Manual linearization and dot gain compensation

[mimosatexas]

I use the bitmap method in photoshop to produce films, and up until now it has mostly worked fine, but I am starting to push myself a bit when it comes to printing style and detail and have run into a few issues with my films.  I was wondering if there was a way to manually linearize and adjust for dot gain using photoshop instead of having to buy a rip.  Basically, I want to knock out the small dots and holes in nearly solid areas without losing detail in the rest of the film.

A related issue happens with choking underbases.  If I choke my underbase by 1px on a 300 dpi film for example, I end up with dots that are too small to print properly or too small to expose, and while it isnt a huge deal for those to be on the film, I would prefer to knock them off completely if it is possible.  Anyone know how this might be done?

[Sbrem]

You are, I suspect, wanting the make the range of your halftones produce the best picture, for each color. I just was reading an article on getting the best grays from the Adobe website, it was a little dated, but the info is still true. It was titled, "Getting the best grays in Photoshop CS3". Most of us have trouble with highlights burning out, and shadows filling in so that the entire image prints dark, and the highlights are gone. You want to control the dynamic range so that the brightest highlight with detail (2% - 5%) not a specular highlight light the sun off a chrome bumper, prints a dot, for t's 5% is fine. You also need to control the shadow, so that the darkest part of the image still shows some detail in the shadows. Offset likes 95% for that, screening on t's is more like 80% or 85%, depending on what you hold. Once you have your individual channels setup that way, you can then still halftone them using the Bitmap method in Photoshop; btw, when I use the Bitmap method, it may be 300 in, but for the output resolution, I use 1200, because it makes the dots smoother. Check out that article...

Steve

[DDSol]

I agree totally that 300dpi for halftones is way too low. Double it, at least. 1200 is very high res and may slow down your system, but it's definitely recommended. Remember to select "Nearest Neighbor" when resizing the image or you get (slightly) jaggy edges from halftoning areas that should not have any 'gradients' (like hard lines).

As far as choking your underbase: The choke will be much smoother at a higher resolution, even though you'd have to choke by more than 1 px.

Besides that, there may be a way to get rid of everything smaller than x pixels, but it's very complicated and time consuming, involving a lot of use of the 'min' and 'max' functions.

First, turn your image grayscale (or layers and filters will not work). Then dupe the white base layer, run a 'max' filter (Filter->Other->Maximum) on the dupe of x pixels, where x is the minimum radius of a dot, in pixels, set the blend mode of the layer to 'Liner Dodge (Add)' then run a 'min' filter on it of x*1.5 px (or so). To control it, make sure you zoom to 'Actual Pixels' (Ctrl-1) and are looking at an area where it transitions from 'all dots removed' to 'dots just visible'. Increase the radius until all your 'good' dots are looking round/elliptical again.

Finally, turn it back into a bitmap, which will flatten your art. Use 50% threshold.

All your small dots are now removed.

Note: You can use the 'opposite' to remove small pinholes as well:
-Grayscale
-Dupe
-Maximum x radius
-Set blend to Multiply
-Minimum x*1.5 radius
-Mode->bitmap

That said, it's not the best or cleanest way to get rid of small dots and pinholes. A better way to do it is to prevent them from occurring in the first place:

Select white
Select->Color Range
  Select: Sampled colors
  Pick the white you selected
  Adjust the range to select all the areas that are too light and should be white
  Ok
Select->Edit in Quick Mask Mode
Image->Adjustments->Levels
  Move the 'gray tick' in the center 'way to the left'. Look at the image to see what you're selecting.
  Put the black and white ticks close together and see what you're selecting. This area will have no dots.
  Ok
Select->Edit in Quick Mask Mode (turn off)
Hit delete
  Contents:Use: White
  Ok
Select->Deselect

You just selected the lightest areas and made them white. Dots will not be generated for these areas, and you won't have to clean them up.

Again, you can do a similar thing for pinhole removal:
-Color range
-Quick mask edit on
-Adjust levels
-Quick mask edit off
-Delete, make black
-Deselect

Underbase choking:
You should not choke a halftoned underbase, just like you should not do small dot removal, unless you Only have the Halftoned version.

Really, you should adjust the levels to do choking. Just make it slightly brighter.
Image->Adjustments->Levels
  Input levels: 0 - 1.0 - 235
  Output levels: 10 - 255

Adjust for your needs, but the point is that the levels work like this: The left input level is change to the left output level, so all 0's become 10's in the above example. So all blacks become very dark gray. Similarly, 235 and up will become 255, which means everything light and lighter will become white. NOTE that you should do this before small dot removal and pinhole removal. As a matter of fact, it Creates pinholes and small dots (blacks become dark gray=pinholes and light colors become very light colors=tiny dots).

Play with it a little . You'll find you can control the dots just fine without a RIP. As a matter of fact, you'll probably do a better job .

[jsheridan]

  I was wondering if there was a way to manually linearize and adjust for dot gain using photoshop instead of having to buy a rip. 

you need a RIP to linearize (one that will allow this info to be entered, some don't allow this) as that is the driver that controls how the printer prints.

You will also need a densitometer to read the dots printed on the film. These readings are then fed into a RIP so it can control the amount of ink/dye applied to the film to achieve a calibrated dot.

IE.. we linearize so that a dot is actually printed at it's specified tonal range and so on all the way to 100%.

Once you have your dot linearized, then you can go into PS settings and manually adjust for dot gain on each channel. This is a tricky setting as you have to account for smash from prior plates. To get the best results here, you'll need to do running testing from job to job until you determine the amount of dot gain your printing process produces and adjust for that.

[mimosatexas]

DDSol: exactly the kind of post I was hoping for, and the reason this forum is invaluable.  Thank you!  I will be testing this out on my next job to see how it all works in practice.  I figured there was a way, I just wanted to see how others would approach the process.

In response to the comments about resolution for output, I almost always work at 600 dpi or 1200 dpi when working with halftones, but most spot stuff I do at 300 and it comes into the shop at 300.  I just mentioned 300 as it is a pretty standard resolution for a lot of art and the 1 px at 300dpi is how I choke spot underbases usually.

[Dottonedan]

I love good responses like this!  LOVE them.

I'm in Sbrems camp for the most part. The other two guys have stellar info as well.

DDSol is showing me new methods and I'm intrigued but I fear much of that is either overkill or a bit too long of a method for me. Im sure it's much faster once you get the hang of the process like most of this is.

I also agree with John fundamentally. On a side note while on this topic, I had recently been at a shop in Mich that did excellent some process work. Award winning to say the least. Great full color people photos.
None of them were linerized. In fact, they never heard of it. Their films were from an Epson and did no adjustments in their rip. How do I know? Because I had to disect their process and match ours to theirs si they could compare apples to apples.  I'm sure they compensate in the separation process by experience but they didn't know if the 40 was 40 or 60. They just knew what to do to get it to look right.

Now I agree that this was not the best way to deal with dots, but it worked for them and for years, I avoided most if the technical gadgets like densatimitors just for that reason. I did it all manually and got good results.

Today, i know its a must have simply for the reason that others don't always know what they don't know so it's a good idea to show them with a densitoniter.

[Dottonedan]

I forgot to mention, I would use 2 pixels at 300 Rez.

I also use the same Rez as your film printer is set at like 1200x
 Or 600x for the file Rez. They produce cleaner dots.

Also, the photoshop bitmap line screen (count) is resolution dependent,  this means that it's like a box of chocolates. You never know what you're going to get. Your 55 could be 49 at 300 and could be 61 at 1200 ppi.  Still tho, it's at least close.

[mimosatexas]

interesting note on the line count being rez dependent.  That doesnt seem logical as dpi is independent from actual size in inches, which is what the lpi should be referencing for output.  Any idea why it would change?

[Colin]

The math in photoshop is wonky... why, I dunno.

Make a 25% fill in a one inch square box.

Bitmap it to 50 lpi halftone at 0 or 90 degrees at 300/600/900/1200 dpi.

You will see the line count increase as the resolution goes up...... yea, it's wonky.

[Sbrem]

There's a lot to gobble up here I see, thanks DD; now I just have to find some time... and as Mimo says, what a great forum this is... and lastly, we use AccuRIP, and are very happy with it. It lets you fine tune your ranges, as said before, so you get the dot on the film that represents the tone you're trying to create. This is strictly to control what the printer prints, so that what you send comes out correctly on the film. What happens in the screen room and on press is 2 more whole subjects.

Steve

[blue moon]

save your pennies and get a RIP as soon as you can afford it. You'll wonder how you manged without it once you get it.

pierre