How? Creating a ruled line screen of high precision, 1000 lp/in

[holmburgers]

I need to create high resolution black & white stripe patterns that can be saved as TIFF files; to ultimately print out as transparencies using a laser recorder. The patterns are simply series of parallel lines; having maximum density and zero density.

The trick is that it needs to be quite fine... like 1000 lines per inch. To make an 8x10" film with this pattern, well.... let's just say I'd like to avoid hitting *paste* 10,000 times. I need a way to create these line screens more automatically.

The proportion of black to white, opaque:transparent, is important and should be variable. For instance, within a given unit distance, having a black line that is twice as wide as the transparent portion (1/3 is clear), and alternatively, having equal areas of black & transparent (1/2 & 1/2).

Assuming that this kind of thing is computationally rather simple, how about the ability to make these lines wavy; following a sinusoidal function?

I hope this isn't hopelessly obtuse. There are a number of reasons for such a screen, one of which is in the making of a parallax parallelogram, or a diffraction color photograph, but I do have a specific and ulterior idea in mind. . . .

Conceptually this is a rather menial task, but when it comes to actually creating this kind of file I have no idea where to begin. There are a number of "stripe generators" online that, although not up to this task, suggest some savvy programmer might be able eat this for lunch.

 

[TheFlyingCamera]

Are you using Photoshop CS4 or newer? You could try to script this as an action and then automate it. Or create a background pattern of a 4px x 4px square, have two rows of 4 px be black, two rows white, and then use it as a fill on a 1000ppi file. I've not done either before, but the theory is there for you to try.

 

[holmburgers]

Sadly, I'm not using Photoshop... though I do have friends with it. Not sure what version it is. That being said, focusing on PS techniques is probably wise considering it's the 'standard'.

This scripting thing seems very promising, thanks for the tip.

 

[gmikol]

Or create a background pattern of a 4px x 4px square, have two rows of 4 px be black, two rows white, and then use it as a fill on a 1000ppi file. I've not done either before, but the theory is there for you to try.

I've done this on a much smaller scale (1" x 1") , and it works great and is really easy in PS.

While I use PS, might one of the other commercial image editing programs do this (Paint Shop Pro, as an example), or even GIMP (free/open source) be able to do this as well?

I may also have a fragment of PostScript code that could do this (bar patterns only), but I'd have to dig it up first.

One clarification on what you mean by "laser recorder"...do you mean film recorder (full-color, pseudo-continuous tone), or do you mean imagesetter (litho film, on or off dots only).

My brief experience with using film recorder output is that they can't necessarily image their full stated resolution. i.e. a 4000 PPI film recorder might only be able to do 2000 PPI of real data. Be sure to test on a smaller frame to make sure it's doing what you want. Last time I checked, 8x10 film recorder output could run $80 a sheet or more. Imagesetter output might be less expensive. Make sure the operator understands what you're trying to accomplish. Maybe have them try a litho film instead of a con-tone film if you just want on-off patters. There may be dot gain issues. You'll achieve best results using an integer divisor of the native resolution of the device. In the 4000 PPI example, that means 2000 (div by 2), 1333.33 (div by 3), 1000 (div by 4), etc. Any other resolutions and interpolation will seriously mess up your bar patterns.

Sinusoidal patterns might be more difficult, as you probably want at least 4 pixels per cycle (if not many more), so a 1000 cycle/inch pattern would require really good 4000 PPI output. Sinusoidal output would need something a little more computationally intensive than simple bar patterns.

Lastly, I feel like I may have mentioned this to holmbugers before, so I apologize if this is repetitive, but you may want to use "ronchi ruling" as a search term. There are some scripts out there that generate that sort of thing for people who build their own telescopes. I think they use them for focus and alignment testing. Anyway, that's were I found the postscript code I mentioned earlier.

Hope that helps some. Good luck.

--Greg

 

[holmburgers]

Hi Greg, thanks for the thorough reply.

I totally forgot about our past discussion of Ronchi rulings, but indeed they'd be perfect for this. However, they're absurdly expensive!!

But, it's a lead, and it's nice to know what might be possible someday.

Well I'm going to look into this scripting biz, and hopefully will come back sometime with some more questions about this. For testing, I might just make a crude set of screens; afterall there's a lot I need to figure out before I scale it down.

The laser recorder I'm referring to, which indeed might not be the right term, is I believe a Durst Lambda. Actually, I PM'd Bob Carnie about this because I know he works with such machines.

A Durst Lambda catalog states, Exceptional image quality at 4000 dpi apparent resolution with no visible dots, but then also says, The Durst Lambda exposes digital information (raster pixel) directly to conventional photographic media at full continuous tone with a linear writing speed of up to 30 cm / 12" per minute with the coice of two resolutions of 200 and 400 ppi (equal to an apparent resolution of 400 dpi) and again, The 400 ppi continuous-tone resolution of the Durst Lambda 130 is comparable with a resolution of 4000 dpi of printers with half-tone dots.

I'm not sure what exactly this means, but it doesn't sound promising...

I was thinking a litho film and a high contrast developer so we can get maximum density and complete transparency. As for sine waves, well I see what you mean..

Very helpful guys, thanks a ton!

 

[gmikol]

Yeah...I think you've figured it out. Lambda's 400 PPI optical output is equivalent (as claimed by Durst) to 4000 dpi halftone screen.

So you're only gonna get 400 PPI out of the lambda...optimistically, that's 200 lp/in, but realistically, probably only 100 lp/in.

Is this for Lippmann-type experiments, or something else?

Also, how many of these do you anticipate needing to make?

--Greg

 

[holmburgers]

Back to the drawing board.

So what are the other options out there? Maybe an image setter is actually the ideal tool for this. If they're able to reach 1200 and 4800 DPI, then that should be more than enough for straight lines.

How about J. Browning's laser recorder? "The recorder both scans the originals (Up to 150 l/mm) and records onto 8x10 film."

If my math is serving me, that's 3810 l/in., or nearly 2000 lp/in. Again, not bad.