Pages: [1]   Go Down

Author Topic: Paper profiling questions: best # patches, etc  (Read 4716 times)

santodx5

  • Newbie
  • *
  • Offline Offline
  • Posts: 17
Paper profiling questions: best # patches, etc
« on: May 24, 2007, 10:05:22 pm »

Hallo,

I am a student from Brooks. I'm taking color management class right now
and I have some questions for you guys:

1. How many target patches are the best for paper profiling. I have
Bill Atkinson's target up to 4000 patches that I can read with
measurement tool. Does more pacthes = better accurate profile?

2. What resolution should I print these target patches? 1440 or 2880?
Does printing with 1440 vs 2880 will create different color? or can I
use one profile to print the other resolution and vise versa. Does the
myth 2880 > 1440 true? Can you tell the differences?

3.Will RIP help me to create a better printing quality?

I use Epson 3800 with Eye One Photo.

Thanks,
Dan Santoso
Logged

Christopher

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1499
    • http://www.hauser-photoart.com
Paper profiling questions: best # patches, etc
« Reply #1 on: May 25, 2007, 04:43:25 am »

Quote
Hallo,

I am a student from Brooks. I'm taking color management class right now
and I have some questions for you guys:

1. How many target patches are the best for paper profiling. I have
Bill Atkinson's target up to 4000 patches that I can read with
measurement tool. Does more pacthes = better accurate profile?

2. What resolution should I print these target patches? 1440 or 2880?
Does printing with 1440 vs 2880 will create different color? or can I
use one profile to print the other resolution and vise versa. Does the
myth 2880 > 1440 true? Can you tell the differences?

3.Will RIP help me to create a better printing quality?

I use Epson 3800 with Eye One Photo.

Thanks,
Dan Santoso
[a href=\"index.php?act=findpost&pid=119487\"][{POST_SNAPBACK}][/a]

Just three short answeres, because i don't have much time right now.
1. Yes and no. On very smooth papers it can be better on roughs you won't see a huge difference. Personally I think around 3000 patches are enough for everything. Even 1500 for faster testing.

2. You print your targets at the same resolution as you want to print you pictures later. I would use 1440 for almost everything, but I would suggest that you make a testprint first with both and then decide for yourself.

3. Yes, but not much. Especially with the 3800 I wouldn't buy a RIP.

Christopher
Logged
Christopher Hauser
[email=chris@hauser-p

Ernst Dinkla

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4005
Paper profiling questions: best # patches, etc
« Reply #2 on: May 25, 2007, 05:13:26 am »

Quote
Hallo,

I am a student from Brooks. I'm taking color management class right now
and I have some questions for you guys:

1. How many target patches are the best for paper profiling. I have
Bill Atkinson's target up to 4000 patches that I can read with
measurement tool. Does more pacthes = better accurate profile?

2. What resolution should I print these target patches? 1440 or 2880?
Does printing with 1440 vs 2880 will create different color? or can I
use one profile to print the other resolution and vise versa. Does the
myth 2880 > 1440 true? Can you tell the differences?

3.Will RIP help me to create a better printing quality?

I use Epson 3800 with Eye One Photo.

Thanks,
Dan Santoso
[{POST_SNAPBACK}][/a]

* I hope this isn't meant to embarrass the teacher next time :-)

1/ depends on the quality of the target creation, the ability to get consistent readings of such large targets and the software that has to cope with the measurements. there are some good arguments for printing more targets of smaller size and do more measurements that are averaged before profiling. Good when the printer isn't that consistent and the measuring has to be done by hand. Bill Atkinson's target is considered as one of the best around but his profiling method is in the same category so suit one another.

2/ there are inkjet models on the market that have claims they can be profiled with one profile for several resolution modes. The Epson 3800 is a first for Epson but the HP models have it all. The Epsons before the 3800 had a different inkload per resolution setting, a compromise in the use of the available droplet sizes per resolution. Check the manufacturer's profiles available for the printer models and you will see whether there's distinction made per resolution.
Be aware that for the Epsons at least the company advises not to use the speed (bidirectional) setting to make profiles for or print targets with, the order of ink lay down is reversed on the way back of the heads and by that the color changes between the strokes. This will be compensated partly by weaving but in the end there's a difference between uni- and bidirectional printing. As I understand it there has been a Canon printer that had two heads with reversed ink order to get that controlled in bidirectional printing. In printers that only do bidirectional and with the modern weaving, dithering methods + only small droplet sizes this may be less of a problem.

2a/ the myth stays a bit obscure in your message but I guess you mean print quality.
Depends a lot on the paper coating quality, printer consistency and what quality is asked for the job. The highest resolution usually suppresses any banding that may exist in the resolution below it (printer consistency, maintenance) may enhance gradations but in few cases it will actually add to better detail if one actually has the eyesight for that quality and the image file contains that detail for the printed size. With Epsons the Dmax may be better at the highest resolution compared to the one below it (the same compromise mentioned before). More a concern for B&W printing.

3/ that's meanwhile a myth. The RGB-device CM controlled normal printer drivers with good profiles will do 90 % of the jobs as good as a RIP will do those jobs. Meaning color and image quality on the usual media. RIPs (and that is a heterogene category meanwhile) still have advantages for special media, printers, workflows. On workflows there are often cheaper solutions available that will use the normal driver combined with cheap but good applications. The price of the software and skill needed to make profiles for CMYK device CM as used in 90% of RIPs is another factor to consider.

Ernst Dinkla

try: [a href=\"http://groups.yahoo.com/group/Wide_Inkjet_Printers/]http://groups.yahoo.com/group/Wide_Inkjet_Printers/[/url]
Logged

santodx5

  • Newbie
  • *
  • Offline Offline
  • Posts: 17
Paper profiling questions: best # patches, etc
« Reply #3 on: May 25, 2007, 03:51:56 pm »

Thanks guys,

I think I got some answer. The only thing that I need to know is to decide how many target patches do I need for my paper profile.

More input will be appreciated.

-Dan
Logged

madmanchan

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2115
    • Web
Paper profiling questions: best # patches, etc
« Reply #4 on: May 26, 2007, 09:20:28 am »

My rule of thumb: for Photo Black papers, I use 4096 patches. For Matte Black papers, I use 1728 patches.
Logged
Eric Chan

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20614
  • Andrew Rodney
    • http://www.digitaldog.net/
Paper profiling questions: best # patches, etc
« Reply #5 on: May 26, 2007, 10:35:15 am »

Generally speaking, more patches will build a 'better' (notice the quotes) profile but there's a point of diminishing return. In a prefect world, you'd measure 16.7 million patches to fully exploit and measure the color space of a device. Obviously this would take forever, you'd have to output a heck of a lot of prints and the resulting profile would be larger in file size than most images! So the idea becomes, output a representative number of patches that doesn't drive anyone nuts, then the software will extrapolate the other data.

I've built profiles using patch samples as small as 18 (the profiles generally suck although, on a very well behaved device, they actually worked.... This is rare) to as many as 11,0000. Generally, with Bills targets, which are excellent, the 918 does a good job for most devices. Personally I use 2700 patch target for devices where paper size is letter and 4096 for larger format printers, and printers who's drivers are not as well behaved (linear) as I'd like (most of the current ink jets using manufacturers drivers). The difference between say 2700 patches and 11000 takes very close examination (your nose has to touch the print), making it only an interesting exercise but pretty useless for everyday work.
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

Schewe

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6229
    • http:www.schewephoto.com
Paper profiling questions: best # patches, etc
« Reply #6 on: May 26, 2007, 01:06:37 pm »

Quote
The difference between say 2700 patches and 11000 takes very close examination (your nose has to touch the print), making it only an interesting exercise but pretty useless for everyday work.
[a href=\"index.php?act=findpost&pid=119705\"][{POST_SNAPBACK}][/a]

Also, don't lose sight of the fact that having all those samples doesn't help if the profile making software can't really use them to produce accurate and smooth rendering profiles. ProfileMaker and Profiler (both now from X-rite) seem to have a sweet spot at the 2700 mark (which is why I think Andrew tends to work at that count, right dog?).

I remember when the first i1 came out with the 288 (is that the right count?) sample target-all the color geeks said it was way too small yet a lot of people produced some pretty useful profiles with it-and if you were draging the i1 over the paper the less time draging the better.

The other thing to factor in is what colors are or are not in the sample set. Bill Atkinson did a lot of testing to try to optimize what colors would produce the most useful data for the creation of a profile. Which is why his 9.18 targets are widely used-they're pretty darn good in terms of the range and relevance of the samples and it seems the data points are really good for ProfileMaker in particular.
Logged

santodx5

  • Newbie
  • *
  • Offline Offline
  • Posts: 17
Paper profiling questions: best # patches, etc
« Reply #7 on: May 26, 2007, 03:19:58 pm »

Thanks guys,

One more thing, it seems Bill uses 203 resolution for his pacthes. Should I print as it is or should I upscale it to 300/360 my general printing resolution?

Also can I resized the patch coz I want to print on A4 paper?

-Dan Santoso
« Last Edit: May 26, 2007, 03:22:12 pm by santodx5 »
Logged

Schewe

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6229
    • http:www.schewephoto.com
Paper profiling questions: best # patches, etc
« Reply #8 on: May 26, 2007, 05:18:43 pm »

Quote
Also can I resized the patch coz I want to print on A4 paper?
[a href=\"index.php?act=findpost&pid=119737\"][{POST_SNAPBACK}][/a]

No, you can not change the size of the color sample on the swatch page...the resolution of the image file doesn't matter at all. What matters is the DPI setting on the printer-you should use the same output DPI as you are planning on using for images...and the EXACT media settings as well.
Logged

Wayne Fox

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4237
    • waynefox.com
Paper profiling questions: best # patches, etc
« Reply #9 on: May 26, 2007, 07:31:12 pm »

Quote
Thanks guys,

One more thing, it seems Bill uses 203 resolution for his pacthes. Should I print as it is or should I upscale it to 300/360 my general printing resolution?

Also can I resized the patch coz I want to print on A4 paper?

-Dan Santoso
[a href=\"index.php?act=findpost&pid=119737\"][{POST_SNAPBACK}][/a]

You should really read Bill's FAQ thoroughly.  In it he explains the reason behind the resolution - here's a pasted quote from his FAQ...

"Why are all your profile targets at 203.2 pixels per inch?
That resolution is exactly eight pixels per millimeter.  
Target dimension requirements for most spectrophotometers are specified in millimeters, including some that use increments
as small as to 0.25 mm.  For reasonable file sizes, only 4, 8, or 12 pixels per millimeter can exactly make 0.25 mm."


BTW, I'm finding that I'm of the same  opinion as others, for high quality photo-paper (Currently I'm liking Kodak's professional Gloss and Lustre best right now), I use the 4096 patch test.  For matte papers/canvas etc.  it gets hard even getting accurate readings with that many patches so I'm using the 1728 targets.  I'm currently using the Eye-One i0.
Logged

madmanchan

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2115
    • Web
Paper profiling questions: best # patches, etc
« Reply #10 on: May 26, 2007, 08:27:16 pm »

Quote
Also, don't lose sight of the fact that having all those samples doesn't help if the profile making software can't really use them to produce accurate and smooth rendering profiles. ProfileMaker and Profiler (both now from X-rite) seem to have a sweet spot at the 2700 mark (which is why I think Andrew tends to work at that count, right dog?).
[a href=\"index.php?act=findpost&pid=119725\"][{POST_SNAPBACK}][/a]

I believe the upper limit for MonacoPROFILER is 1728 patches (that is, the software doesn't let you scan or import more than 1728 patches). The point Jeff is making is right on, though: very little point doing a gazillion readings if the profiling software only makes use of a coarser sampling.

The other problem with reading a billion patches is measurement noise. Despite how good the instruments are, there are still noise in the measurements (and measurement conditions) regardless of how well-controlled the environment is. The more patches, the less likely that a given additional measurement actually contributes any meaningful data.
Logged
Eric Chan
Pages: [1]   Go Up