Pages: 1 [2] 3   Go Down

Author Topic: HR-1 SuperChroma  (Read 6701 times)

sea-speak

  • Newbie
  • *
  • Offline Offline
  • Posts: 8
Re: HR-1 SuperChroma
« Reply #20 on: May 17, 2020, 01:07:50 am »

hi Jim,

Thanks for the post on your site, that is constructive and does help explain the problem. That said, what you don't address is what's the best we *do* currently do (practically speaking) -- not to be confused with what the best we *can* currently do might be.

I.e. out in the messy real world of taking pictures and printing them and trying to get 'what I saw' to 'what I want' to 'what I see on my screen/print', is there a workflow, including a calibration target, that addresses the metameric issues you describe as well as the too-few-sample patches issue (from your post, it sounds like solving this would require calibration targets comprising actual human skin, which is probably not where we want to go...)? If yes, what is it? If no, what is in your opinion the best solution that is commercially relevant and realistically implementable by an enthusiast photographer (rather than, say, a major Hollywood studio, etc.)?

regards,

Brandt




Logged

sea-speak

  • Newbie
  • *
  • Offline Offline
  • Posts: 8
Re: HR-1 SuperChroma
« Reply #21 on: May 17, 2020, 01:15:36 am »

Put another way, there are the following possibilities and ensuing questions:

1 - the problem is insoluble.
Q: what's the best practical solution that we have now, and are there any obvious ways to improve upon it

2 - the problem is soluble, but has not been solved.
Q: Again, what's the best practical solution currently available, and what would be required to improve this solution to the point that it meets the challenge

3 - the problem is soluble, and has been solved
Q: Is this solution commercially available, and if so, practical for individuals to implement.



Logged

Jim Kasson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2370
    • The Last Word
Re: HR-1 SuperChroma
« Reply #22 on: May 17, 2020, 10:02:59 am »

Put another way, there are the following possibilities and ensuing questions:

1 - the problem is insoluble.
Q: what's the best practical solution that we have now, and are there any obvious ways to improve upon it

2 - the problem is soluble, but has not been solved.
Q: Again, what's the best practical solution currently available, and what would be required to improve this solution to the point that it meets the challenge

3 - the problem is soluble, and has been solved
Q: Is this solution commercially available, and if so, practical for individuals to implement.

As long as cameras don't meet the Luther-Ives condition, the problem is not perfectly solvable. However, we can come pretty close, for limited lighting spectra and limited subject matter. For those, the problem is close to solvable, given a camera with a tractable set of color filter array dyes and an appropriate lighting source. A demonstration of this is the success of heritage preservation photographers in color reproduction. The key to successfully doing this is using patches of the materials to be photographed, illuminated by light that is the same as will be used for the actual photography. Gather the patches, measure them with a spectrophotometer, photograph them, input the spectra to the profile making software along with the raw file.

The problem can also be nearly solved, and more accurately, by using a monochrome sensor and a filter wheel consisting of more than three segments. But there is no nonspecialist commercial software to deal with such a camera. As the number of segments in the filter wheel increases, the accuracy of the solution improves.

But most photographers don't want accurate color. Hunt defines possible objectives for color reproduction:

Spectral color reproduction, in which the reproduction, on a pixel-by-pixel basis, contains the same spectral power distributions or reflectance spectra as the original.

Exact color reproduction, in which the reproduction has the same chromaticities and luminances as those of the original.
Colorimetric color reproduction, in which the reproduced image has the same chromaticities as the original, and luminances proportional to those of the original.

Equivalent color reproduction, in which the image values are corrected so that the image appears the same as the original, even though the reproduction is viewed in different conditions than was the original.

Corresponding color reproduction, in which the constraints of equivalent color reproduction are relaxed to allow differing absolute illumination levels between the original and the reproduction; the criterion becomes that the reproduction looks the same as the original would have had it been illuminated at the absolute level at which the reproduction is viewed.

Preferred color reproduction, in which reproduced colors differ from the original colors in order to give a more pleasing result.

Most photographers want the last. That gives a lot of wiggle room, although it introduces an element of subjectivity. All the commercial raw developers are set up to attempt to deliver the last by default, and there are huge controversies about how well they do that, with hotly-debated opinions on all sides.

I see this is getting long. To be continued.

Jim

Jim Kasson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2370
    • The Last Word
Re: HR-1 SuperChroma
« Reply #23 on: May 17, 2020, 10:33:00 am »

3 - the problem is soluble, and has been solved
Q: Is this solution commercially available, and if so, practical for individuals to implement.

If you're not too picky about accuracy, the problem has been solved, in various ways. In my opinion, the most accurate way to generate a color profile for a camera is to measure the spectral responses of the raw channels, and compute the profile from there. However, that way requires equipment that most photographers don't have, and, more to the point, don't know how to use. There are two main ways to do that. One is to photograph -- with no lens on the camera -- the image cast by a grating or a prism. Now we have the spectral response of each channel, but we don't know the amount of energy at each wavelength. To get that we need to measure the output of the rainbow-generating device vs wavelength. Put those to together and we have the spectral response of the channels. There are endless variations on this theme. You can also buy a device that has many individually-filtered spectrally disparate light sources, and that's the other method. I believe, but am not sure, that Adobe and others generate their profiles by looking at the spectral response of the cameras.

Whether doing the above is practical for individuals to implement is highly dependent on the individual. I would say that most people don't have the time, technical ability, mathematical and programming skills, and, most importantly, interest, to do that kind of work. Remember, most people don't want accurate color anyway.

Many people are entirely satisfied by photographing commercial patch sets. However, for best results, you should measure your patch set with a spectrophotometer, since there is variation in the patches and they change with age. The X-rite SG patch set delivers more accurate results than the CC24, in my experience. With some color profile making software, you can add your own patches, measured with your own spectrophotometer.

If your camera obeyed the Luther-Ives condition, you could use patches that covered the range of colors that you're interested in. Since it does not, you must use patches that cover the range of spectra that you're interested in.

Most people are satisfied with commercial profiles, once they find one they like. Many who are not find that they can edit those profiles to their satisfaction. Adobe provides tools to do that.

Here is some background reading:

https://blog.kasson.com/the-last-word/the-color-reproduction-problem/

https://en.wikipedia.org/wiki/Tristimulus_colorimeter

https://blog.kasson.com/nikon-z6-7/camera-differences-in-color-profile-making/

Jim



DP

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 727
Re: HR-1 SuperChroma
« Reply #24 on: May 17, 2020, 11:11:32 am »

I don't believe any serious professional photographer or museum will use cheap chinese lights for their work.
right, right - their for exampe will use chinese LEDs from the likes of Yuji  ;D which will be put in nice boxes by the likes of https://dtculturalheritage.com/dt-photon-custom-cultural-heritage-lighting/

Logged

DP

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 727
Re: HR-1 SuperChroma
« Reply #25 on: May 17, 2020, 11:20:51 am »

. And with high quality lights (Profoto, Broncolor...) the spectral curve is really smooth.

and if you refer to strobes you imagine that they have some special xenon bulbs ;D - what they can boast sometimes is more stable spectrum vs power output range, but then you can use way cheaper PCB Einstein

Logged

sea-speak

  • Newbie
  • *
  • Offline Offline
  • Posts: 8
Re: HR-1 SuperChroma
« Reply #26 on: May 17, 2020, 11:23:29 am »

hi Jim,

These two posts, plus the one on your blog from this week, are a very helpful crash course on this problem. Thank you.

I am a hobbyist photographer, and qua hobbyist I'm a little more interested in 'what I want' rather than 'what my eyes really saw', though I prefer to stay relatively closely tethered to 'what I saw.' I'm not a follower of the Renoir/Kincade/Peter Lik school.

Meanwhile, after 10 years of development, I'm close to launching a very high-end consumer product that will be painted in colors of the customer's choice. This product will be sold online. Ultimately samples of the actual paint on the actual substrate will have to be sent to customers for approval. However it would save a lot of time and aggravation if the colors I see on my monitors match as closely as possible the automotive paints used. Carrying this accuracy through to customer viewing devices and print collateral would also be very helpful. I realize that I can't make customers profile their smartphones, but I want to control what I can. Qua product vendor, I care about accuracy a great deal.

From what you've said, it sounds like the practical answer from your perspective is an X-rite setup using their SG target. (?) One minor question, do you know what technology they use to print their target? They provide basically zero technical information on their site about the product.

regards,

Brandt


Logged

Jim Kasson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2370
    • The Last Word
Re: HR-1 SuperChroma
« Reply #27 on: May 17, 2020, 11:30:17 am »


From what you've said, it sounds like the practical answer from your perspective is an X-rite setup using their SG target. (?) One minor question, do you know what technology they use to print their target? They provide basically zero technical information on their site about the product.


I'll get to the practical answer in another post, but I can deal with this on fairly simply. The patches in the X-Rite targets consist of paints that were mixed to have the desired spectrum. The spectra that they chose consist of what they consider to be commonly-photographed objects.

If you know anything about offset printing, they are spot colors, not process colors.

I don't know for sure, but I imagine that they spray paint substrates, then cut those up into the patches, and then glue the patches to the backing board of the target.

Jim

Jim Kasson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2370
    • The Last Word
Re: HR-1 SuperChroma
« Reply #28 on: May 17, 2020, 11:32:39 am »

Meanwhile, after 10 years of development, I'm close to launching a very high-end consumer product that will be painted in colors of the customer's choice. This product will be sold online. Ultimately samples of the actual paint on the actual substrate will have to be sent to customers for approval. However it would save a lot of time and aggravation if the colors I see on my monitors match as closely as possible the automotive paints used. Carrying this accuracy through to customer viewing devices and print collateral would also be very helpful. I realize that I can't make customers profile their smartphones, but I want to control what I can. Qua product vendor, I care about accuracy a great deal.

From what you've said, it sounds like the practical answer from your perspective is an X-rite setup using their SG target.

I'd use some actual paint samples for the targets. Have you looked at the paint sample spectra to see how well-behaved they are?

Jim

Jim Kasson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2370
    • The Last Word
Re: HR-1 SuperChroma
« Reply #29 on: May 17, 2020, 11:36:38 am »


Meanwhile, after 10 years of development, I'm close to launching a very high-end consumer product that will be painted in colors of the customer's choice. This product will be sold online. Ultimately samples of the actual paint on the actual substrate will have to be sent to customers for approval. However it would save a lot of time and aggravation if the colors I see on my monitors match as closely as possible the automotive paints used. Carrying this accuracy through to customer viewing devices and print collateral would also be very helpful. I realize that I can't make customers profile their smartphones, but I want to control what I can. Qua product vendor, I care about accuracy a great deal.


If you want to put colorimetrically correct patches on your web site, you don't need a camera at all. Just measure the paint chips with a spectrophotometer, calculate the colorimetric values in sRGB when they are lit by D65 light, and use those values for your posted color patches.

Jim

DP

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 727
Re: HR-1 SuperChroma
« Reply #30 on: May 17, 2020, 11:41:57 am »

From what you've said, it sounds like the practical answer from your perspective is an X-rite setup using their SG target.

if you want to create profiles using targets - get a decent spectrophotometer, decent light (to go above high 2K or very low 3K spectrum you need to get a strobe or if you want a constant light then Solux / https://www.solux.net/cgi-bin/tlistore/index.html / or for example Yuji sells some ready LED solutions if you can't DIY  = https://store.yujiintl.com/collections/high-cri-led-lights ),  rawdigger & free dcamprof ( https://torger.se/anders/dcamprof.html ) and you can use "synthetic targets" buy shooting shooting various materials separately and combining the data (also read _completely_ the megathread @ https://forum.luminous-landscape.com/index.php?topic=100015.0 )
Logged

DP

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 727
Re: HR-1 SuperChroma
« Reply #31 on: May 17, 2020, 11:45:53 am »

I'd use some actual paint samples for the targets.

enterprising person who is capable of DIY work can hit a proper supply shop and buy actual pigments to make his own paints to paint patches for targets
Logged

DP

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 727
Re: HR-1 SuperChroma
« Reply #32 on: May 17, 2020, 11:50:59 am »

If you want to put colorimetrically correct patches on your web site, you don't need a camera at all. Just measure the paint chips with a spectrophotometer, calculate the colorimetric values in sRGB when they are lit by D65 light, and use those values for your posted color patches.
indeed - I'd just suggest to Sea-Speak to get software like BabelColor Patchtool to automate the work with that to some extent (it is not that expensive product)
Logged

Jim Kasson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2370
    • The Last Word
Re: HR-1 SuperChroma
« Reply #33 on: May 17, 2020, 11:51:13 am »

enterprising person who is capable of DIY work can hit a proper supply shop and buy actual pigments to make his own paints to paint patches for targets

Sounds like the poster already has access to samples: " Ultimately samples of the actual paint on the actual substrate will have to be sent to customers for approval. "

DP

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 727
Re: HR-1 SuperChroma
« Reply #34 on: May 17, 2020, 12:12:08 pm »

Sounds like the poster already has access to samples: " Ultimately samples of the actual paint on the actual substrate will have to be sent to customers for approval. "

indeed, as you noted above he does not need to photograph anything... unless he wants to photograph the ready product as a whole
Logged

sea-speak

  • Newbie
  • *
  • Offline Offline
  • Posts: 8
Re: HR-1 SuperChroma
« Reply #35 on: May 17, 2020, 01:39:56 pm »

hi,

Now we're cooking with gas. A few items.

-I'm a hobby letterpress printer. Totally get the spot vs process distinction.
-I haven't looked at actual paint samples yet, nor measured, to see how well-behaved their spectra are.
-I can definitely get some test panels painted and measured with a spectrophotometer. I'm a few months away from that point but will get there.
-Babelcolor patchtool is reasonably priced, no problem to get that software.

Questions
-Re lights - I put Soraa Vivid lamps in my house. Wondering whether you have looked at these and whether they would be suitable. I suppose I need to figure out whether Soraa lamps can be used to construct a D65 source.
-Re spectrophotometers, are the Xrite products the best practical product available, or is there a different vendor with a better mousetrap?



Logged

Jim Kasson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2370
    • The Last Word
Re: HR-1 SuperChroma
« Reply #36 on: May 17, 2020, 01:54:48 pm »


-I'm a hobby letterpress printer. Totally get the spot vs process distinction.
-I haven't looked at actual paint samples yet, nor measured, to see how well-behaved their spectra are.
-I can definitely get some test panels painted and measured with a spectrophotometer. I'm a few months away from that point but will get there.
-Babelcolor patchtool is reasonably priced, no problem to get that software.

Questions
-Re lights - I put Soraa Vivid lamps in my house. Wondering whether you have looked at these and whether they would be suitable. I suppose I need to figure out whether Soraa lamps can be used to construct a D65 source.
-Re spectrophotometers, are the Xrite products the best practical product available, or is there a different vendor with a better mousetrap?

I use the X-Rite spectros, and have for about 30 years, although the manufacturer names have changed. I am currently using two i1Pros and an iSis. The newest i1Pro, the 3, doesn't have a lot of software support yet, though.

For illumination of the samples, ideally I'd use the same lighting I expect the user to use to evaluate the product. You don't need a D65 source to calculate the spectrum of the sample illuminated by a D65 light. Take the reflectivity spectrum of the sample and perform a wavelength-by-wavelength multiplication by the spectrum of D65.

It's less accurate, but you can use the Bradford adaptation algorithm to change white points, too.

I don't know how you're going to control the adaptation of the viewer when she is looking at the samples on the web, but adaptation can make a huge difference in color appearance.

Jim

Jim Kasson

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2370
    • The Last Word
Re: HR-1 SuperChroma
« Reply #37 on: May 17, 2020, 01:58:16 pm »

-Re lights - I put Soraa Vivid lamps in my house. Wondering whether you have looked at these and whether they would be suitable. I suppose I need to figure out whether Soraa lamps can be used to construct a D65 source.


If you're going to be taking photographs, I'd recommend a Xenon flash tube from a brand name vendor. And I'd go out on a limb and put Godox into that category.

The Soraa lamps I've seen have a big peak in the blue part of the spectrum.

https://blog.kasson.com/the-last-word/further-adventures-in-observer-metameric-failure/

Jim
« Last Edit: May 17, 2020, 02:01:44 pm by Jim Kasson »
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20701
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: HR-1 SuperChroma
« Reply #38 on: May 17, 2020, 02:14:39 pm »

Indeed, the Soraa LED I measured wasn't anything to write home about and not close to as ideal as the Solux.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

sea-speak

  • Newbie
  • *
  • Offline Offline
  • Posts: 8
Re: HR-1 SuperChroma
« Reply #39 on: May 17, 2020, 03:54:23 pm »

ok, good to know the Soraa lights can be ruled out. Any experience with Waveform? They have a claimed 99CRI D65 LED strip source that I could use to build a lamp.

https://store.waveformlighting.com/collections/led-linear-modules/products/absolute-series-99-cri-led-linear-module?variant=8190565777510

Thanks for the pointer to Solux, and to Godox for bulbs. I will need to take some photos of actual products at some point - won't only be able to use the spectrophotometer directly on paint sample panels. That said - for sample panels - if I were to buy an off the shelf solution rather than build a box, is Just Normlicht clearly better/worse than Pantone?
Logged
Pages: 1 [2] 3   Go Up