 2005-2006, but Lightroom did have most of the facilities I wanted. Lightroom is like an 85% match for my needs.
Digital images have a very large dynamic range. Like 13-14EV. Our monitors have like 9EV and prints even less. So, we need to map down the wide DR of digital sensors to what we can display or print. HDR techniques were around a long time, but the tone mappings were often excessive.
2012 pipeline in Lightroom/ACR has put an end to that. New algoritms in Lightroom/ACR made it possible to map highlight and shadow detail into credible images. It could be seen as dirty tricks, but it works. Until this day, I don't think C1 can match.
I have also some bad experience with colour profiles in C1 mixing up colours: http://echophoto.dnsalias.net/ekr/Articles/OLS_OnColor/SimpleCase/
Here Phase One Capture One fully screws up colour, yielding this:
(http://echophoto.dnsalias.net/ekr/Articles/OLS_OnColor/SimpleCase/20150107-CF046070_C1_vsmall.jpg)
Instead of this:
(http://echophoto.dnsalias.net/ekr/Articles/OLS_OnColor/SimpleCase/20150107-CF046070_DI13125_vsmall.jpg)
The latter one being pretty much of the real colour. How I know that? I have seen the flower and measured spectrum. That flower is deep blue purple, end of discussion.
(http://echophoto.dnsalias.net/ekr/Articles/OLS_OnColor/SimpleCase/Violet_vsmall.jpg)
Best regards
Erik
Yes and no.
Erik has repeatedly expressed that he wants exact colours. Erik, if I understand his ideas correctly, wants to use a colour chart and believes that with the correction all colours will be the same and, in turn, will be the same as the colours of the subject. Even if that is not really possible, Erik wants the best approximation. In his ideas, the camera with the colours closest to the subject will be the best camera. Note that this is a perfectly valid workflow if one reproduces paintings, for example.
Exact colour is a different thing than best colour. It seems to me that you are advertising the latter: Phase camera and software would be tuned to give the most pleasing colours. Pleasing may not be exact. For example: skin tones that make the model look younger are considered to be better than accurate skin tones.
Obviously, "more pleasing" is subjective, so what one considers "best colour" depends on the person, even if there are some general lines most people agree on. OTOH, "accurate colour" is objective and there even is an ISO procedure to note it. That procedure is independent of the software.
I should probably add a disclaimer, knowing that I am talking to an official representative from Phase One: a camera - software combination can have both "best" and "accurate" colour, when the software can be told to switch between one output and the other. I am not saying that Phase One cameras cannot have "accurate" colour. In truth, I don't know but I suppose they can since the reproduction of heritage documents is a part of Phase One business.
-
Here is a layered image: http://echophoto.dnsalias.net/ekr/Articles/Temp/Tricolore/Stacked.tif
With that layered image, I redid a gif animation of the colour chart and of the tangerine orange to better show what I think are the most obvious differences. Just clic on the icon to animate...
-
Hi,
Two things that you may keep in the mind.
1) That part of the image may be mapped outside both Adobe RGB and sRGB
2) How that part of the image may depend much on your display and display setting.
Downloading the original image, here: http://echophoto.dnsalias.net/ekr/Articles/Temp/Tricolore/Stacked.tif
and opening in Photshop may yield much more accurate information, as Photoshop may honour your display settings while a web browser may not! Probably not an issues on MacOS X, but a real issue on Windows 10.
Best regards
Erik
-
With that layered image, I redid a gif animation of the colour chart and of the tangerine to better show what I think are the most obvious differences. Just clic on the icon to animate...
That was even visible on a smartphone without superposition.
-
Hi,
Two things that you may keep in the mind.
1) That part of the image may be mapped outside both Adobe RGB and sRGB
2) How that part of the image may depend much on your display and display setting.
Downloading the original image, here: http://echophoto.dnsalias.net/ekr/Articles/Temp/Tricolore/Stacked.tif
and opening in Photshop may yield much more accurate information, as Photoshop may honour your display settings while a web browser may not! Probably not an issues on MacOS X, but a real issue on Windows 10.
Best regards
Erik
On Windows 10 Mozilla displays correctly, even ProPhoto RGB and no shift from PS. It is very much browser dependant.
-
Hi,
You may have noticed the numbers on the fruits. Those correspond to samples I have read with my ColorMunki Photo spectrometer.
I have added most of those samples as a layer on top. Choosing mixing mode "color" overlays the color ignoring the luminousity, I think.
Here is that link: http://echophoto.dnsalias.net/ekr/Articles/Temp/Tricolore/Stacked_with_samples.tif
The fruit that some readers interpret to be a tangerine was called an orange where I bought it.
What I can see the Habeneros 3 and 4 show the largest deviations.
Please note that some colours are outside sRGB.
Best regards
Erik
-
Hi,
Your comments are much appreciated.
I think I commented on this before. The images were processed identically and there was no adjustment directly based on the ColorChecker in the image. The image checker was used for white balance and and exposure adjustment.
On the other hand, I have used color profiles generated by Lumariver Profile Designer and that profiling was based on the standard ColorChecker.
It can be argued, that different raw converters should be used, but say Capture One has a different tone curve and Adobe has hue shifts. With the LumaRiver profiles everything is identical, except the color differences coming from the sensor as converted by the compromise matrix. That would mean that I would need to manipulate the images to have the same contrast.
So, I think that this probably shows the basic differences between the sensors. To clarify, the following controls were used in LR/PS
- White Balance (in LR)
- Exposure adjusted so L in Lab is at 51, using exposure slider in LR
- Camera Calibration is set to LumaRiver based profiles in LR. That is the large equalizing factor and yes, those profiles are based on the ColorChecker
- Exported as TIFFs 1050 high
- TIFFs imported into Photoshop and resized to 1050x1500
- The tiffs copy pasted into a new image as layers
- The layers were algned. This introduces distortion, but makes the images easier to compare
- Text "A", "B" and "C" added
- In the latest release a new layer was added containing sampled data from the fruits. Activating this layer with "color" blending mode shows the colors, without hiding the structures.
The lastest image is here: http://echophoto.dnsalias.net/ekr/Articles/Temp/Tricolore/Stacked_with_samples.tif
Best regards
Erik
The way you present the pictures makes a direct comparison difficult, so I had to extract the 3 images to be able to flip between them. When I flip between the images, I notice the following:
-the colours on the colorchecker chart are identical almost identical (*). Presumably, you used that chart to calibrate the images. Please note that, by calibrating the images, you may have destroyed some of the original colour differences.
-B and C look almost identical on all subjects.
-A looks markedly different on the tangerine (one of the two items 8 ) and a bit different on the chilis marked 4 and 3. Therefore, I vote for A.
-I see almost no difference on the other items. While your images have a range of colours from natural objects (and that is a good thing), I miss skin tones. Interestingly, the tangerine is the object that comes closest to the area where "caucasian" skin tones would be, no other object is pink-orange. This is also the object where the difference is more important.
I attach a smaller version of the separated images, if somebody else wants to do a quick comparison flip.
(*) Edit: with the later posted animation, I see some differences in the blue and violet squares of the chart on image C. I suppose that your calibration was not perfectly correct with that camera on that camera, which may also explain the darker flower petals. Still: the orange patches are matched between all cameras, while the tangerine is not, a proof that cameras do not match the Luther–Ives conditions...
-
I am not sure I understand the new layer with "samples". If I wanted to characterise numerically the fruits colours, I would define a zone to be averaged and let PS tell me about the colour. I did not do any of that, however, I understood that you wanted a visual test.
-
Hi,
Samples are measured data directly of the fruits using a spectrometer. So, these correspond to real measurable colours. The colour boxes were created using Babelcoors PatchTool from the spectral data.
I enclose two images, the first one is showing the samples only, and the second shows the samples layer composed with the A-layer in "color" blending mode.
You can see that the Orange (8) should be a little more orange and the lemon should be a little more green (7).
Best regards
Erik
I am not sure I understand the new layer with "samples". If I wanted to characterise numerically the fruits colours, I would define a zone to be averaged and let PS tell me about the colour. I did not do any of that, however, I understood that you wanted a visual test.
-
Samples are measured data directly of the fruits using a spectrometer.
So you laid your color munki directly on the fruits? I had not understood that.
-
Yes.
So you laid your color munki directly on the fruits? I had not understood that.
-
Hi Doug,
I think you misinterpret this poll a bit. I admit that the title of the poll is a bit misleading.
But, some folks are saying there is a typical MFD color that say the X1D posesses, while the GFX has typical DSLR color. Those cameras use the same sensor even if it is conceivable that they may have different CFAs. Or, maybe not as folks using the X1D with Capture One seem to get very good results.
The intent here was to see if folks recognise MFD colour, if such a colour exists. I would assume that my P45+ would have MFD colour?
Obviously a lot happened since the P45+ and the Alpha 900 were released. The P45+ was from 2007 and the A900 from 2008. So in a sense they are same generation. In another sense the P45+ is proven CCD technology and the A900 is Sony's first full frame CMOS sensor.
Some folks argue that CCD colour is superior to CMOS colour. I don't think that is the position that Phase one has now.
I have also include a lime in the setup, as you indicated that the limefruit has a very high near IR content. Which it indeed has:
(http://echophoto.dnsalias.net/ekr/Articles/Temp/Tricolore/LimeSpectrum.JPG)
So, I asked myself if the high IR content in the lime would cause a yellowish cast on the limefruit. This was not the case with the three cameras tested here. They all reproduced the limefruit extremely well.
The color patches shown above were inserted in the picture on a layer above the lime and blended in "color" mode, that keeps L channel from the base image but uses the "a" and "b" channels from the sample. The patches are not visible in any of the images, so match is virtually perfect.
This comparison is of course limited to what I have here. If you arrange with Phase One to lend me a Thrichromatic over a few days, I will gladly redo the demo.
In this demo I wanted to eliminate raw converter as a factor. In engineering you try to eliminate as many variables as possible. Here I used a tool developed by Anders Torger that is based on pure and honored math, that treats all sensors similarly.
I could have used C1, but C1 has another tone curve and I am pretty sure adds quite a bit more saturation, so, I would need to correct for that manually.
Or, I could have processed all images in C1, but I doubt that the image quality professor spends the same effort on an oddball Sony product, like the Alpha 900 as on the P45+, not to mention the Thrichromatic.
As things are, I can use Capture One with my P45+, as the MFD version is free and that also applies to Sony. But, I prefer to pay for a subscription to the Adobe product. I think that photographers own their images and can use whatever raw processor they prefer.
There are some reasons I prefer Lightroom over Capture One:
- LR has content aware handling of highlights and shadow, maintaining local contrast. Capture One doesn't have that.
- I prefer DNG over natives formats. Capture One has not supported DNG for a long time and their support for DNG may be halfhearted.
- Both DNG format and DCP profiles are well documented that is not the case with C1 colour conversion pipeline.
- I prefer to use my own profiles.
- But, that is just me, I like to have freedom of choice...
Regarding the Thrichromatic, Jack Hogan did a very trough analysis of the sensor, just based on a few sample shots, from yourself and Dave Chew. He found that the Thrichromatic has some significant advantages. His analysis is published in three chapters:
http://www.strollswithmydog.com/phase-one-iq3-100mp-trichromatic-linear-color-i/
http://www.strollswithmydog.com/phase-one-iq3-100mp-trichromatic-linear-color-ii/
http://www.strollswithmydog.com/phase-one-iq3-100mp-trichromatic-linear-color-iii/
Jack's analysis makes sense, perhaps Phase One should consider getting permission to use it in their marketing. Sharing good and honest info is not a bad idea.
And just to say, Jack likes the Trichromatic.
Best regards
Erik
By making profiles for each camera and then processing them in Adobe software you've robbed the Phase One of one of its core assets: that the same company makes the lenses, camera, software, and color profiles. All components are designed together for the best end results.
It's like you've taken a ready-to-race dragster and stripped it down, replaced the tires with generic tires, replaced the chassis with a square block of metal, and put in low-octane fuel. And now you're running a test on the engine versus other engines. But the engine wasn't designed to run on its own; it was designed as part of the overall car.
Also notably this back is from 2007. The improvements Phase One has made in the 11 years since are also pretty profound. The Trichromatic (https://digitaltransitions.com/phase-one-trichromatic-part-2-results/) has the best color of any camera I've worked with (in my highly biased and subjective opinion).
-
So, when do we get the answer to the question? ::)
-
Hi,
Voting is open until April 2, so I guess that I will share the info the third of April.
For those who cannot wait, the answer is here: http://echophoto.dnsalias.net/ekr/Articles/Temp/Tricolore/Answer.jpg
Best regards
Erik
So, when do we get the answer to the question? ::)
-
Hi, I don’t have or tried those cameras, but I have used Hasselblad 31 and 60mp as well full frame Nikon&Canon cameras, and speaking about “color” the biggest difference is hue gradations...
-
Hi,
Voting is open until April 2, so I guess that I will share the info the third of April.
For those who cannot wait, the answer is here: http://echophoto.dnsalias.net/ekr/Articles/Temp/Tricolore/Answer.jpg
Best regards
Erik
I won. Erik, next time you do that, make sure to prize a P45!
-
I won.
Really? ::)
-
Hi,
Voting is open until April 2, so I guess that I will share the info the third of April.
For those who cannot wait, the answer is here: http://echophoto.dnsalias.net/ekr/Articles/Temp/Tricolore/Answer.jpg
Best regards
Erik
Hi Erik -
What sort of light source was utilized for the images in the contest?
Steve Hendrix/CI
-
Hi Steve,
This was not intended to be a contest, I would more see it as demo.
Illumination was electronic flash, it was some time ago I shot the images, so I don't remember setup. But, as far as I recall main light was a Godox Wistro AD200 using a reflective umbrella and fill light was an Elinchrome D-Lite with a soft box.
Best regards
Erik
Hi Erik -
What sort of light source was utilized for the images in the contest?
Steve Hendrix/CI
-
Hi,
The answer is:
A) P45+
B) Sony A7rII
C) Sony Alpha 900
There were 20 votes (I did a vote of my own to test the polling - on "Neither is P45+")
So we could see that the P45+ got four votes to be P45+, the Sony A7rII got six and the Sony Alpha 8.
I am not sure all the votes were placed for the right reason. This was intended as a colour demo.
I am thankful for all good comments.
Now, a few reflections of my own.
The major reason of choice of cameras was that I own all three, but there were other factors.
- The P45+ is known to have tricky color, would it show up?
- The P45+ is medium format, it is often said that there is a medium format color. Also, it has been said on these forums that Hasselblad X1D has Medium Format Color while the Fuji GFX has DSLR color. Would observers be able to discern that MFD color in something that has some elements of a controlled comparison?
- The P45+ is probably said to have a CCD look.
- The P45+ was released 2007 and the A900 2008, so they are similar generation.
- Some experts on color regard the Alpha 900 to have the best compromis in CFA design in commercially available cameras.
- The Alpha 900 was taken for P45+ most frequently. That may be because for good or different color. Don't know!
- The sampled colours were generally well rendered, with some deviations on the red chilly peppers.
- The lime fruit was not affected by high near IR reflection spectra, kit reproduced extremely well with all sensors.
It would be nice to include like thing like human skin in test scene, but that is nothing I can buy in a grocery store, thanks god for that...
Best regards
Erik
-
It would be nice to include like thing like human skin in test scene, but that is nothing I can buy in a grocery store, thanks god for that...
You could hire a model. More simply, you could leave one of your hands in the picture.
From a look at the thread, it seems that people who voted C did so because of the flowers. But it seems that the flowers appeared different to their eyes because of the influence of their surroundings. When I posted about the effect at least one person changed his mind to A. Deep purple is also poorly calibrated on camera C (look at the colour target), which again influences the flower.
Also: when MF cameras are said to have different colours, users refer to (caucasian) skin tones. Fashion is one of the key market for MF cameras. Your simple test would indicate that there could be some truth to that as the object with the most noticeable change in colour (the orange) is also the one closer to skin tones (I would say the skin of a famous US president is the same colour...). In any case, it proves that there is noticeable metamerism, as the orange patch on the colour target does not change while the fruit does.
-
Hi,
The answer is:
A) P45+
B) Sony A7rII
C) Sony Alpha 900
There were 20 votes (I did a vote of my own to test the polling - on "Neither is P45+")
So we could see that the P45+ got four votes to be P45+, the Sony A7rII got six and the Sony Alpha 8.
I am not sure all the votes were placed for the right reason. This was intended as a colour demo.
I am thankful for all good comments.
Now, a few reflections of my own.
The major reason of choice of cameras was that I own all three, but there were other factors.
- The P45+ is known to have tricky color, would it show up?
- The P45+ is medium format, it is often said that there is a medium format color. Also, it has been said on these forums that Hasselblad X1D has Medium Format Color while the Fuji GFX has DSLR color. Would observers be able to discern that MFD color in something that has some elements of a controlled comparison?
- The P45+ is probably said to have a CCD look.
- The P45+ was released 2007 and the A900 2008, so they are similar generation.
- Some experts on color regard the Alpha 900 to have the best compromis in CFA design in commercially available cameras.
- The Alpha 900 was taken for P45+ most frequently. That may be because for good or different color. Don't know!
- The sampled colours were generally well rendered, with some deviations on the red chilly peppers.
- The lime fruit was not affected by high near IR reflection spectra, kit reproduced extremely well with all sensors.
It would be nice to include like thing like human skin in test scene, but that is nothing I can buy in a grocery store, thanks god for that...
Best regards
Erik
Erik,
If I understand the nature of your topic and efforts to put it to test in a scientific methodology, I think this essay, as presented, could only be but a dead end road for various reasons.
First we'd had to define colour accuracy and once defined for the all chain involved and not just camera sensors; in wich scenarios such coulour accuracy will influence or not the par photographer/photographic tools.
Is there such thing as absolute coulour accuracy? considering that the camera sensor and color science is only a small part of the chain in which we have to consider displays, lightning sources, printing/viewing but also retouching within the nature of the work ( Museum reproduction has very little to do with fashion photography) and the psychologic nature of mankind.
It is all about interpretation and mind is conditionned.
As mind is conditionned from the very roots, perception is therefore altered by the programming within the limitations of the human apparel.
The conditionned perception alters the way engineers consider colour accuracy as correct and for a certain purpose and within a determined space and time. (Cultural codes, learned values etc...)
One little alteration in all the chain, such as displays calibrations, will break accuracy.
And accuracy is also linked to the nature of the work. Where this accuracy matters? Skin tones? But who shoots skin tones?
If black people were rulling the industrial world and were the onces making gear, skin tone references will be something very different than what is being accepted as "pleasing". The sensation that a skin tone remains pleasing to our eyes is completly conditionned by the dominant societies.
As a matter of fact what we have are as many colour sciences as brands, targetting a specific consummer.
Instead of accuracy, we can merely talk about "optimized for".
So there are of course no magic MF coulour the same way as there aren't magic dslrs colours either.
All what people see are illusions.
Being able to perceive differences in a blind test between colours is doable for everyone but does not bring any sort of conclusion in a way or another.
Then we have the question if it influences the choice of a system when it comes to invest money in.
As we are talking about systems and costs, there are so many different parameters that come into consideration before colour and even resolution.
Disponible cash is one of them.
Pro service, flash system implementation performances, electronic aspects, AF, lenses etc...that we could consider practical facts (a tank built camera or a consummer one are facts. AF, flash capabilities, and costs are facts). Then, there are those more volatile concepts, interpretations and preferences such as colour, handling, photographic style, and something as simple as falling in love with a gear can make one perceive more goodies than there really are. (Same with relashionships).
And more importantly. Nobody ever decides anything here. You do not decide to be Erik. You are this Erik despite yourself. You never decided to be an engineer oriented. It just happened. There has never been a will, a complete consciousness about the fact that you became interested in images and photography more than in geology or litterature. There is no decision on colour tastes either. It happens despite yourself and evolves with datas. Datas are conditionning, information is conditionning. It transforms perception, ideas, beleifs and so on.
Nobody ever controls anything. We just have the sensation that there is a chooser, but there is no such thing as a chooser in the sense generaly accepted.
Pizza or pasta?
Red wine or white wine?
Nike or Adidas?
Phase one or Hasselblad or Nikon? Nobody can even tell scientificaly how this ends to be once's tool more than this other. Some people loves Hasselblad and think that the Fuji is crappy dslr overdone colour. Other see exactly the opposite. Why? It happens by itself. There is no thruth whatsoever.
So one do researches, test gear, try to find a rational explaination in order to "choose" this upon this other.
A choice is made apparently based on this investigation. But the choice was already made before all that happened.
All the ruminations are a parade, a show of the choice itself. In other words, the conclusion comes first (but we don't know it), then apoears the argumentation, and finaly the conclusion is presented again in consciousness as the result.
This is the big illusion. This is only a play.
Will science irrupts to erradicate the human factor and bring datas on the table so we can rely on absolute facts instead? That is relevant for the camera and software maker. Then, the camera brand will alter this celestial order using the very same engineering because they are targetting a specific client. Therefore, we are in marketing.
Colour matrixes, complex science will be involved in order to alter the look and have a signature that others do not.
Objects are not what they are because this is what they are but because of a choice of an intention.
Therefore, everything is distortionned and nothing is accurate as such.
Anything that people say you mentionned is right nor wrong.
All there is are interpretations of the same play.
I tend not to like Sony colour science. Each time I have a Sony image I say "I don't like the files!"
Is there a rational reason why I like Nikon and not Sony? Nope. It happens.
I can not come to the conclusion that Sony is crap and Nikon is best. It would be absurd.
I don't like Sony the same way I don't feel atrracted by this striking blonde other guy find attractive.
May I extract complex argumentation in order to explain my point of view, someone would bring the antitesis
With a no less valuable argumentation.
This is a dead end road, not because the way you did the test, but because
There can not be an answer reliable enough.
Best regards.
-
Hi Fred,
- The idea was not to have a competition.
The posting was a reaction to a statement that Hasselblad X1D has MFD color och Fuji GFX has DSLR color. I don't know who used those words.
But, if there is an obvious difference between MFD color and DSLR color than it must be easy to which one is which. Four voters have guessed the correct camera. That was Landscapephoto and yourself and two others. But the other 16 voters were not able to pick the P45+, for whatever reason.
I was not really interested in the outcome. The question asked was not "Which camera has the best color?" but "Which of the three cameras is the P45+?".
Of course, the issue was muddled by using Lumariver generated profiles, instead of using Capture One.
I am pretty much aware that accurate color reproduction seldom is the aim in commercial photography, that does not hinder me in trying to achieve natural colors in my landscape photography.
Also, I would suggest that this forum is not about professional photography, but about "Medium Format / Film / Digital Backs – and Large Sensor Photography" and I am pretty sure that a large part of the posters here shoot mainly for pleasure.
Best regards
Erik
-
Also, I would suggest that this forum is not about professional photography, but about "Medium Format / Film / Digital Backs – and Large Sensor Photography" and I am pretty sure that a large part of the posters here shoot mainly for pleasure.
Best regards
Erik
I didn't think your thread as a competition. I was of course kidding when I said "I won". It was to tease a bit and make you smile. I did understand it the way you described.
And you are correct: you did not ask which has the best colour but which is which. This is IMO the only possible correct way to present the topic.
But I was refering more to the thread that inspired your test, in which yes there were
Interpretations and valorations about best/better/worst etc...me, being one of them also.
(Then I reread myself and find a complete nonsense of my own words and other's too. ;D)
The fact that you might demostrate that if there is a difference it should be visible easily (and you are right)
And the fact that many made mistake (as expected), will not prevent those claims to keep going.
We will still have best-better-worse and astrovisions, magical differences etc...
That is inevitable.
In fact I like your posts. Your mind is of a scientist, and I'm not.
Sometimes I think that scientists think too much about details, but I like the way you focus it in general,
And more importantly the tone is always of a gentleman.
One test that I'd like to do is this: 2 people with the same camera are seeing the same flower.
In the exact same moment, angle and location.
A picture is taken. Will those 2 persons see exactly the same accuracy on the same display?
We have the flower in front of us, the image in front of us and compare.
We could compare using several softwares.
Will there be a consens? Sometimes yes but sometimes probably no.
I hope the pros photographers who frequent regularly this forum shoot also for passion/pleasure.
Just that they make their incomes with it.
Photography is probably an area one can not really do without pleasure as it happens in many other jobs where people hate what they do.
Best regards.
-
First we'd had to define colour accuracy and once defined for the all chain involved and not just camera sensors
Actually, colour accuracy is perfectly defined and there are iso standards to measure it. Colour accuracy is a fact.
You are thinking about "more pleasing" colours. Obviously, "more pleasing" is subjective, so what one considers "best colour" depends on the person, even if there are some general lines most people agree on.
Both are very different matters and taking one for the other is not going to bring us anywhere. Also: both are perfectly valid in photography: accuracy is necessary for the pro reproducing paintings, pleasantness is essential for the portraitist.
-
Actually, colour accuracy is perfectly defined and there are iso standards to measure it. Colour accuracy is a fact.
You are thinking about "more pleasing" colours. Obviously, "more pleasing" is subjective, so what one considers "best colour" depends on the person, even if there are some general lines most people agree on.
Both are very different matters and taking one for the other is not going to bring us anywhere. Also: both are perfectly valid in photography: accuracy is necessary for the pro reproducing paintings, pleasantness is essential for the portraitist.
I'm not denying this and that wasn't my point really. You are talking about procedures and isolate a phrase maybe?
What I'm saying are 2 things.
1) there are color sciences involved in both camera and softwares that alter color representations and focus on certain aspects, within the same gamma and colour spaces. Therefore we can not talk about colour accuracy
In a stric term but merely ”optimized for”, which is the term I used.
Standardized procedures aren't here to provide absolute reliability but to establisb acceptable limits
Necesary for interchange between collaborative environements.
Procedures in themselves change, evolve constantly. You can expand or reduce colour spaces etc...
2) Accuracy is entirely display dependant. Just one element in the all chain, not just sensors, that fails will alter colour accuracy inevitably.
So it's all about getting close to an ideal situation without never reaching
Absolute accuracy.
3) what we can see in test images is not colour accuracy but about different colour sciences
Applied by different manufacturers.
Colour accuracy is a myth. Just like the speed of light, the more you want to get close, the more it will escapes.
All we do is establishing some boundaries as standarts so that it's manageable within margin of acceptable "errors".
The defined and established standarts that help us to reproduce with accuracy in print or in screens the intended colour decisions that we make based on displays wasn't IMO the topic but if there was some sort of better accuracy between 2 systems from capture? If so, everyone should be able to see the same properties. Erik's experiment shows that there is no such thing.
-
I didn’t vote because Erik’s question was ,”Which one is the P45+?” I’ve never owned a P45+ so how would I know??
He didn’t ask which had the better color, more accurate color, the most pleasing color or the magical MF color. Just which was the P45+. I’ve never owned one, never shot one, never seen a raw image from one so there is now way I could possibly guess.
Dave
-
He didn’t ask which had the better color, more accurate color, the most pleasing color or the magical MF color. Just which was the P45+. I’ve never owned one, never shot one, never seen a raw image from one so there is now way I could possibly guess.
Notably, even if you had owned and used a P45+ the test wouldn't be any more relevant.
You'd have to be a P45+ owner who preferred to use 3rd party software rather than use Capture One, the software whose main purpose for being is to get the most out of Phase One raw files. AND you'd have to be a P45+ owner who preferred to create your own color profile to using the profiles made by the team who made the hardware and software.
In 10 years working with many many Phase One owners, I can't think of a single client of mine who fits this description. Maybe 95% of my clients are using Capture One and 1% are making their own profiles, and the ones who are making their own profiles are doing so in Capture One.
It's an interesting test of the LumaRiver Profile Designer, and anyone who is really interested in that software should greatly appreciate Erik's time in writing it up. But it's not a test that is particularly relevant to any current or prospective Phase One owner I've met.
-
Notably, even if you had owned and used a P45+ you wouldn't be any better off.
You'd have to be a P45+ owner who preferred to use 3rd party software rather than Capture One AND preferred to create your own color profile to using the profiles made by the team who made the hardware and software.
In 10 years working with many many Phase One owners, I don't think I can think of a single client of mine who fits this description.
Exactly. That is my point. Any changes on the chain will affect results. If I shot Arri log and
Use a Sony or Redlogfilm to rec709 then colour will be affected. And there are more people that do that than we might think.
But again, Phase coulour science is not about absolute colour accuracy myth but about correct maths within
A look. For some it is the look there are looking for, others wlll prefer Hasselblad colour science.
The only think that ensure you when work with P.45 + capture one + team profiles is that the underlying maths are correct.
It does not garantee one will "see" accuracy of some sort.
It works as expected. That is all.
-
Exactly. That is my point. Any changes on the chain will affect results. If I shot Arri log and
Use a Sony or Redlogfilm to rec709 then colour will be affected. And there are more people that do that than we might think.
But again, Phase coulour science is not about absolute colour accuracy myth but about correct maths within
A look. For some it is the look there are looking for, others wlll prefer Hasselblad colour science.
The only think that ensure you when work with P.45 + capture one + team profiles is that the underlying maths are correct.
It does not garantee one will "see" accuracy of some sort.
This is true for the commercial profiles.
Notably, when using the CH profiles and the CH Color Guide Reproduction Workflow (https://dtdch.com/color-reproduction-guide-cultural-heritage/) (for which I was a main author), the intention (and result) is accuracy.
Using that workflow for pictures of people, places, and things makes you very quickly aware of why 99.9% of photographers don't benefit from a profile and workflow engineered for brutal and unforgiving accuracy.
-
This is true for the commercial profiles.
Notably, when using the CH profiles and the CH Color Guide Reproduction Workflow (https://dtdch.com/color-reproduction-guide-cultural-heritage/) (for which I was a main author), the intention (and result) is absolute accuracy.
Doug...when the word "absolute" appears, I have a big caution warning flashing light that switch on. Absolute zero, absolute beauty. But I see what you mean.
Specialy in our display based industry. Always is a display between us and the object.
This is why I prefer talking about correct maths. Or better maths provided by the company based on their researches and knowledge of their products.
-
Doug...when the word "absolute" appears, I have a big caution warning flashing light that switch on. Absolute zero, absolute beauty.
Absolutely :).
I actually modified my post to remove absolute before your post objecting to the word. I should have said "Damn good" or similar and defined what that means in this context.
In the case of the Cultural Heritage profiles the goal is to consistently provide color accuracy better than the requirements of FADGI-4 star or ISO 19264 imaging guidelines for preservation-grade imaging (which in turn require specific mean and maximum deltaE2000 values as measured on specific targets). Based on extensive experience at a variety of high-end institutions (https://dtdch.com/our-clients/) the system (back + body + lens + software + CH profiles) great exceed those quality requirements both in theory and in practice.
-
I'm not denying this and that wasn't my point really. You are talking about procedures and isolate a phrase maybe?
What I'm saying are 2 things.
1) there are color sciences involved in both camera and softwares that alter color representations and focus on certain aspects, within the same gamma and colour spaces. Therefore we can not talk about colour accuracy
In a stric term but merely ”optimized for”, which is the term I used.
Standardized procedures aren't here to provide absolute reliability but to establisb acceptable limits
Necesary for interchange between collaborative environements.
Procedures in themselves change, evolve constantly. You can expand or reduce colour spaces etc...
2) Accuracy is entirely display dependant. Just one element in the all chain, not just sensors, that fails will alter colour accuracy inevitably.
So it's all about getting close to an ideal situation without never reaching
Absolute accuracy.
3) what we can see in test images is not colour accuracy but about different colour sciences
Applied by different manufacturers.
Colour accuracy is a myth. Just like the speed of light, the more you want to get close, the more it will escapes.
All we do is establishing some boundaries as standarts so that it's manageable within margin of acceptable "errors".
The defined and established standarts that help us to reproduce with accuracy in print or in screens the intended colour decisions that we make based on displays wasn't IMO the topic but if there was some sort of better accuracy between 2 systems from capture? If so, everyone should be able to see the same properties. Erik's experiment shows that there is no such thing.
I don't think we are talking about the same thing. I was saying that colour can be measured, and the measuring error can be quantified. I am saying that colour can be reproduced (and indeed that also implies a display) and that the error between input and reproduction can also be quantified. The magnitude of the error defines "accuracy".
-
Hi Doug,
The software is not Luminar but LumaRiver Profile Designer. It is an excellent piece of software that deserves to be called by the right name.
So, I would appreciate if you could correct your posting.
Other than that, I would not argue with your viewpoints.
Best regards
Erik
Notably, even if you had owned and used a P45+ the test wouldn't be any more relevant.
You'd have to be a P45+ owner who preferred to use 3rd party software rather than use Capture One, the software whose main purpose for being is to get the most out of Phase One raw files. AND you'd have to be a P45+ owner who preferred to create your own color profile to using the profiles made by the team who made the hardware and software.
In 10 years working with many many Phase One owners, I can't think of a single client of mine who fits this description. Maybe 95% of my clients are using Capture One and 1% are making their own profiles, and the ones who are making their own profiles are doing so in Capture One.
It's an interesting test of the Luminar profiling software, and anyone who is really interested in that software should greatly appreciate Erik's time in writing it up. But it's not a test that is particularly relevant to any current or prospective Phase One owner I've met.
-
I don't think we are talking about the same thing. I was saying that colour can be measured, and the measuring error can be quantified. I am saying that colour can be reproduced (and indeed that also implies a display) and that the error between input and reproduction can also be quantified. The magnitude of the error defines "accuracy".
This kind of essay (link below) covers some aspects involved
In the topic.
But it's not for everyone and reflects the complexity of this subject.
I guess beyond the scope of what a forum conversation generaly allows.
(proprietary matrixes from XYZ etc...the principles explained remain relevant
To understand the aspects involved in colour representation/accuracy.)
In motion pictires there are even more complications.
Also, I don't use gamma 2.2 on my displays but 2.4 and prefer to rely on perception.
Digging on colour accuracy is on the very edge of what my brain legitimaly tolerates.
Link to part 1 at the beginning.
http://www.odelama.com/photo/Developing-a-RAW-Photo-by-hand/Developing-a-RAW-Photo-by-hand_Part-2/
-
Absolutely :).
;D ;D
That was an absolute good one!
-
Hi Dave,
The software is not Luminar but LumaRiver Profile Designer. It is an excellent piece of software that deserves to be called by the right name.
So, I would appreciate if you could correct your posting.
Other than that, I would not argue with your viewpoints.
Noted and corrected.
Ironically though... my name is not Dave but Doug. :D
-
Hi Doug,
I am very much aware of that! Deeply sorry!
Best regards
Erik
Noted and corrected.
Ironically though... my name is not Dave but Doug. :D