Pages: [1]   Go Down

Author Topic: If you print profiling targets for a Canon Pro1000 Read this caution  (Read 716 times)

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197

I have recently been working on a program that can compare multiple profile patch sets for accuracy. It works by combining the patches one wishes to compare along with an independent set of patches to allow evaluation of profile accuracy against the full gamut, a low saturation subset, and 52 device near neutrals. Additionally, it puts a common set of 10, repeating colors, 4 neutrals, 3 light CYM colors, and 3 heavier CYM colors along the right edge of each sheet for easy consistency testing with a spectro. It also prints both a top bar and bottom bar in i1isis format so that the pages can be read normally and in reverse. This is useful for detecting the fairly rate, but annoying, times when the charts get a bit of lint. A re-read fixes it.

The Pro1000 Problem

The Pro1000 tends to print the first page of a group with heavier inking. L* on dark grays can often be 1 L* or more lower on the first page. Similar effects occur on the color patches,

This can result in significant degradation of profile accuracy. So much so that a 1000 patch set will usually have worse accuracy than a 500 patch set where the first page is printed twice and the initial one tossed.

Getting the best profile patch prints should be done this way:

1. Use the print spooler. Do not print individual pages one at a time. Print them as an aggregate using the print spooler.

2. Print the first page twice, the throw the first one out.

3. Do not let each page stay in the tray. As soon as a page is ejected in the tray remove it and set it aside to dry. The reason is that when paper sits on top of another one that isn't completely dry, the ink dries slower. Turns out this is more important than whether you let the paper dry 2 hours or 24 hours. It's the drying in the first few minutes after printing that matters. Don't let it sit in the hopper.

Drying time isn't the problem on a Pro1000. It's these other issues. Especially the first page effect.

Here's the results comparing 12hour v 30 min. drying as well as the first page effect. The latter is overwhelmingly dominant.

Drying time: half day

Using initial "first page":
Device Neutrals  Low Saturation    Full Gamut
 dE76  dE2k       dE76  dE2k       dE76  dE2k
 0.61  0.53         0.87  0.68       1.07  0.64       581_patches.icm
 0.56  0.52         0.76  0.60       0.92  0.56       957_patches.icm

Skip first and using second "first page":
Device Neutrals  Low Saturation    Full Gamut
 dE76  dE2k       dE76  dE2k       dE76  dE2k
 0.36  0.35         0.72  0.56       0.90  0.50       581_patches.icm
 0.48  0.47         0.62  0.49       0.76  0.43       957_patches.icm


Now we look at the effect of drying time. Making profiles from
patches read after drying only 30 mins., here's how well these profiles read
the patches dried for half a day:

Device Neutrals  Low Saturation    Full Gamut
 dE76  dE2k       dE76  dE2k       dE76  dE2k
 0.37  0.35         0.73  0.58       0.91  0.51       581_patches.icm
 0.46  0.45         0.63  0.50       0.76  0.43       957_patches.icm

Even after only 30 minutes drying there is almost no difference!
Logged

rasworth

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 473
Re: If you print profiling targets for a Canon Pro1000 Read this caution
« Reply #1 on: October 16, 2021, 11:27:55 am »

Good detective work, interesting information.  This seems to me to be a printer defect, have you reported to Canon?

Richard Southworth
Logged

arobinson7547

  • Full Member
  • ***
  • Offline Offline
  • Posts: 187
Re: If you print profiling targets for a Canon Pro1000 Read this caution
« Reply #2 on: October 16, 2021, 01:47:22 pm »

To the extent that you can, I woud ALWAYS try to make Profiling Targets, all part of the same print job; preferably on the same sheet (2-up on 12x18) x-up across a 44" roll.

In the case you describe, once that Printer Stops and sort of 'reset' the starts again, that's job to job; page to page variability. I'm sure that 'can' exist, in in printing system.


But I would still isolate 'Targets for Profiling', is if it were one image printing all at the same time, the same job, spooled as one; once.
Logged

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: If you print profiling targets for a Canon Pro1000 Read this caution
« Reply #3 on: October 16, 2021, 04:59:41 pm »

To the extent that you can, I woud ALWAYS try to make Profiling Targets, all part of the same print job; preferably on the same sheet (2-up on 12x18) x-up across a 44" roll.
That's what I do on my 9800 when profiling rolls. The 9800 doesn't have the "first page" problem though but it does have a different one that shows up when it's been a few weeks since I've used it. It's a smaller effect than on the Pro1000 but there can be gradual changes over multiple pages. Best results I've had with the 9800 when not used for a while is to print 5 pages of US  letter size random patches first. After that it's quite consistent run to run even with a few days in between. And no first page effect. A larger effect with the 9800 was sensitivity to vacuum level. I made a special set of patches for that so I could set the vacuum to produce the smallest variation across sheets.

Quote
But I would still isolate 'Targets for Profiling', is if it were one image printing all at the same time, the same job, spooled as one; once.

That works well for the Pro1000. All the patch charts including a duplicate of the first page which then gets tossed.

Worth noting that these differences are relatively small. One really won't see them with normal printing. I'm a bit hypercritical because one of the things I was doing was to make a process for creating highly consistent and accurate special targets for calibrating/testing production cameras. This required a high degree of repeatability when comparing different profiling patch sets.
Logged

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: If you print profiling targets for a Canon Pro1000 Read this caution
« Reply #4 on: October 16, 2021, 05:14:11 pm »

Good detective work, interesting information.  This seems to me to be a printer defect, have you reported to Canon?

Richard Southworth

I don't believe it's a defect but just a characteristic of the print head technology. The 9500 II had even larger effects. But these issues aren't really visible. They do, however, exaggerate the improvements seen with larger patch sizes since all the pages after the first are more self consistent. Larger patch sizes have better average dEs as the first page anomaly becomes less of a factor when more target pages are printed.

I've been working on code that automates all this so people can test this on their own printers. Eventually I will post it to github along with a Windows executable. For now here's an example of outputs of two printed pages showing the first page effect. The r.txt file is the result of averaging fwd/rev scans and extracting two embedded profile measurement files as well as an independent set of random patches for evaluating profile accuracy. The c.txt file is the evaluation of profiles made from those extracted measurement files against the random set.
Logged

arobinson7547

  • Full Member
  • ***
  • Offline Offline
  • Posts: 187
Re: If you print profiling targets for a Canon Pro1000 Read this caution
« Reply #5 on: October 17, 2021, 09:19:55 am »



I'm a bit hypercritical because one of the things I was doing was to make a process for creating highly consistent and accurate special targets for calibrating/testing production cameras. This required a high degree of repeatability when comparing different profiling patch sets.

To the end I follow Don Hutcheson and no matter the target, for Camera/Scanner Profiling, I ALWAYS make a custom reference file (read the target with a Specto). I've done it for the Hutch Color Target (Reflective), as well as, my Colorcheckers

This way you can me less 'hypercritical' about the age of your Targets; just make a fresh reference file.

Not to mention, I never really trusted those BATCH reference files, I use to spend so much time hunting down and downloading; just make my own; and since I'm using the same Spector for Profiling, that does help.
Logged

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: If you print profiling targets for a Canon Pro1000 Read this caution
« Reply #6 on: October 17, 2021, 11:22:20 am »

To the end I follow Don Hutcheson and no matter the target, for Camera/Scanner Profiling, I ALWAYS make a custom reference file (read the target with a Specto). I've done it for the Hutch Color Target (Reflective), as well as, my Colorcheckers

This way you can me less 'hypercritical' about the age of your Targets; just make a fresh reference file.

Not to mention, I never really trusted those BATCH reference files, I use to spend so much time hunting down and downloading; just make my own; and since I'm using the same Spector for Profiling, that does help.

I make my own reference charts measured with a spectro too for calibrating scanners and cameras. In my experience batch produced scanner charts with an averaged CGATs file were pretty bad.

But my needs differed and unique to a set of products.

The charts I needed to make for verifying in manufacturing and QC/QA had to cover many areas such as detecting registration errors where the sensor chip wasn't exactly perpendicular as well as measuring other geometrical unit variations. Color accuracy was only a small part of what the target/chart had to provide. And the areas of colors as well as neutrals, were often irregular in shape and had gradients which made spectros useless. Most of the color measurement stuff was to make sure the CFAs of the devices being tested had consistence across the sensor as well as were within product spec. limits.
Logged
Pages: [1]   Go Up