Need Help Profiling - curves not smooth, less-than-ideal gradients

[Seth Griffin]

I'm getting some odd readings on my calibrations - the generated curves aren't smooth. When I try to import the profile to another identical printer and then calibrate for that printer, it says that my readings are out of range, and can't be clipped. (In other words, I'm going to have to rebuild the profile for the other Roland printer we have.)

I'm using Onyx Thrive, an ancient and uncalibrated X-rite Eye-One spectrophotometer (rev. D), and Roland Pro II SJ-1045EX printers converted to dye sublimation. I'm printing on Beaver TexPrint paper that's (I think) 140gsm, and transferring to Fischer tri-poly fabric through a rotary heat press. (I'm not sure what the time under heat is off the top of my head; I think that it trips out the heat strips at 206C.)

I'm setting my initial gamut for extended or coated (both appear to be the same), because uncoated gave me washed-out colors and shadows. I've got it set for FDRP enhanced at the moment (rather than stochastic). The initial ink restrictions seem to go okay, more or less, as did ink limits. I set my ink limits a little higher than I might normally (i.e., seeing a very small amount of artifacts from pooling ink, but it's still drying okay) to allow for more saturated colors. Once I do the calibration, it all goes sideways. As you can see from the attached screenshots, my curves look more like steps, with small spikes all over the place. For instance, patch A1 has a density of 1.510, but then A5 has a density of 1.509, and so on. It doesn't matter whether I read in strips, or individual patches; I get results that still show spikes.

Any ideas what's going on?
Link to images:
https://imgur.com/a/fXEVnLn

 

[Jim Hancock]

Hello Seth!
I've done a lot of sublimation profiling using Onyx. I'm going to ask a stupid question, because sublimation profiling requires just a touch of faith. Are you reading the cloth or the paper? I ask because the weird jaggedness of your readings is very strange.
I would like to see pictures of each print step on the paper before transferring and on the cloth after transferring. You want to look at both the paper and the cloth when setting your ink restrictions & limits. On the paper - first step without artifacts and edge pooling on the patches, and on the cloth (for settings determined by eye) the point where you see no visual density increase. Look at both to make a judgement call. If you do send pics, please send highest possible resolution and send to my email in my signature.
Another consideration is your resolution. In my opinion, the following is all that is needed for sublimation. Hard substrates like mugs - 720x720. For soft substrates - 360x720 or 720x720. You will reduce ink saturation on the paper and considerably reduce print time. All the resolution does for you is to gain increased detail definition and smoothness. In your case, your cloth is your limiting factor. You will not see a difference between 720x720 and 720x1440, and probably won't see much usable difference even at 360x720 vs 720x1440. This is assuming your printheads are in perfect working.

 

[Seth Griffin]

Hi Jim,

Hello Seth!
I've done a lot of sublimation profiling using Onyx. I'm going to ask a stupid question, because sublimation profiling requires just a touch of faith. Are you reading the cloth or the paper? I ask because the weird jaggedness of your readings is very strange.
<snip>.

Hey, sorry I didn't get back to this sooner; we've been on short weeks due to the lack of business via covid-19.

When I'm reading the calibration - before the ink limits, but after the initial ink restrictions - I'm reading fabric-only. The Ink restrictions are the same kind of swatches (I think I have it set to 216 swatches each of CMYK) as the calibration, but the calibration has to occur before setting the ink limits. I don't think that there's any kind of manual reading I can do on the ink restrictions. I don't usually transfer the ink limit test; I look at what dried correctly without pooling or cracking, and set that as my upper limit. I'm going to be trying to profile our other 1045EX today (it didn't like the profile for the first printer, so, start all over...), and I'll get photos at each step, and put them together with the screenshots. It may just be that the particular printer I was working on is wonky; the test print is always significantly lighter than it is from our other Roland of the same model.

Our fabric is Fischer tri-poly, which has a very smooth satin face. I've tried dropping the resolution before--I think I was using v12 when I did that?--and the difference in details was visible to me. There seemed to be a difference in saturation and vibrancy as well; shadows ended up looking dull, and colors didn't look as lifelike. Years before I took the job, they went through months of negotiations with our single largest client, and the current resolution (720x1440) and the tri-poly fabric were what they wanted, because almost everything we do for them are large-scale photo installations. We've just kept it the same for everyone else, because of inertia.

 

[P Wagner]

The original X-Rite i1Pro is best suited to taking measurements from flat medias like paper or SAV, as opposed to woven or knit fabrics. Using a large aperture device such as a Barbieri LFP QB or new i1 Pro 3+ would be vastly preferable for measuring fabric. Because of the small aperture on your measurement device, there will be much more variability in readings on fabric, dependent on measurement direction, speed, user error, etc. Using iterations averaged across multiple Calibration charts will provide improved accuracy. That is issue number one.

The second issue is that you are using far too many patches in the Calibration chart. The default number of patches is 31 per channel (if I recall correctly), and it sounds like you increased that by a factor of 7x to over 200! I would suggest going the other way (reducing the number of swatches in the Calibration chart) to not more than 25 per channel. This should smooth the readings and largely eliminate the stair-steps in the Calibration density values.

 

[Seth Griffin]

Sadly, I don't have a lot of choice on calibration device, unless I want to buy one myself. (Which I do, since my boss doesn't seem interested.Probably an X-Rite, since I can't afford a Barbieri anything.) I'm typically making four passes over each individual swatch (A1 L->R, R->L, L->R, R->L, A2 repeat), so it should be averaging results. I can see what you're getting at though; if I was doing single pass with strip or patch readings, then variability in the fabric and grain changing the way that light is reflected would make perfect sense. But since I'm doing 4 passes, I would have thought that it would average out to the 'true' value.

I'll try reprinting my ink restriction swatch with fewer patches and see what it looks like. 31 patches would end up being less than a single page of patches though; one full page would be about 72/channel.

As an aside, I keep crashing and locking up Media Manager. There's something about the way it's handling Java that my system doesn't like. I'll have to reboot the system once I'm done with my current print run.

 

[P Wagner]

Sadly, I don't have a lot of choice on calibration device, unless I want to buy one myself. (Which I do, since my boss doesn't seem interested.Probably an X-Rite, since I can't afford a Barbieri anything.) I'm typically making four passes over each individual swatch (A1 L->R, R->L, L->R, R->L, A2 repeat), so it should be averaging results. I can see what you're getting at though; if I was doing single pass with strip or patch readings, then variability in the fabric and grain changing the way that light is reflected would make perfect sense. But since I'm doing 4 passes, I would have thought that it would average out to the 'true' value.

I'll try reprinting my ink restriction swatch with fewer patches and see what it looks like. 31 patches would end up being less than a single page of patches though; one full page would be about 72/channel.

As an aside, I keep crashing and locking up Media Manager. There's something about the way it's handling Java that my system doesn't like. I'll have to reboot the system once I'm done with my current print run.

I'd suggest physically printing several copies of the Calibration chart across the width of the media, and then measure each one to average from as opposed to multiple measurements of the same chart.

For fabric printing, there is generally little benefit to charts with more than 25 patches per channel. For example, IIRC Wasatch SoftRip for the Linearization chart used just 17 per channel with no ability to change the patch count. 20 patches yields measurements at 5-percent intervals, and 25 patches yields readings at 4-percent. By way of contrast, at your 217 patch chart, measurement values are at < 0.5 percent intervals.

In this case, less is more.

 

[Jim Hancock]

Agree with P Wagner, you are going overkill on the number of patches, which could account for the jaggedness of your curves. Go with the default of 31 patches. I get it on the resolution, it's what works for your customer. I would also suggest you press the total ink limits and make your decision based on the fabric. The paper will give you an idea of where you "might" set your ink limits, but you will probably be surprised at there being a difference based on the fabric. This will probably explain why your single color numbers are so high and your two color limits are kind of low. Remember, you are looking for the point where there is no real discernable color change in each color row on the fabric. You are probably putting down more ink than needed by using the paper as a guide. This can possibly give you some issues when printing the profile patches, as well.
Remember sublimation acts differently than vinyl, so you can't necessarily make any final total ink limits judgement based on the paper. This is what I mean by a touch of faith. The first time I did a sublimation profile, I did exactly that - used the paper to decide total ink limits and the result was awful. I finally decided to press the total ink limits swatch and decide from the fabric and the result was a very nice profile. Also, once you print out the ICC patches, look for artifacts and pooling - on the paper. This can be an indication you need to go back and tweak your total ink limits. I think between the advice from P Wagner and myself, we can help you get to a good profile.