Transmissive or Reflective Denistometer for CTS imaged screens?

[ZooCity]

What's the way to go about linearizing CTS output?  The positive is black but the emulsion is pink, blue, etc. so I'm not sure if transmissive or reflective is the correct tool. 

[blue moon]

place some film on the screen and measure that. then just for good measure (pun intended) do the same with several different types of paper just so you see what's going on with the wicking of ink. I am thinking that photography paper might be the best option as it should not wick the ink. Film's WP layer will make it spread.
Another option is to burn the screen and print on the platen then measure that. We'll be trying most of these on tuesday.

pierre

[Dottonedan]

Pierre is on the right track. Even with the photo paper it does not represent the final results. If you are going to test for exact results, you want to use the transmissive densitomitor to measure the final result on your substrate.

To take that further, You will have different results on various subtracted. 100% cotton, 50/50.  Underbase or no underbase. So if you want to go in the direction of s Mark Coudrey, you might have 4-6 different cuves to compensate for different production runs.

I would not do your test on film. First, film in no way represents any type of obsobtion that goes on with wet ink and emulsion. To get close, you would have to test your prints on a screen coated with your same emulsion (with no coloring) or a clear emulsion. Then zero that out on the clear and measure your dark area. I doubt you can get a hold of your brand or any, with no coloring. The next best thing would be photo paper. This does bleed a tad, so be aware. Very close but not exact. Then, Even with photo paper, your true answer is on the press. Once wash out and ink added and squeegees pass only then will you get a true dot gain measurement. Then, you can compensate in your curves but your parameters all need to be consistent to repeat that.

Same squeegee duro, same print speed, same mesh, same thickness on emulsion, same pressure, same tension, same garment type.


I've seen Dmax  (if I remember correctly), at 3.75- 4.25 depending on print modes and speeds.

[Full-SpectrumSeparator]

I would never linearize with a densitometer. 

You print a test with your 100% values preferably as 1" squares of your desired halftone shape and LPI and on the mesh you are linearizing for.

Then you scan that in at high resolution like 1200dpi or 2400dpi.   

Then you would take the colors and perform various functions to make greyscale and make the pink or color of the emulsion to white, and set your black and white levels for the ink color and background.

Then you select each area of the 1" squares of your halftone percentages,  and you perform an "Average Blur" to find the average grey value of the resulting halftones with their dot-gain or dot-loss etc...  and that is the value for that percentage which you would know as your input/output value, and reverse it to create the linearization but this is not absolute so you usually have to check again.

However, there are a few main points you want to linearize -- the separation to the halftone, the halftone to the film or CTS, and also the final print because you will have different results from the linearized CTS when it actually exposes and prints, depending on mesh and other factors like your LPI etc.   

A densitometer is not going to give you the averaged values of the halftone percentages and their resulting grey values, as far as I know, it is for measuring your black-level density for a solid area on film...   it might automatically sample an average of the area it captures, but this is not the same as a density value for black on clear film.     

I don't need a densitometer or a spectrophotometer etc to perform linearizations at each stage... it is all just in using logic and science to find your input-output changes and correct for them, using a scanner and knowing how to actually perform these calibrations.     

CTS is notorious for problems in how the ink deposits on the emulsion compared to inkjet film or thermal imagesetting.   So you want to really scan your resulting screens and see what is happening to your dots or details or the information you send to print.       Also you can scan the screen AFTER you expose and wash out, and linearize the actual screen from the CTS,   then you can linearize the print from the screen.     But you want to follow each step and linearize first your sep to your halftone (various halftone LPI and resolution settings will affect whether you even generate the right % value you wanted digitally to begin with),   then you can linearize the resulting film or CTS etc halftone "mask" and see what kind of input/output changes happened and account for them,   then linearize the washed out screen, then linearize the final print and make sure to record for what settings like mesh and EOM and exposure time and the ink being used, tension, shirt fabric,  squeegees and speed/angle/pressures etc etc...    or you can also just attempt linearizing from the final print only and try to compensate and get the best range you can out of your full 100% values.       


The main problem with linearizing your dot-response curve with halftone print tests, is that you are linearizing for something that rarely happens with actual artwork -- the large areas of single % values of halftones which would create various % and sample and measure them to get their values...   

In an actual art and separation and halftone for printing, there are rarely large areas of the same halftone %,  it is usually lots of various % of halftones all mixed around in different areas...    and especially with index or hybrid interlocking separation halftone screens you would want to actually linearize to your final dot-cluster/pixel resolutions  and there are ways of adjusting the size of your actual halftone dots,  compensating for dot-gain or dot-loss on the actual artwork separations rather than just test patterns,  although you can still utilize test patterns to capture the general deviations from the desired value you want, and account those into your final rip halftone cluster compensations.   

But anyway, most of this can be ignored if you generate accurate separations and have the halftone dots in the right places to begin with, then you have a lot of control in printing variables to make it work on press, without ever having to go into all the crazy technical detailed science behind how or why it works or doesn't work.     You can go to a lot of trouble linearizing everything but if your actual separations and the composite of the halftones being printed are not accurate in the first place then what is the point?   

To answer your question again, personally I would never use a densitometer for screenprint linearization, it is like apples and oranges.     If you know you have good density in your solid black, and it exposes, then you don't need to measure it with a densitometer.     The densitometer is really not "smart enough" for my purposes in linearizing, and especially with a CTS screen you have a lot of other factors involved.    Does the densitometer measure an area of halftones and still give a density reading?   Probably, but I would not rely on it as an accurate measure of your resulting grey-value that would be those halftones.       Also, the printed screen CTS is different than the exposed and washed out screen, and also what about the final print on the shirt?     

The simplest approach I have found is scan the final print from a test with areas of the same % halftones, and average-blur the halftones after correcting for the shirt white level and ink black level,  and then use those values you measure and the amount of which they deviate is the amount you compensate, then test again and try to get to that linear point where "What You See Is What You Get"....   but for 2-color and 3-color and more there are a lot of other factors involved in what resulting average color-values you are getting from what you want.   

This is all really overkill if you just put the most effort into the variables that have the greatest impact on the final print.     You can linearize perfectly and then still output a perfect halftone gradient at the same angle with dot printing on top of dot, and everything goes out the window because that is just not how you reproduce a blend between two inks with halftones.    If you want to over-print the inks and know their opacity settings and utilize the overprint to generate the 3rd color blend, then that is an entirely different method of blending colors together, but where they are halftones they would need to be in exactly the same places such as with CMY or CMYK with flamenco printing which produces more accurate results but your registration can throw it off,  this is why offset paper printing went with rosette patterns a very long time ago.     Flamenco in CMY or CMYK where the CMY are all at the same angle is actually the most color-accurate way to print process or overprint blends where you want for example a red from a magenta and yellow printing together,  but because of the need for exact or perfect registration, this caused way too much misprinted material.    Rosette patterns resulting from the angles being different then creates a less color-accurate result but viewed from a distance and when it blurs together it still creates similar tones, however this works more consistently when you have registration shifting from print to print such as what happens in offset and paper printing,  while giving more "passable" final prints and less "rejects",  so along with many other things in our world it comes down to profit and what generates more profit and less waste, while sacrificing color accuracy is allowed because of better profit margins or you could say better quality-control in terms of what is capable in business terms.     Interlocking halftones of blending colors where you don't need or want overprint - such as red and yellow making orange, will create more accurate values and actually be more friendly to registration and dot-gain/dot-loss shifts,  than flamenco printing with all the dots at the same angles.        All of this is dependent upon the separations being run correctly in the first place.    You can linearize everything but if your halftones are not even in the right places to create the right blend then it is not going to really help final multi-color prints much.    It is mostly for when you have 1-color halftone prints and want to capture the most dynamic range from only 1 color and 1 screen having a full range of % shade values.    2-color and 3-color and more type of prints have a very different scheme of calibration and linearization, especially whether you are printing process inks and overprinting like a CMY or CMYK or hybrid, or opaque inks or any variation between.     

    Linearization should be really easy for anyone to perform as long as you have a high resolution optical scanner or maybe even close-up pictures with your phone (the megapixel resolution these days can capture halftones at the right focal distance) and a graphics program to take the measurements.    Think of it this way,   make a 1" square and fill with 50% grey,  then convert to halftones being sure your profile converts exactly (50% would be a checkerboard pattern with certain types of halftone algorithms),  then you photo or scan that area and select the square of all the halftones, and perform an average-blur (photoshop has this blur filter option), and this blurs all the halftones the way your eyes would from the right viewing distance where your visual acuity defaults to the constrast-sensitivity-function for your LOD - level  of detail in your vision.    The resulting blurred value, for example a 50% grey halftone, will most likely gain and could be even up to 30% or more gained, so as an example the 50% medium grey would become an 80% dark grey.... so you would create a curve effect to be applied to greyscale images where the 50% level reduces down to 20%.... but this is not absolute again, because the 20% may actually only gain to become 25% or 30%,  not 50%, so you would run a test with all the values and see what each input becomes as an output... .perhaps you would see that your 30% dot pattern area becomes 50% grey, so you would know that is how you should set your input-output value in the compensation-curve for 50% so it becomes the right value on output --- but keep in mind, are you just linearizing to the CTS?   Then this would only be making your image mask to be more accurate... you need to measure and get the values accurately for the actual exposed screen or more importantly the actual printed result.    The way dots and holes expose to emulsion and mesh bridging or not affects things, all of the screen and emulsion and exposure variables etc affect the outcome, and then the print variables all affect what those things in the screen become on press.     But still, try to make it easier on yourself, just linearize first for your final print, and only if you have extreme dot-gain or loss then look at the steps along the way to see where you are just not even putting that information onto the screens or its getting lost or gaining too much etc.    For a CTS imaged screen, I would scan it in and then force the hues of the emulsion color to white, and perform average-blur on the background emulsion white, and on the ink solid areas, to get the levels to apply to white/black levels, then go in and perform the average-blurs to the various % levels of halftones to test.    The test-pattern vs. actual-art halftone cluster linearizations are a bit deeper than this and would make more sense being explained or shown in a better way.     Hope this sheds some light on linearization.

[ZooCity]

Unit is wax if that makes a difference.

Clear emulsion and a transmissive might be the way to go here.  I'll reach out and see if I can get ahold of some.   A transmissive would work here?

Do the reflective densitometers really work if you try to measure a print on jersey weave fabric?  Sounds like a tall order for any device given the inconsistent substrate.

Wikipedia talks about densitometers that can do both, anyone have experience with one?

I agree that your on press result is the one that matters.

However, our RIP (Harlequin) has first a general calibration that I want to dial in to the output direct to screen.   From there you can apply Tone Curves as they're called for specific substrates or specific jobs. 

There are multiple reasons for starting with "at device" calibration and then curving the output as needed for the application on press.  My biggest reason is that my staff is not going to print as well with the same linearization curve that I might.  You hear this a lot where the pre-press dept knows that production likes to use too much pressure as a solution or maybe the pm won't buy new blades or use better mesh or whatever it is but my point is you end up tweaking to fit the average, regular ol' joe, production run in your shop.  I think starting with firm knowledge of what's hitting the emulsion is the way to go since it's tough to control what's going to happen once those screens land on press.

[Dottonedan]

Correction, it's reflective densitomitor that would read the printed garment. The transmissive is for reading film.

[Dottonedan]

The tough thing about going to using a rip and compensating is that ...for those who have not created art and then separated in the past (for a divice that will compensate for you, the art can come out way too light....because they have already been compensating as they build. Therefore, adding excessive compensation.

The way and probably only real way to measure correctly or as best you can...is to use multiple readings of the same % area and average them together to give you a good reliable number.  I'd use 2" squares filled with a 3% 5%, 10, 15, 20, 25 etc.

[ZooCity]

Spectrum separator - damn, that's a post right there.  Need to re-read it tomorrow but it appears you are familiar with the tree I've been looking to bark up.   

I share the perspective of "linearizing" in a non-traditional sense by focusing on where it counts.   The suggestion to scan up prints or even the screen itself is super cool and I'll try it out.  Thanks.

Dan - you are correct.  I've outputted art for us linearized that went nowhere but looked great unlinearized. 

In my scenario this is where the Tone Curve settings come in.  I comp every channel we output in photoshop but it's a pain to do it in there given it's super limited representation of channel opacity/percent areas on screen.  I know that's how we've all learned to roll with channel seps but it would be nice to ditch the constant levels and curves adjustments that eat up a good portion of time spent doing sep work.  To me, the whole meaning of linearizing your output is to be able to sep a job quickly and know that your output will compensate.  It's really the same thing,  just a different adjustment you are making for the linearized output.   It also allows you to be way more consistent in how files are handled for output, i.e., try training someone in pre-press who's never printed and you'll see what I mean here.  They don't have the years of intuition to know how far back to pull what and why and where.  Adjusting output to match best what appears "naturally" on screen sounds better to me.

[blue moon]

our densitometer will read both transmittive and reflective. It also has different modes for different readings, halftones and ink density being just two of them.

You can try reflective readings off the imaged screen, but I am not expecting usable results. I tried before and got nowhere. We'll try with our CTS next week, if you remind me I'll post results.
Reading the shirts is also a no go. the weave interferes with readings.

The easiest thing to do would be to grab a linearized piece of film and place it over the imaged screen then compare the dot sizes with a microscope. That should get you very close.

pierre

[blue moon]

also, adding film or paper to the screen changes the head distance and that will impact the actual dot size (I would imagine it spreads more as the distance increases), so paper or film are really not that ideal.

pierre

[Dottonedan]

The final beauty of knowing what you are getting on press and having your rip compensate for you (across the board on all jobs the same way ) is as you say, "consistent". Imagine a perfect world where your artist only had to create and then can expect to get very close to what they created (without ever having to think ahead) about gain.

[jvanick]

The final beauty of knowing what you are getting on press and having your rip compensate for you (across the board on all jobs the same way ) is as you say, "consistent". Imagine a perfect world where your artist only had to create and then can expect to get very close to what they created (without ever having to think ahead) about gain.

The rheology of the ink will play into this too...

you'd almost need a curve for each garment type, ink, LPI setting,  screen mesh, squeegee type and angle, possibly even each emulsion.

Something really cool would be a reflective densitiometer of some sort attached to the computer hosting the rip.
put a fidicual image on the bottom of each screen (Especially simple 1 color or whatever stuff).  read the printed output and then use that to calibrate the rip over time.

(you may even have just an image generated.. as a 'calibration' image that the computer knows how to read/process)

there would certainly be some fuzzy logic involved, but with the proper programming resource, it could certainly be done.

[ZooCity]

That would be sick.  Isn't that what a lot of offset printers used to do with the color bars and what not?

[Alex M]

The final beauty of knowing what you are getting on press and having your rip compensate for you (across the board on all jobs the same way ) is as you say, "consistent". Imagine a perfect world where your artist only had to create and then can expect to get very close to what they created (without ever having to think ahead) about gain.

The rheology of the ink will play into this too...

you'd almost need a curve for each garment type, ink, LPI setting,  screen mesh, squeegee type and angle, possibly even each emulsion.

Something really cool would be a reflective densitiometer of some sort attached to the computer hosting the rip.
put a fidicual image on the bottom of each screen (Especially simple 1 color or whatever stuff).  read the printed output and then use that to calibrate the rip over time.

(you may even have just an image generated.. as a 'calibration' image that the computer knows how to read/process)

there would certainly be some fuzzy logic involved, but with the proper programming resource, it could certainly be done.

On top of that you are thinking of a shop with one auto... Imagine having to calibrate 4,6,12,20 autos exactly the same to accomplish this!
Off contact, squeegee pressure, pallet level... Etc...
You can truly get lost in this if you want. It is just a matter of what point you call it done... It Will be at a different point for everyone.


Sent from my iPhone using Tapatalk

[DannyGruninger]

My opinion..... It's great to linearize and get a good baseline setup that shows smooth transitions, tonals, etc. BUT I firmly believe screen printing is an art form of itself and should be treated as such. Meaning that the best form of linearization is done by eye and lots of testing on the press. Understanding how colors peak, buildup, gain, blend, etc is a complete art form to me and not two jobs are the same requiring different parameters for them. We've spent a lot of time making our sim process what it is and I will say there's not much science behind what we do as our curve for outputting and the way we sep is highly based upon doing things by hand based on what we "see" on screen and press. Once you get your baseline that you can control its all about making slight adjustments to the output device until your prints look amazing. At least thats how we have done it here. I dont care if a dot is true to the computer as long as our prints look killer which most of the time they do.