Imagesetter vs Inkjet comparison

Here is a question for all you techincal types.
Imagesetters will print at a resolution of up to 3600, possibly 4000 dpi. My Epson 4000 prints at a resolution of 2880 dpi. But 2880 is the resolution for each individual ink, isn't it. If I use Quadtone RIP to make the printer print with nearly equal amounts of all seven inks, and if each ink is printing with a resolution of 2880 dpi, how many effective dots per inch am I getting? 2880x7 ???? If this line of thinking is correct, it would seem that a modern inkjet printer would blow any image setter out of the water, in terms of resolution? What am I missing?
Cheers, Ron-san

I can't speak to the technical reasons, only the results. I have done a lot of imagesetter negatives, and even an imagesetter neg output at 2400dpi prints sharper than a 2880 inkjet neg. Whether this is due to microscopic ink spreading or what, I can't say. But the imagesetter negs have more print acuity than the inkjet versions printed from the same file. But they cost a whole lot more.

clay said:

I can't speak to the technical reasons, only the results. I have done a lot of imagesetter negatives, and even an imagesetter neg output at 2400dpi prints sharper than a 2880 inkjet neg. Whether this is due to microscopic ink spreading or what, I can't say. But the imagesetter negs have more print acuity than the inkjet versions printed from the same file. But they cost a whole lot more.

Click to expand...

Clay-- Actually, I am not much worried about sharpness or acuity. I am wondering how they compare in tonal smoothness. Would an inkjet printer printing with all seven inks show less dot pattern than an imagesetter printing with one ink?? I have never used an imagesetter and have no reference for comparison. Thanks, Ron-san

As far as smoothness goes - with the imagesetter negatives a lot depends on the curve you use and the operator of the machine. I used one outfit for years and the Fokos curve and I got beautiful results. Then that operator left, and I got some pretty substandard results. In fact, this episode has made me far more interested in getting the inkjet negatives to the same point that I had the imagesetter negs.

My intution is that a properly curved inkjet negative will probably be smoother in the highlight tones than an imagesetter. But that is pure seat-of-the-pants SWAG at this point. I have not done a direct comparison with your quadtone negs and an imagesetter neg of the same image. Need to create an eight day week.

FWIW, I used that 2200 curve you downloaded to the site here and I got some pretty good results. Actually results so good that I question whether I ought to try refining them or just use them as-is and make a bunch of prints. I can't wait to see the book.

Ron-san said:

Here is a question for all you techincal types.
Imagesetters will print at a resolution of up to 3600, possibly 4000 dpi. My Epson 4000 prints at a resolution of 2880 dpi. But 2880 is the resolution for each individual ink, isn't it. If I use Quadtone RIP to make the printer print with nearly equal amounts of all seven inks, and if each ink is printing with a resolution of 2880 dpi, how many effective dots per inch am I getting? 2880x7 ???? If this line of thinking is correct, it would seem that a modern inkjet printer would blow any image setter out of the water, in terms of resolution? What am I missing?
Cheers, Ron-san

Click to expand...

But how many dots are required for a pixel if you are printing something that isn't just black or white? The true resolution might be a tenth of the nominal dpi. That's why a Lightjet at 400 ppi can look sharper than an inkjet at 2880 tpi (thingybobs per inch). It's easy enough to check - just print a test pattern with features one pixel wide and see how well it prints at different dpi settings. You might find different optimum values for QTR vs the Epson driver, and you might find that the horizontal and vertical optima are different with QTR - ie one value gives the best horizontal resolution and another value gives the best vertical resolution. QTR and the Epson driver handle the issue in different ways.

Best,
Helen

Ron-san said:

Here is a question for all you techincal types.
Imagesetters will print at a resolution of up to 3600, possibly 4000 dpi. My Epson 4000 prints at a resolution of 2880 dpi. But 2880 is the resolution for each individual ink, isn't it. If I use Quadtone RIP to make the printer print with nearly equal amounts of all seven inks, and if each ink is printing with a resolution of 2880 dpi, how many effective dots per inch am I getting? 2880x7 ???? If this line of thinking is correct, it would seem that a modern inkjet printer would blow any image setter out of the water, in terms of resolution? What am I missing?
Cheers, Ron-san

Click to expand...

I don't think theory and reality are going to always be the same when comparing output methods. From what I have seen image setters produce sharper looking negatives than inkjets. Continuous tone film recorders capture much more fine detail than either, but may not look sharper than an image setter.

Keep in mind that an image setter's resolution is for only black and white dots. Thus to get 256 tones you need a 16x16 square of pixels, so your real resolution for a 3600dpi image setter is 3600/16 or 225ppi (image pixels per inch). Inkjet printers are similar and usually have real resolutions of 360ppi or maybe 720ppi. Continuous tone devices such as film recorders, lambdas, dye sub printers and LED printers have dpi measurements that directly correspond to ppi.

L Gebhardt said:

I don't think theory and reality are going to always be the same when comparing output methods. From what I have seen image setters produce sharper looking negatives than inkjets. Continuous tone film recorders capture much more fine detail than either, but may not look sharper than an image setter.

Keep in mind that an image setter's resolution is for only black and white dots. Thus to get 256 tones you need a 16x16 square of pixels, so your real resolution for a 3600dpi image setter is 3600/16 or 225ppi (image pixels per inch). Inkjet printers are similar and usually have real resolutions of 360ppi or maybe 720ppi. Continuous tone devices such as film recorders, lambdas, dye sub printers and LED printers have dpi measurements that directly correspond to ppi.

Click to expand...

Thanks to all for the replies. I thought my way of looking at the matter was way to simplistic. Ron-san

I am interested in this very thing, I would like to make very large separation negatives on FP4 or equivalent film for alt printing.
I think the film from the lambda will be very good for this purpose. Right now I image at 400ppi and when making a 30x40 cibachrome*very high gloss* louping the print only shows me film grain and not a pixelation that I suspected.
Therefore I believe that if one prints onto a rag paper with platinum and maybe a colour gum over hit the results would be nothing short of spectacular.
At this point I am gathering my pennies to buy a 30inch roll of film from Ilford or Jand&C and hope to see if my hunch is correct.

I have tried both imagesetter and inkjet negatives.

Imagesetter negatives:
You need to find a good imagesetter lab - many don't use the newer RIPs to get super high resolutions - and most are not consistent enough to get reliable results.
Expensive per negative - about $50 for a 20" x 30" neg. - but you can group negatives.
Sizes go up as large as 30"x40" (or maybe larger)
Takes time - you need to drop of the files and come back to pick up the negatives.
Imagesetter negatives are really just clear film and black dots - at very high resolutions. These work fine, but they don't work like traditional negatives (that have grey values). It can be diffucult to tweak your prints - but adding a but more time for example.
I think these are best for production type work. If you have a specific process that you want to print professionally, you can make a set of negatives and then print away on these.
Printing onto silver gelatin produces noticable dots under a loop.

Inkjet negatives:
These can be very sharp.
You can print negatives right at your desktop.
Your prints are limited by the size of your printer.
It is easy to tweak your prints and experiment.
I'd strongly recommend using inkjet negatives to start - and likely you'll never want to go to using an imagesetter.

Hope this is useful

It would seem that you could layer image setter produced negs which would conceivably increase the tonal range (like a duo tone). it would also seem likely that you could do this with stochastic patterns which would help alleviate registration issues and further increase the tonal range and the ability to layer more negs.

Just a thought...

I've had the same thought: An inkjet printer can produce 255 distinct tones. ---The imagesetter the same? They say the best lithographers can detect up to 2000? Part of what we may be seeing in an image printed from a digital negative that qualitatively differentiates it from one originating from an in-camera negative is this limitation in tonal range. I have seen where Mark Nelson achieved better tonal range in a print by limiting the range within the image: an exclusively high key image, for example. Why not allocate the full tonal spectrum of an image among two or more negatives, exposed seperately in registration? I'd love to hear from anyone who has experimented with this.

Ron-san said:

Here is a question for all you techincal types.
Imagesetters will print at a resolution of up to 3600, possibly 4000 dpi. My Epson 4000 prints at a resolution of 2880 dpi. But 2880 is the resolution for each individual ink, isn't it. If I use Quadtone RIP to make the printer print with nearly equal amounts of all seven inks, and if each ink is printing with a resolution of 2880 dpi, how many effective dots per inch am I getting? 2880x7 ???? If this line of thinking is correct, it would seem that a modern inkjet printer would blow any image setter out of the water, in terms of resolution? What am I missing?
Cheers, Ron-san

Click to expand...

You're missing alignment, the bane of inkjets, where different colors don't align, leaving "holes" in the dither.

Also that using near equal amounts of all inks will cause the image to be a mosaic of different inks, each having a different density, and will look incredibly grainy.

QTR negs look best when you lay down a lot of ink of a single color, moving from color to color to increase density. You start with the least dense color and run it up to the ink limit before crossfading to the next least dense color, etc, across the band. So a good curve is never using more than two colors at any one density.

wiz said:

You're missing alignment, the bane of inkjets, where different colors don't align, leaving "holes" in the dither.

Also that using near equal amounts of all inks will cause the image to be a mosaic of different inks, each having a different density, and will look incredibly grainy.

QTR negs look best when you lay down a lot of ink of a single color, moving from color to color to increase density. You start with the least dense color and run it up to the ink limit before crossfading to the next least dense color, etc, across the band. So a good curve is never using more than two colors at any one density.

Click to expand...

Wiz-- I have enjoyed reading your posts. It is good to hear from someone who obviously knows a lot about the Quadtone RIP. I myself know just enough to get in trouble.
I understand your point that a good curve uses one ink at a time, up to its max density, before cross fading to the next higher density. In my case, I did something a bit in between. I used the light inks (LK,LM,LC) all at the same time and then cross faded them into the dark inks (K,M,C,Y) which then all printed at the same time up to the specified default ink limit. This approach did not seem to increase graininess. What it did was eliminate an annoying banding (about 1/2 inch periodicity) that was ruining any area of smooth tone when I made negs using just K and LK (which was how I first started out). I have been very happy with negs printed by this all color approach.
My only problem right now is another banding artifact. In areas of even light tone I see a faint but obnoxious banding with a periodicity of about1 1/2 to 2 inches. I see it when printing with QTR or with the Epson driver so my suspicion is that it has to do with my Epson 4000 itself. Have not yet found out if such banding shows up on other printers or if it is my own little problem. I actually do not see it when printing positive paper prints, only when printing negs on Pictorico.
Keep up the good posts. Cheers, Ron-san

jd callow said:

It would seem that you could layer image setter produced negs which would conceivably increase the tonal range (like a duo tone). it would also seem likely that you could do this with stochastic patterns which would help alleviate registration issues and further increase the tonal range and the ability to layer more negs.

Just a thought...

Click to expand...

It's a very good thought. I've been doing that with my Epson R2400 negatives. One sheet for the negative and one, in registration, a very light positive. In works along the lines of a traditional pin-registered unsharp mask or constrast mask. I'm having good results on my homemade silver gelatin emultion. Also, I rotate one 90 degrees to the other to further dither out printer lines.