Custom Curve Necessary?


Just starting with a mostly selenium setup. Using a Mac, 3880 and Print-Tool.

I’m wondering if, given the variability of Hahnemuhle Photo Rag, a custom curve would improve my linearization on that paper. Or would the custom curve just tend to be off differently? I’ve attached a screenshot of my latest linearization-checker (hope that’s OK).


Added screenshot (I hope).

I for one would be very happy with that linearisation! And I suspect that if you repeated it daily for several days you’d get slightly different results each time, due to sampling variation, but looking roughly the same. But I don’t know how much variation you’d get sheet to sheet or batch to batch with HPR. It’s supposed to be highly variable.

Thanks for the reply Brian. I have made a few of these a couple of days apart and, like you say, just a little different day-to-day. My lin for Canson Baryta is a bit farther off and I suspect it would benefit from a Custom Curve. I would just like to get a few observations from others—I’m in the “not sure what to expect” phase at this point. Here’s the Canson:

Not the worst I’ve seen, but it’s heading towards a custom curve. How much drift can you tolerate before you get a custom curve made? There isn’t really an answer to this question, at least not that I’ve seen. I have a little technique where I create a Photoshop curve that applies the inverse of the linearisation curve to the image. I tested this approach by reprinting the 21x4 using my PS curve and remeasuring and it works fairly well, although I wouldn’t use this approach for large departures from linearity.

That kink at the bottom would concern me the most. I’d be tempted to repeat the exercise to see if it was just a measurement glitch. Did you apply GO to the Canson? My understanding is that you’re supposed to measure gloss with a coating of GO. There was a recent posting by Jon discussing how much GO for different papers. See this thread.

Here’s something that puzzles me, and also one other piezo user I’ve been talking to. Why does my printer and his and yours behave linearly on some papers and not on others? We know that printers are supposed to be fairly linear when they’re new, unless you’re unlucky. Certainly mine seemed to be on the matte papers that I was using at the time. So you’d like to think that IJM made all these standard profiles that ship with QTR on new printers. But they must also have done them at different times, after all gloss became an option much later than matte. And Canson is a recent addition to the papers that IJM sell. So is a new printer used for each new standard curve? If not then that might explain the phenomenon that we’re seeing - where one paper is linear and another not. But that would also mean that the standard curves are not always the gold standard in linearity. Or is there another explanation?

[This is sure to provoke Jon and so I will take cover and don a flame-proof suit.]

Regarding the CBP lin—yes that reversal at the bottom bugs me. If the curve was off all in one direction it would seem less problematic. I’ve thought about applying a PS curve to try and straighten this out like you did (like a transfer function applied right before printing) but I’d rather it just be correct, more or less.

I hadn’t applied GO before reading this target so I guess I will go back and try that. Maybe that’s the reason for the reversal?

My printer isn’t new (three years old) but it’s had infrequent use until now so the printhead should be fairly fresh. I will post the new Canson lin when I have it tomorrow.

Thanks for your conversation Brian.

If you use this approach then you’ll need to turn the PS curve on whenever you turn soft-proofing on. Which presumably is whenever you’re editing for printing on that paper.

Have you seen the technique whereby you control where on the page GO is printed, rather than the page’s entire printable area? This means you can reuse a sheet to print several 21x4 and just cover one 21x4 with GO at a time. (I’m not a total Scrooge, but I like to minimise waste.)

You’re welcome.

Here’s what my HPR does on a brand new 3880. You just have to love a paper with variability. I’m heading for the Canson as soon as I use up my HPR. On the subject of Canson Rag, none of the available IJM curves are anywhere near as good as this. I’ve ended up using a custom curve for a 2400 which is the best of the 4 that I tried.

I have two new Canson Baryta linearization checks—one using the Selenium curve and one using the carbon curve. Both have GO applied.
Looks like I’m going to get a custom curve made.




I’ve been printing the GO layer by using a QTR Flush image for the GO channel and printing through Print-Tool’s “RIP Calibration” setting. I like the GO to print on top of the image only, so I just crop the same and let it go. Not sure if I read that in a post or where it came from.

Have you seen the technique whereby you control where on the page GO is printed, rather than the page’s entire printable area? This means you can reuse a sheet to print several 21x4 and just cover one 21x4 with GO at a time.

Well that’s a new one for me. I hadn’t seen that idea. The standard amount of GO is 30,000 units although in that other thread Jon suggest 40,000 for recent batches of Canson Baryta. I wonder how you control the amount of GO using this approach? My limited understanding of calibration mode (I’m on Windows, but it should be the same) is that it doesn’t use any curves and if you wanted to control how much GO / ink is put down then you use the ink calibration slider, but how do you set this to get 30K or 40K units of GO?

There is a possibility that your problem is that you’re putting down too much GO. I saw the linearisation plots from another user who had put down too much GO and while I don’t recall exactly, I think it produced results like yours. Before you pay for a new profile I’d be inclined to repeat the exercise but printing GO the conventional way.

The conventional way to print GO is to print a 1" square white image and QTR automatically expands it to cover the printable area of the page. The variation on this that controls where you print GO is to have a TIFF that matches the printable area of the page exactly. This is harder than you think. You need to allow for the 1/8" non-printable border and not a pixel more. I could send you my A4 version but I assume you’re on Letter. Then where you want GO printed you have white and elsewhere you have black. It’s a little tedious to set up, esp locating the white rectangle exactly, by which I mean allowing for the 1/8" border. But it works.

If you still end up with a linearisation plot like the last ones, I’d consider them candidates for my Photoshop curves approach, given that they’re not extreme and that they’re relatively smooth. But I have a hunch that some of your problem may be excess GO. If you plump for a custom curve, make sure you print it with the amount of GO that you intend to use in practice.

The GO instructions are in the Piezography Manual. Basically, you make a small TIF and print it with the GO curve instead of the normal curve. The GO curve is a special curve which will only print at 255.

Here’s a test with the 30000GO curve applied as per the manual. Looks even wonkier—worth a try though. Maybe come Monday Dana would venture an opinion about these.

Using my earlier, unconventional method, I was using the ink slider til the coverage looked like I wanted (fairly glossy).

I’ve also posted today about how matte the extreme blacks look even after applying GO. I tried posting a picture of it but it’s hard to see.


Yours looks like a workflow issue. Are you printing from Mac or PC?

The 21 step file has either been converted when opening, or you are not printing from the Gamma 2.20 workflow, or not printing with color management off, or possibly a combination of these in order for it to print so dark. Your mid-tones appear to have been converted into a space other than Gamma 2.20. IS that possible?


Hi Jon—

Thanks for chiming in.

I’m on a Mac and using Print-Tool. The Spectro is an ISIS.

I think I’m good with the rest but here are a couple of screenshots of my setup:

really small pics but I think I can see.
Do you know if the grayscale was converted when opening? Or if it was in a different gamma and you applied 2.2 to it?

I made it in ColorPort. I think I had to apply 2.2 to it in PS (it was untagged)—the squares still read the correct dot-percent numbers.

I will re-make it the morning if it’s incorrect.

That looks rather odd to me as well and doesn’t look just like a non-linear printer, although I guess in theory anything is possible. At least that kink in the deep shadows is gone. Since I’m on Windows I don’t have unintentional profile conversion issues, and can’t comment. All my conversions are intentional! I did once get a plot roughly like that by letting the inks settle in some desktop carts for a couple of months before using them, but since your HPR plot is pretty darn good that can’t be a factor here. I defer to the maestro.

Hi Williston~

Sorry for not responding sooner, I was out with the stomach bug for several days, and am now catching up on emails.

After reviewing your order history, I see you purchased the set of 3880 refillable carts and Piezography Special Edition ink in Nov 2014, then purchased SEL shades 1-4, and new C, LC and M position carts to set up your printer with a pure Selenium inkset.

Your latest linearization with GO looks horrible, and VERY dark in the mid-tones. Please confirm, are you getting close to linear results (top linearization, in your post# 2) with this system and Hahnemuhle Photo Rag paper/curve, but extremely dark results with gloss paper/curve? Are you using the latest QuadTone RIP version 2.7.5? QTR has a 21 step strip image located in the EyeOne folder.

Please let me know, so I can help you figure this out and get back to happily printing.
Best regards~ Dana