Pages: 1 2 [3] 4   Go Down

Author Topic: Pro 1000 Color Management Investigation  (Read 7662 times)

Mark D Segal

  • Contributor
  • Sr. Member
  • *
  • Offline Offline
  • Posts: 12512
    • http://www.markdsegal.com
Re: Pro 1000 Color Management Investigation
« Reply #40 on: February 18, 2019, 09:09:10 am »

Maybe Canon provides more tools than Epson for third party media.  Scott Martin published something useful about a decade ago that I found useful for my Epson 3880.  However, Epson only has some limited choices and then one might have to fine tune things in the print driver in terms of ink lay down and dry times which is difficult and time consuming.  As my 3880 nears the end of life, I'll be faced with the choice of Canon and Epson 17inch printers.  Doug's comments on this thread will be useful information in that regard.

You can customize a range of media settings in both printer lines, but how refined the one is versus the other wouldn't be at the front line of my considerations in selecting a printer model to replace a 3880. The very first thing you need to look at is what the printers can do relative to what you want them to do. For example, if you want to print panos or prints that exceed a 26 inch length limit, if you want to print on inflexible media, then you are buying an Epson P800, not a Canon Pro-1000. If how easily the media feeds into the printer is your highest priority consideration, then you are more likely buying a Canon Pro-1000. Each model has a different way of handling maintenance/cleaning and that may be important to you. Both of them can be profiled to make excellent prints. On the whole, the Pro-1000 has slightly larger gamut volume than the P800; whether this matters to the kind of photos you will be printing is something you would need to determine and that could influence a decision. We can get all bent and twisted over the finest points of profiling and fine-tuning the media types, but in the final analysis what will matter the most to most people are usage factors and how the prints look. Some differences of file rendition just aren't noticeable enough to swing a buying choice.
Logged
Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....."

Alan Goldhammer

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4344
    • A Goldhammer Photography
Re: Pro 1000 Color Management Investigation
« Reply #41 on: February 18, 2019, 03:09:48 pm »

Thanks Mark, very useful comments!
Logged

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Pro 1000 Color Management Investigation
« Reply #42 on: February 20, 2019, 11:08:16 am »

General observations to date:

The Pro 1000 delivers a significantly larger gamut over both the 9500 II and 9800. Especially lower black points and better rendition of darker colors.

The Pro 1000 driver has a better, and less lumpy, mapping of device RGB values than either the 9800 or 9500. Fewer patches are needed to create good profiles. The standard iSis 957 patch, letter size target creates better profiles. Good profiles for the 9800 requires significantly more patches.

However, the 9800, to my surprise, produces slightly more consistent colors. Printing random, in gamut, colors in different locations, the 9800 still produces the smallest differences. A large patch set overcomes the lumpiness of the 9800 driver so, when combined with a patch set optimized for the 9800, the 9800 delivers more precise, in-gamut, colors. However, this effect is not visible (to me anyway) though it is measureable. Says a lot for Epson's design of the rather old 9800.

The Pro 1000 prints appear to have less gloss differential and bronzing though, at this point, I don't have any good way to quantize the differences. This is an important, but highly subjective, effect.

Things I have not yet investigated:
How well the Pro 1000 prints fine textures. Does it do a better job interpolating rapid changes in pixel values? How much does the printer change over time and how much change occurs when not printing for a week or so.

Logged

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Pro 1000 Color Management Investigation
« Reply #43 on: February 22, 2019, 03:33:49 pm »

After further refinement, I have come up with a set of patches that addresses some of the unique aspects of Pro 1000 color mapping. And have attached the CGATs file that can be loaded into I1Profiler. It consists primarily of 1914 patches that include the original default set of 957 iSis patches as well as about 600, I1Profiler "Optimized" patches created from the initial 957 set. Additionally, I added patches concentrated at small RGB values from 0 to 49 in steps of 7 as well as redundant near neutrals. This was because the Pro 1000 exhibits a strong shift in a* and b* in the range of L* from 3 to 20. Otherwise, the Pro 1000 has very smooth responses along the neutral axis and is quite well modeled by the default plus optimized sets. However, the Pro 1000 exhibits somewhat more patch variation depending on location as well as a slight "warm up" phase which mostly affects the L* shifting most of the neutral L* by about .3 L* when printing the first page in a group.

So I also duplicated the patches as two sets of 1914 patches, each set randomized differently.

The result is a 4 page iSis letter size target set attached to this post. One can just edit the patch to the first 1914 but including the full 3828 patches reduces slightly the variations from where individual patches are printed.

Using my CC data set which also includes an L* neutral tone set in increments of L*=1, resulted in the following accuracies:

Using the default set of 957 patches and measuring the printed CC target excluding the white CC patch as it's OOG :  Ave. dE00 for the CC set: 0.54. Ave dE for the neutrals: .71

Using the  set of 3848 patches and measuring the printed CC target excluding the white CC patch as it's OOG :  Ave. dE00 for the CC set: 0.39. Ave dE for the neutrals: .44

Overall, this is similar to what I was able to achieve with the 9800, but there the 3828 patches were unique, not duplicated and randomized 1914 sets. The 9800 is a bit more consistent printing specific colors at different locations but it is also more "lumpy" and benefits more from a larger set of patches clustered around where the "lumps" are greatest.

One note: the neutrals show a somewhat higher dE00 because dE00 is much more sensitive to variations in a* and b* than even slightly saturated colors. This is why much of my efforts have been improving the neutral color's accuracy.

I don't know how applicable this set would be to the Pro 2000/4000 but since the inks are the same it might be of some use with these.

EditToAdd:
I just compared the mapping of Pro1000 and Pro2000 profiles for the same Canon Paper (210 GSM Photog rag) they are quite different. The Pro1000 exhibits the rapid change in a*b* at low L* that the patch set is optimized for, while the Pro2000 has quite a different response. So I do not recommend this patch set for anything other than Pro1000 printers. Odd that it would have a significantly different mapping.
« Last Edit: February 22, 2019, 11:37:31 pm by Doug Gray »
Logged

Panagiotis

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 441
    • Fine Art Print
Re: Pro 1000 Color Management Investigation
« Reply #44 on: February 23, 2019, 06:43:35 am »

After further refinement, I have come up with a set of patches that addresses some of the unique aspects of Pro 1000 color mapping. And have attached the CGATs file that can be loaded into I1Profiler. It consists primarily of 1914 patches that include the original default set of 957 iSis patches as well as about 600, I1Profiler "Optimized" patches created from the initial 957 set. Additionally, I added patches concentrated at small RGB values from 0 to 49 in steps of 7 as well as redundant near neutrals. This was because the Pro 1000 exhibits a strong shift in a* and b* in the range of L* from 3 to 20. Otherwise, the Pro 1000 has very smooth responses along the neutral axis and is quite well modeled by the default plus optimized sets. However, the Pro 1000 exhibits somewhat more patch variation depending on location as well as a slight "warm up" phase which mostly affects the L* shifting most of the neutral L* by about .3 L* when printing the first page in a group.

So I also duplicated the patches as two sets of 1914 patches, each set randomized differently.

The result is a 4 page iSis letter size target set attached to this post. One can just edit the patch to the first 1914 but including the full 3828 patches reduces slightly the variations from where individual patches are printed.

Using my CC data set which also includes an L* neutral tone set in increments of L*=1, resulted in the following accuracies:

Using the default set of 957 patches and measuring the printed CC target excluding the white CC patch as it's OOG :  Ave. dE00 for the CC set: 0.54. Ave dE for the neutrals: .71

Using the  set of 3848 patches and measuring the printed CC target excluding the white CC patch as it's OOG :  Ave. dE00 for the CC set: 0.39. Ave dE for the neutrals: .44

Overall, this is similar to what I was able to achieve with the 9800, but there the 3828 patches were unique, not duplicated and randomized 1914 sets. The 9800 is a bit more consistent printing specific colors at different locations but it is also more "lumpy" and benefits more from a larger set of patches clustered around where the "lumps" are greatest.

One note: the neutrals show a somewhat higher dE00 because dE00 is much more sensitive to variations in a* and b* than even slightly saturated colors. This is why much of my efforts have been improving the neutral color's accuracy.

I don't know how applicable this set would be to the Pro 2000/4000 but since the inks are the same it might be of some use with these.

EditToAdd:
I just compared the mapping of Pro1000 and Pro2000 profiles for the same Canon Paper (210 GSM Photog rag) they are quite different. The Pro1000 exhibits the rapid change in a*b* at low L* that the patch set is optimized for, while the Pro2000 has quite a different response. So I do not recommend this patch set for anything other than Pro1000 printers. Odd that it would have a significantly different mapping.

Thank you. I am looking forward to use your patch set when I will have an ι1προ2.
Logged

howardm

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1984
Re: Pro 1000 Color Management Investigation
« Reply #45 on: February 23, 2019, 10:01:07 am »

You're going to have a great afternoon of manually scanning all those patches ;)

The file has it's colors in RGB space.  I was under the impression that i1Profiler *really* wants
Lab and will not accept RGB values in CGAT or .cxf

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Pro 1000 Color Management Investigation
« Reply #46 on: February 23, 2019, 11:00:38 am »

You're going to have a great afternoon of manually scanning all those patches ;)

The file has it's colors in RGB space.  I was under the impression that i1Profiler *really* wants
Lab and will not accept RGB values in CGAT or .cxf

I1Profiler requires RGB (device space) values when printing through an RGB driver. The initial patch set is used by I1Profiler on the first pass to create a profile. Once that is done I1Profiler can now map colors from Lab to close RGB triplets. It can add additional nearby colors to improve the interpolation accuracy. Added spot colors and those automatically generated are all converted to RGB and printed in device space.

Once a set of RGB values that is optimized around a  printer and paper type, that complete set of RGB values can be printed, scanned, and a profile created without the need to repeat the two step process. My basic two sets are for matte and glossy types optimized for the printer.

By saving the initial and optimal RGB values at each step as CGATs files, one can concatenate them to create a single patch set for future profiling. I have scripts in Matlab that automate most all of this.
Logged

Panagiotis

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 441
    • Fine Art Print
Re: Pro 1000 Color Management Investigation
« Reply #47 on: February 24, 2019, 04:10:47 am »

You're going to have a great afternoon of manually scanning all those patches ;)

Is it difficult to scan 6 A4 pages with i1pro2?
Logged

vikcious

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 79
Re: Pro 1000 Color Management Investigation
« Reply #48 on: February 24, 2019, 04:21:49 am »

Is it difficult to scan 6 A4 pages with i1pro2?

Nope. Not difficult but incredibly boring and time consuming... I would probably fall asleep three times in doing that!
Keep in mind that you'd be scanning 186 lines, of 21 patches per line... TWICE! with an average of 8-10 seconds per line  :o
If only I had that patience!
Logged
|| One screw-up is a mistake, two screw-ups is a trend, three screw-ups is a style ||

Panagiotis

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 441
    • Fine Art Print
Re: Pro 1000 Color Management Investigation
« Reply #49 on: February 24, 2019, 04:29:28 am »

A question for Doug.
Do you believe that the media type plays a role in the quality of the profile or the gamut volume?
I ask this because I have the impression that there are different ink mixing recipes in some media settings.
For MK papers, the Fine Art Smooth media type seems to me visually that produces darker blacks that the Highest Density Fine Art paper setting. (I printed samples with the BW driver mode)
Or there are MK paper types that use both blacks PBK/MBK.
Thanks!
Logged

Panagiotis

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 441
    • Fine Art Print
Re: Pro 1000 Color Management Investigation
« Reply #50 on: February 24, 2019, 04:42:01 am »

Nope. Not difficult but incredibly boring and time consuming... I would probably fall asleep three times in doing that!
Keep in mind that you'd be scanning 186 lines, of 21 patches per line... TWICE! with an average of 8-10 seconds per line  :o
If only I had that patience!

Hm.. this is (186*10)/60 = at least 30 minutes. Difficult to do it multiple times.
Logged

howardm

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1984
Re: Pro 1000 Color Management Investigation
« Reply #51 on: February 24, 2019, 08:17:24 am »

And you're not including any time for setting the page up or G*D FORBID that i1P croaks or throws errors during the scanning and you.have.to.start.over.again :(

Alan Goldhammer

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4344
    • A Goldhammer Photography
Re: Pro 1000 Color Management Investigation
« Reply #52 on: February 24, 2019, 08:29:33 am »

Nope. Not difficult but incredibly boring and time consuming... I would probably fall asleep three times in doing that!
Keep in mind that you'd be scanning 186 lines, of 21 patches per line... TWICE! with an average of 8-10 seconds per line  :o
If only I had that patience!
My Argyll profiles require four letter size pages.  I think my scans per line are about 1/2 the time mentioned above. 
Logged

Mark D Segal

  • Contributor
  • Sr. Member
  • *
  • Offline Offline
  • Posts: 12512
    • http://www.markdsegal.com
Re: Pro 1000 Color Management Investigation
« Reply #53 on: February 24, 2019, 08:50:48 am »

My usual i1Profiler 2371 patch set requires 4 letter-size pages and usually takes about 12 minutes to scan all of it with an i1Pro2. This is doing a two-pass scan per row in dual scan strip mode. I make sure not to scan too quickly in order to assure the reading of an adequate number of samples per patch. If there is an error in a row the software notifies it immediately and one simply rescans the row.
Logged
Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....."

Panagiotis

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 441
    • Fine Art Print
Re: Pro 1000 Color Management Investigation
« Reply #54 on: February 24, 2019, 09:54:07 am »

Thanks Alan and Mark. Sounds easier now.
Logged

Mark D Segal

  • Contributor
  • Sr. Member
  • *
  • Offline Offline
  • Posts: 12512
    • http://www.markdsegal.com
Re: Pro 1000 Color Management Investigation
« Reply #55 on: February 24, 2019, 10:04:10 am »

A question for Doug.
Do you believe that the media type plays a role in the quality of the profile or the gamut volume?
I ask this because I have the impression that there are different ink mixing recipes in some media settings.
For MK papers, the Fine Art Smooth media type seems to me visually that produces darker blacks that the Highest Density Fine Art paper setting. (I printed samples with the BW driver mode)
Or there are MK paper types that use both blacks PBK/MBK.
Thanks!

While you asked Doug, I have some understanding of this I could share here. The Media Type choice does influence the density of ink laydown and whether the Black is to be PK or MK. It also affects other media settings in the printer driver. Profile quality and gamut volume should be driven more importantly by other factors. Insofar as gamut volume is related to Black density, one could say that the Media Type influences it, insofar as the combination of matte surfaces and MK ink has grayer maximum black and less reflectivity the gamut volume would be smaller than for luster or gloss papers using PK ink. But the primary influence on gamut volume is the kind of paper and ink. If by profile quality you mean the accuracy of colour rendition relative to the image file values and smoothness of tonal gradation, I don't believe the choice of Media Type in and of itself is a factor.
Logged
Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....."

Panagiotis

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 441
    • Fine Art Print
Re: Pro 1000 Color Management Investigation
« Reply #56 on: February 24, 2019, 10:15:01 am »

While you asked Doug, I have some understanding of this I could share here. The Media Type choice does influence the density of ink laydown and whether the Black is to be PK or MK. It also affects other media settings in the printer driver. Profile quality and gamut volume should be driven more importantly by other factors. Insofar as gamut volume is related to Black density, one could say that the Media Type influences it, insofar as the combination of matte surfaces and MK ink has grayer maximum black and less reflectivity the gamut volume would be smaller than for luster or gloss papers using PK ink. But the primary influence on gamut volume is the kind of paper and ink. If by profile quality you mean the accuracy of colour rendition relative to the image file values and smoothness of tonal gradation, I don't believe the choice of Media Type in and of itself is a factor.

Thanks Mark.
I asked because there are so many choices in the Canon driver. If someone wants to build a custom media type for a PK or MK paper it is difficult to choose where to start from. And the choices of the major paper brands (Hahnemuhle, Canson) are confusing. For the same paper they choose for Pro-1000 different media type than the Pro-4000. I guess extensive experimentation is needed.   
Logged

Alan Goldhammer

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4344
    • A Goldhammer Photography
Re: Pro 1000 Color Management Investigation
« Reply #57 on: February 24, 2019, 10:32:02 am »

Thanks Mark.
I asked because there are so many choices in the Canon driver. If someone wants to build a custom media type for a PK or MK paper it is difficult to choose where to start from. And the choices of the major paper brands (Hahnemuhle, Canson) are confusing. For the same paper they choose for Pro-1000 different media type than the Pro-4000. I guess extensive experimentation is needed.
Go back to my earlier POST on this thread.  It links to a good test chart from Scott Martin that can help you choose what media setting to use.
Logged

Panagiotis

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 441
    • Fine Art Print
Re: Pro 1000 Color Management Investigation
« Reply #58 on: February 24, 2019, 10:40:28 am »

Go back to my earlier POST on this thread.  It links to a good test chart from Scott Martin that can help you choose what media setting to use.

I show that. Thanks! I believe this is mainly a chart for choosing the ink load. To check if there is any bleeding.
The thing is that for every base media type (and there are many) you can set the ink load in 5 steps, so the original question remains: What other differences (if any) are there?
Logged

dehnhaide

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 87
Re: Pro 1000 Color Management Investigation
« Reply #59 on: February 24, 2019, 11:15:40 am »

Thanks Mark.
I asked because there are so many choices in the Canon driver. If someone wants to build a custom media type for a PK or MK paper it is difficult to choose where to start from. And the choices of the major paper brands (Hahnemuhle, Canson) are confusing. For the same paper they choose for Pro-1000 different media type than the Pro-4000. I guess extensive experimentation is needed.
+1 I share your confusion.


Sent from my iPhone using Tapatalk
Logged
Pages: 1 2 [3] 4   Go Up