Free ICC profile creation tip. (patch count)

[Pauly]

We all love free stuff right?
I know 90% of you who make your own profiles most likely use i1 Profiler.

This information comes straight from X-Rite. It's on the internet, so it's not to difficult to find. But it was mentioned not long ago about how many patches to measure and everyone coming up with different numbers.

How to determine how many printer patches to use in the test charts you print.

This a topic that can generate a lot of discussion and there's a number of schools of thought on how to determine an exact "ideal" patch count.

We typically would recommend a minimum of 400 patches per channel for a quality printer profile, so if you were printing using an RGB printer (which would describe almost all inkjet printers, when using the printer driver), that would mean a minimum of 1200 patches.

We've found that going much beyond the 2033 patch count (the default patch count in the Advanced mode) gets into the area of diminishing returns pretty quickly. You won't gain significant improvement in gamut volume and could actually reduce the profile's quality if you go too high. We would typically suggest you could get a better result by measuring a target more than one time in the Advanced mode, saving each measurement separately, and then letting the i1Profiler software average them for you. That would tend to smooth out any irregularities in individual patch measurements. if you're doing hand measurements using an i1PRO2, it can be possible to introduce some unintentional variation in measurement if doing a single set of readings.

The reason you've read that "too many patches are overkill" is due to the fact that i1Profiler permits up to 6000 patches, and many end users feel that they need to use that overly large patch count. That would simply be unnecessary for any Epson, Canon or HP inkjet printer.

As said in the quote, i1 Profiler defaults to 2033 patches and it already has it as a saved target.
There's also other charts like:
IT8/7.3: ISO 12642 target
IT8.7/4 - 2005: a newer version of the IT8/7.3 chart which has 1617 patches (formerly 928 patches)
ECI2002: which is an extended version of the IT8/7.3 chart with added patches developed by ECI.

All those targets are in i1 profiler, and on the internet to download with a quick search.

Fell free to discuss, or add your own experiences.

 

[ColorCrest]

Xrite being Xrite, maybe they failed to mention a very important factor regarding patch count.

What number of patches provide the best neutral and near-neutral patches in the layout will provide a more accurate result from the ICC profile?

For machines running as RGB, I can offer 2371 as an example and it produces 168 near-grays which is enough to get a good average for gray balance. 2945 will produce 195 near-grays. These particular numbers will work even while sampling heavy canvas with the i1 or iSIS. If you’re familiar with G7 you might already have optimum CMYK layouts using their optimum gray patches.

Any typical patch count will utilize the gamut of the machine. One of the only benefits of high patch counts is producing averages of sample readings due to rough media textures, nozzle deflections, head mis-alignment, etc. Otherwise, high counts should be unnecessary.

 

[ikarasu]

The highest in my onyx is 2383, which is usually what I use. I havent had time to try a lower patch count... But I might make time just to compare.

 

[Pauly]

Xrite being Xrite, maybe they failed to mention a very important factor regarding patch count.

What number of patches provide the best neutral and near-neutral patches in the layout will provide a more accurate result from the ICC profile?

For machines running as RGB, I can offer 2371 as an example and it produces 168 near-grays which is enough to get a good average for gray balance. 2945 will produce 195 near-grays. These particular numbers will work even while sampling heavy canvas with the i1 or iSIS. If you’re familiar with G7 you might already have optimum CMYK layouts using their optimum gray patches.

Any typical patch count will utilize the gamut of the machine. One of the only benefits of high patch counts is producing averages of sample readings due to rough media textures, nozzle deflections, head mis-alignment, etc. Otherwise, high counts should be unnecessary.

There is no "1 solution for everything"

If you're creating targets in i1p, You can see the bottom few rows of neutrals, When you keep adding, you'll see patches keep getting added to the bottom rows until it jumps to the next large target set, then you'll see neutrals shrink until you add more again.
I've never had bad neutrals from i1p with low patch counts, software these days is that advance that you don't really need excess amount of neutrals in default charts.

High patch count does not produce averages unless you have multiples of the same patches. You are way better off doing a lower patch count and measuring twice or make the patches wider.

FYI Barbieri mostly use ECI2002 and IT8.7/4 for most of their targets. Theres only 1 target with 3536 patches

The highest in my onyx is 2383, which is usually what I use. I havent had time to try a lower patch count... But I might make time just to compare.

I've always used the 944 target with great results. It was a while since i tried a high patch count in onyx, i'll do it again and see what happens. last time i did it, i had the same results as the lower patch count.

 

[ColorCrest]

If you're creating targets in i1p, You can see the bottom few rows of neutrals, When you keep adding, you'll see patches keep getting added to the bottom rows until it jumps to the next large target set, then you'll see neutrals shrink until you add more again.

Yes, and particular patch counts "fill" the void evenly to build the matrix of not only near-grays.

I've never had bad neutrals from i1p with low patch counts, software these days is that advance that you don't really need excess amount of neutrals in default charts.

Different shops may have different and specific requirements for neutral color balance for their customers. Many problems stem from the sample reading of media from devices and not as much from software calculations.

High patch count does not produce averages unless you have multiples of the same patches.

Which is what high patch counts produce, nearly the same color upon which to average.

You are way better off doing a lower patch count and measuring twice or make the patches wider.

To expect better results from measuring twice, one would need to print two sets of swatches or turn a single set 90 or 180 degrees for the second reading.

High patch counts are usually unnecessary but certain patch counts will provide more near-grays upon which to build optimum profiles.