Individual icc vs generic icc on Canon pro 4600

[HCHeyerdahl]

I am taking delivery of a new Canon 4600 next week to replace my HP Z 3200.

If I understand correctly, the Canon printer is calibrated to a Canon referance standard apon instalation and then I should do that every 6 months or so.

Presumably then, every similar Canon printer is "identical" with respect to using icc profiles as long as they are created on a calibrated printer. Ideally then, I can simply rely on the icc profiles provided by Canson etc.

Well, what is your experience? Is the calibratiion on new Canon printers good enough to use icc profiles provided by paper producers?

 

[digitaldog]

Is the calibratiion on new Canon printers good enough to use icc profiles provided by paper producers?

Calibration places a device into a consistent state, profiling is different, and:
http://digitaldog.net/files/Not_All_Profiles_are_created_equally.mp4

[HCHeyerdahl]

Thank you,

that was a useful video!

I understand the difference between calibration and profiling.

After looking at the video, I realise that my question is really: How similar are the new Canon printers when calibrated? If one takes data created from a sample of Canon pro printers using the same software, same spectro etc, how different/similar will the data be? 

From the video, I understand that even if the data is identical, the resulting icc will still depend on how the data is processed/ converted by the paper producer and thus depends on the software used etc. and to some extent even the personal preferences of the profile creator.

[digitaldog]

After looking at the video, I realise that my question is really: How similar are the new Canon printers when calibrated? If one takes data created from a sample of Canon pro printers using the same software, same spectro etc, how different/similar will the data be? 

Short answer: It depends.
Longer answer: You have to print the same target over time with multiple printers if that's what you want to evaluate, measure them, and produce a dE (deltaE report).
When I built the profiles for Epson's Exhibition Fiber more than a decade ago, we used half a dozen of the same printers across the USA over a span of months, and the dE was less than 0.5 for all the units. IOW, multiple Epsons of the same model over time are very similar and consistent. So was the resulting profile for that paper and model. Paper consistency would also play a role.

[Panagiotis]

I am taking delivery of a new Canon 4600 next week to replace my HP Z 3200.

If I understand correctly, the Canon printer is calibrated to a Canon referance standard apon instalation and then I should do that every 6 months or so.

Presumably then, every similar Canon printer is "identical" with respect to using icc profiles as long as they are created on a calibrated printer. Ideally then, I can simply rely on the icc profiles provided by Canson etc.

Yes but did they (Canson) used a calibrated printer?

Well, what is your experience? Is the calibratiion on new Canon printers good enough to use icc profiles provided by paper producers?

I don't know how good is the color calibration (I mean if it can actually "equalize" different machines) on Canon printers but I know that the combination of canned icc profiles + media types chosen by the paper manufacturers are not ideal. You will leave print quality on the table. Especially with Hahnemuhle.

[Czornyj]

I don't know how good is the color calibration (I mean if it can actually "equalize" different machines) on Canon printers but I know that the combination of canned icc profiles + media types chosen by the paper manufacturers are not ideal. You will leave print quality on the table. Especially with Hahnemuhle.

Canon color calibration is as good as it gets, and it equalizes printers perfectly. The „aesthetics” of perceptual rendering used by 3rd party profile makers and custom profiling software solutions may vary

[HCHeyerdahl]

Thank you all, this is very useful. I learned a lot from your comments and the video! it took some time to digest it all.

I never realised that there are so many variables to creating an icc profile. I was under the delusion that there is one ideal target and that the quality of the profile mostly depended on the equipment used to read the colour patches.

The question then becomes, how involved do I want to get in creating profiles?

For almost 18 years I have been perfectly happy with the profiles generated by the HP Z 3200. This is basically a one trick pony with few options to fine tune the result when converting the date to the icc file. It seems really to just create a HP profile specific to the printer and paper combo.

Now, if the Canon is so well calibrated/linearised, there seems to be less use for custom profiles for a specific printer. Any profile created for these printers should generate the same prints on all similar printers. The choice between profiles then seems to become a matter of taste, or skills and preferences of the profile creator - if one may assume paper manufacturers use the same equipment as other custom profile makers. I guess I can start and see if I like the prints I get with profiles from Canson and Ilford.

I feel a slight temptation to jump down the rabbit hole and start try to create my own profiles and learn how to tweek them. From previous post I have seen how the Z can be used to create files with spectral data from colour patches printed on other printers, and the spectro on my old Z still works. The challenge then is really only converting the data into an icc profile. I have had a look at the documentation for Argyll, and it seems there is quite a bit I need to learn and understand to be able to use this in a way that consistently creates better profiles than I will get from paper manufacturers. I would appreciate suggestions for easier or more intuitive software that can convert the data created by the Z into an icc file.

[gchappel]

I have been printing with the ipf8400 and then pro 4100 for the last 12yrs or so.   
I found most of the profiles provided by the companies for their papers work well.   
I primarily shoot wildlife and sports, so absolute color accuracy is not something I need.   It needs to look like I remember it, it needs to look real.
With that in mind, the supplied profiles worked fine.   
I had several papers that I could not find profiles for, so I had profiles made for the ipf8400.     When I moved to the 4100, those same profiles were ok, they worked fine.   I thought they were off a little, but no one else could tell, so I kept using them.   I am not really even sure which are more "accurate", they did look a little different side by side.   
I did not want to go down the $2000 rabbit hole to profile a couple of papers. 
The profiles I tried made with a color munki were not great.   
I just started playing with Mike Chaney's Prism.   I use qimage for printing.   Prism is a $50 program that builds profiles using a epson 8550 scanner- which I already had.    I am still in the review process, but so far so good.   The profiles are good on almost all the papers I have played with.
Just another option.
gary

[Kyle Jackson]

Now, if the Canon is so well calibrated/linearised, there seems to be less use for custom profiles for a specific printer.

As mentioned above, be careful not to mix up calibration (or linearization) with profiling. Calibration simply puts the printer into a known state for that paper / print settings. Ideally those results are repeatable within a narrow band but actual results probably vary. Calibration can depend on the paper, the print settings used (in HP land that's the "paper preset", plus any subsequent changes to the paper settings made by the user, such as ink load), changes in humidity, changes in printheads / health, etc. For these reasons periodic recalibration is required (HP prompts for it), and is mandatory before attempting any profiling.

Profiling on the other hand builds "translation" tables so your system can "speak colour" with your printer, so results are ideally consistent and predictable (soft-proofing). In theory the profile is good "forever", as long as you keep the printer in calibrated state. (In practice I reprofiled pretty much every time I recalibrated because HP makes it stupid easy to do.)

I don't think there's anything revolutionarily "new" here in the Canon 4600's calibration that you didn't already have with HP. Maybe it can hold a bit tighter tolerance, but the marketing claim of "same results on any two printers anywhere" was also HP's almost 20 yrs before (tho ofc HP could also *profile* itself), so I think none of the considerations for profiling are really any different because of that. If you weren't already making your own profiles on your HP and you were happy with mfr ones, then it's probably the same with the Canon.

The choice to use a canned mfr profile vs your own is a personal one. The variables above are some of the reasons why a mfr profile might not always be the "best" fit for your situation (but "best" is a personal judgment too). And if you want to adjust print settings such as ink load then you pretty much "need" your own profile (unless again the differences don't bother you). But you can also just use a mfr profile and if you're happy with the results then just move on in life. Again HP made profiling so easy that there was no reason not to make our own (and when I compared results, I found some of HP's canned profiles were poor or at least very conservative).

As for using your HP's built-in spectro to profile the Canon, yes you can do that and I think one of the many PDFs/manuals from HP's site explains how. If you have an HP Z3200ps then I recall you choose a profile target image in the HP Colour Centre, then print it on your Canon, and then use HP Colour Centre again to scan it with the HP. You then get an output file that you need to take into another (non-HP) software to create the profile for your Canon, whether that's Argyll or whatever. The hows and whats of that depends on the specific software.

Edit: as someone here mentioned recently, there's also some "overhead" costs/effort/space associated with keeping an HP Z-series operating solely for use as an occasional profiling scanner.

[Czornyj]

And if you want to adjust print settings such as ink load then you pretty much "need" your own profile (unless again the differences don't bother you).

Adjusting print settings, such as ink load, is also calibration, not profiling. Profiling is usually only a characterisation, done at the end of the calibration process. And you can also adjust the ink load in Canon without reprofiling.

There's nothing revolutionary about Canon calibration; it's just a different approach for a different purpose. A built-in spectrophotometer was primarily intended for contract proofing in the high-volume printing industry to make profiling and validation faster and more foolproof. 

The real difference is that Canon LUCIA Pro II ink has higher pigmentation and is more forgiving, so the results vary less on 3rd party papers, even the shittiest out there. Custom profiles mostly influence the nuances in perceptual rendering.

[HCHeyerdahl]

Canon color calibration is as good as it gets, and it equalizes printers perfectly. The „aesthetics” of perceptual rendering used by 3rd party profile makers and custom profiling software solutions may vary

The real difference is that Canon LUCIA Pro II ink has higher pigmentation and is more forgiving, so the results vary less on 3rd party papers, even the shittiest out there. Custom profiles mostly influence the nuances in perceptual rendering.

Just to make sure I understand correctly: When you write "The „aesthetics” of perceptual rendering" and "nuances in perceptual rendering" you are referring to how the choices made during the conversion of spectral data (from reading the colour patches) to an icc file, will determine the look of the print? (IOW, You are not talking about choosing a rendering intent in say Lightroom before you print).

What do you mean by  "The real difference is that Canon LUCIA Pro II ink has higher pigmentation and is more forgiving, so the results vary less on 3rd party papers"?

[Kyle Jackson]

Adjusting print settings, such as ink load, is also calibration, not profiling. Profiling is usually only a characterisation, done at the end of the calibration process. And you can also adjust the ink load in Canon without reprofiling.

Yes, with the HPs if I create a new ink load, a new calibration is ofc mandatory, but then a new profile is also needed, that's what I meant.

So now I'm confused how (or why) someone would change the ink load on the Canon (or any printer) and not make a new profile for that?

Edit:

A built-in spectrophotometer was primarily intended for contract proofing in the high-volume printing industry to make profiling and validation faster and more foolproof.

But that built-in spectro can't actually be used for profiling, right? Or is there some add-on option that "unlocks" that feature? I've not found one.

[Kyle Jackson]

Just to make sure I understand correctly: When you write "The „aesthetics” of perceptual rendering" and "nuances in perceptual rendering" you are referring to how the choices made during the conversion of spectral data (from reading the colour patches) to an icc file, will determine the look of the print? (IOW, You are not talking about choosing a rendering intent in say Lightroom before you print).

What do you mean by  "The real difference is that Canon LUCIA Pro II ink has higher pigmentation and is more forgiving, so the results vary less on 3rd party papers"?

If you use Perceptual RI to print, it's a "secret sauce" conversion (iirc there's no "standard"), so profiles made in different software from the exact same target print might give different print results.

But practically speaking I don't see how you'd ever encounter this if you're simply using the profile provided by the mfr, and not comparing with other profiles (3rd party or your own).

And ofc if you don't use Perceptual RI then it's moot (I don't).

[Czornyj]

Just to make sure I understand correctly: When you write "The „aesthetics” of perceptual rendering" and "nuances in perceptual rendering" you are referring to how the choices made during the conversion of spectral data (from reading the colour patches) to an icc file, will determine the look of the print? (IOW, You are not talking about choosing a rendering intent in say Lightroom before you print).

What do you mean by  "The real difference is that Canon LUCIA Pro II ink has higher pigmentation and is more forgiving, so the results vary less on 3rd party papers"?

The way the perceptual rendering intent B2A table is created varies across profiling engines and can be further tweaked by the user.

Higher pigmentation means you need to put less ink to achieve the same result, and it also soaks less into the substrate - so it dries faster, it's less prone to over-inking issues and the paper characteristics have less influence on results.

[Czornyj]

Yes, with the HPs if I create a new ink load, a new calibration is ofc mandatory, but then a new profile is also needed, that's what I meant.

So now I'm confused how (or why) someone would change the ink load on the Canon (or any printer) and not make a new profile for that?

Edit:
But that built-in spectro can't actually be used for profiling, right? Or is there some add-on option that "unlocks" that feature? I've not found one.

You can change 5 levels of ink usage amount in Media Configuration Tool, and it doesn't change the overall appearance of print, so there's no need for reprofiling (I guess there are 5 separate 3DLUT's for each ink load level)

Canon has only built-in spectrodensitometer (densitometer with RGB LED emitters), so of course it can't be used for profiling, which requires a spectrophotometer.

[Kyle Jackson]

You can change 5 levels of ink usage amount in Media Configuration Tool, and it doesn't change the overall appearance of print, so there's no need for reprofiling (I guess there are 5 separate 3DLUT's for each ink load level)

Canon has only built-in spectrodensitometer (densitometer with RGB LED emitters), so of course it can't be used for profiling, which requires a spectrophotometer.

Ah ok, good detail to know about that Canon's "spectro".

Ok so if changing the Canon's ink level doesn't change the overall print appearance, then what purpose is it used for? Is it just some kind of economy vs durability slider?

Asking because when I'm profiling papers (esp matte) on my HPs I sometimes increase the ink level to expand the gamut volume and/or black point (Dmax). The HPs permit a very wide range of ink levels. Sometimes there's a significant improvement to both without adverse effects (well aside from ink consumption lol). So it can definitely change the print appearance (assuming ofc that it's a source image that can actually take advantage of it).

(Aside: I found this was actually necessary when trying to achieve results on the Z3200 comparable to Z3100. I gather the Z32's firmware is tuned differently and I found I had to push ink levels higher on Z32 to achieve comparable results to Z31 "out-of-the-box".)

[Czornyj]

Ah ok, good detail to know about that Canon's "spectro".

Ok so if changing the Canon's ink level doesn't change the overall print appearance, then what purpose is it used for? Is it just some kind of economy vs durability slider?

Asking because when I'm profiling papers (esp matte) on my HPs I sometimes increase the ink level to expand the gamut volume and/or black point (Dmax). The HPs permit a very wide range of ink levels. Sometimes there's a significant improvement to both without adverse effects (well aside from ink consumption lol). So it can definitely change the print appearance (assuming ofc that it's a source image that can actually take advantage of it).

(Aside: I found this was actually necessary when trying to achieve results on the Z3200 comparable to Z3100. I gather the Z32's firmware is tuned differently and I found I had to push ink levels higher on Z32 to achieve comparable results to Z31 "out-of-the-box".)

When you decrease the ink amount, it may reduce ink consumption at the expense of slightly worse dmax. It's also to prevent potential overinking issues, but - as I already had mentioned - it uses a relatively small amount of ink from the start, so overinking issues virtually never occur since introduction of LUCIA PRO and PRO II.  Increasing the ink amount in most cases doesn't improve dmax as Canon did the homework, so the standard amount is usually optimal, and it doesn't really need much ink to reach its highest dmax

[Kyle Jackson]

When you decrease the ink amount, it may reduce ink consumption at the expense of slightly worse dmax. It's also to prevent potential overinking issues, but - as I already had mentioned - it uses a relatively small amount of ink from the start, so overinking issues virtually never occur since introduction of LUCIA PRO and PRO II.  Increasing the ink amount in most cases doesn't improve dmax as Canon did the homework, so the standard amount is usually optimal, and it doesn't really need much ink to reach its highest dmax

Ah ok, thanks for that info about Canon's new inks. It sounds like ink level adjustment on the Canon doesn't have much use then, so most users can just ignore it. It sounds quite different from the old HPs, where the paper presets have ink levels ranging from 21 to 60 picolitres per dot (at 600 dpi) and the user can adjust those by an additional +/-20%. I generally just ignore ink level too, it was only when I discovered Z32 output was "weak" compared to Z31 that I tried ink level and it made a useful difference (and as a result I was kinda disappointed in the Z32's firmware tuning).


Kinda strayed off OP's topic, sorry @HCHeyerdahl ... But it sounds to me like it's still a personal choice, and if the results with a canned mfr profile are acceptable for needs, then go for it. The only way to really "prove" otherwise would be to make custom profiles and compare the results, and show that the custom profile is somehow "better" than the mfr one. That's at least partially subjective tho, and might not always be true for every combo of paper and source image. Like on one project it might seem useful and the other 99% it doesn't. So maybe we just use the canned mfr profiles and be blissfully unaware what is "missing" (if anything).

I'm def taking notes from this because I don't know how much longer I'll play with the HPs (my Z31's formatter drive literally died during this thread...), and the "loss" of a built-in spectro for profiling was kinda worrying for me if I eventually move to the Canon 2600. Nothing causes anxiety like taking something away lol...

[HCHeyerdahl]

I have been printing with the ipf8400 and then pro 4100 for the last 12yrs or so.   
I found most of the profiles provided by the companies for their papers work well.   
I primarily shoot wildlife and sports, so absolute color accuracy is not something I need.   It needs to look like I remember it, it needs to look real.
With that in mind, the supplied profiles worked fine.   

I think that is a good point. I realise I mostly do not require accurate colours. This of course seems to suggests that I too will be quite happy with the canned profiles.

I did not want to go down the $2000 rabbit hole to profile a couple of papers. 
The profiles I tried made with a color munki were not great.   
I just started playing with Mike Chaney's Prism.   I use qimage for printing.   Prism is a $50 program that builds profiles using a epson 8550 scanner- which I already had.    I am still in the review process, but so far so good.   The profiles are good on almost all the papers I have played with.
Just another option.
gary

That resembles my experience with trying the Color Munki, and besides reading all the patches is not my idea of having a good time :-)

I have also purchased Print Prism. I was sceptical since previous attempts at using scanners to create profiles left a lot to be desired. However, after reading a thread on DP I decided to purchase it since I already have the Epson V 39 II. I have not had time to start playing with it yet. Very interesting to hear about your results. It seems to mirror the thread on DP: https://www.dpreview.com/forums/threads/print-prism-first-impressions.4826932/

Kinda strayed off OP's topic, sorry @HCHeyerdahl ... But it sounds to me like it's still a personal choice, and if the results with a canned mfr profile are acceptable for needs, then go for it. The only way to really "prove" otherwise would be to make custom profiles and compare the results, and show that the custom profile is somehow "better" than the mfr one. That's at least partially subjective tho, and might not always be true for every combo of paper and source image. Like on one project it might seem useful and the other 99% it doesn't. So maybe we just use the canned mfr profiles and be blissfully unaware what is "missing" (if anything).

I'm def taking notes from this because I don't know how much longer I'll play with the HPs (my Z31's formatter drive literally died during this thread...), and the "loss" of a built-in spectro for profiling was kinda worrying for me if I eventually move to the Canon 2600. Nothing causes anxiety like taking something away lol...

No problem, it has all been quite relevant. And yes, maybe we should just use the canned profiles an be blissfully unaware of whatever might be missing. I am certainly going to try that first!

Thank you all, this has been very helpful!