Epson 3800 Digital Neg Success

sanking · Mar 20, 2007

Kerik said:
Clay - Did you add the 4 points of green to your B&W RGB neg? I didn't do that on the step wedge I made and got boo-koo density.

Are you seeing any increase in grainess when adding the 4 points of green? There is a lot of yellow in green and Clay mentions in another post that this is the biggest UV blocker that may contribute to grain.

Also, are you having any problem with pizza wheel marks? I noticed some pizza wheel marks in printing B&W negatives in RGB with Don Hutton's 3800. I would have purchased already but am concerned about how fast the inks dry, though I understand there is some kind of slow-down mechanism in the printer driver?

Sandy

clay · Mar 20, 2007

I don't think there is enough yellow being added with Gr4 to make any difference. In fact, the QTR profile that is working so well has some yellow in the very upper end of the scale just to add a little boost to the density. I think the problem comes in when you are using just the yellow and cyan throughout the whole scale and not dithering it with some of the other colors and the blacks. My theory is that when you plot the UV densities of yellow and cyan, they diverge quite a bit in the higher ink percentages. When the epson driver has just two colors to dither, you get a large UV transmission density contrast between adjacent dots of yellow and cyan. This is what I think can lead to the graininess reported in the emerald green negatives. However, the difference between yellow and PK is not as great, so the density difference at the same ink percentage is not as pronounced, so there should be (I think) less propensity for this grainy effect. I have gone blind looking at these negatives under a loupe the last two weeks, and I am now convinced that intelligent mixing of all the colors and minimization of yellow is the key to a smooth negative. The reason that Kerik's getting such smooth results is because he is converting a monochrome file to RGB, and the Epson driver is throwing the whole ink range into the mix. I got one of Kerik's prints in the mail yesterday, and it is SMOOOOOOOOVE. As are the ones I'm getting from the QTR approach.

sanking said:
Are you seeing any increase in grainess when adding the 4 points of green? There is a lot of yellow in green and Clay mentions in another post that this is the biggest UV blocker that may contribute to grain.

Also, are you having any problem with pizza wheel marks? I noticed some pizza wheel marks in printing B&W negatives in RGB with Don Hutton's 3800. I would have purchased already but am concerned about how fast the inks dry, though I understand there is some kind of slow-down mechanism in the printer driver?

Sandy

sanking · Mar 20, 2007

clay said:
I don't think there is enough yellow being added with Gr4 to make any difference. In fact, the QTR profile that is working so well has some yellow in the very upper end of the scale just to add a little boost to the density. I think the problem comes in when you are using just the yellow and cyan throughout the whole scale and not dithering it with some of the other colors and the blacks. My theory is that when you plot the UV densities of yellow and cyan, they diverge quite a bit in the higher ink percentages. When the epson driver has just two colors to dither, you get a large UV transmission density contrast between adjacent dots of yellow and cyan. This is what I think can lead to the graininess reported in the emerald green negatives. However, the difference between yellow and PK is not as great, so the density difference at the same ink percentage is not as pronounced, so there should be (I think) less propensity for this grainy effect. I have gone blind looking at these negatives under a loupe the last two weeks, and I am now convinced that intelligent mixing of all the colors and minimization of yellow is the key to a smooth negative. The reason that Kerik's getting such smooth results is because he is converting a monochrome file to RGB, and the Epson driver is throwing the whole ink range into the mix. I got one of Kerik's prints in the mail yesterday, and it is SMOOOOOOOOVE. As are the ones I'm getting from the QTR approach.

Well, based on Kerik's reports on how smooth the results are I have decided that the 3800 is the one.

BTW, it is available on some sites now for about $1100, delivered.

Sandy

Kerik · Mar 20, 2007

sanking said:
Well, based on Kerik's reports on how smooth the results are I have decided that the 3800 is the one.

BTW, it is available on some sites now for about $1100, delivered.

Sandy

Now, if I can just get Epson to throw me a few bucks every time one of these sells for digital negs... That's a good price! I paid about $1,250 delivered a few weeks ago, and had to shop hard to find that price.

BTW, as Clay said, the 4 bits of green adds no perceptible grain. And I'm not even sure if it's necessary once I get the QTR thingy figured out.

I've not noticed any pizza wheel marks. If they're there, they definitely do not show up in the prints.

sanking · Mar 20, 2007

Kerik said:
Now, if I can just get Epson to throw me a few bucks every time one of these sells for digital negs... That's a good price! I paid about $1,250 delivered a few weeks ago, and had to shop hard to find that price.

BTW, as Clay said, the 4 bits of green adds no perceptible grain. And I'm not even sure if it's necessary once I get the QTR thingy figured out.

I've not noticed any pizza wheel marks. If they're there, they definitely do not show up in the prints.

Well, as soon as I get mine I am going to have some questions about QTR because I suspect that this is the best direction for me to take with this printer, given the fact that I am primarily intrerested in digital negatives and making monochrome prints.

I think the Red ink array with PDN is still an option, however, as results on that track are also extremely smooth. Just not sure it is possible to lay down enough ink, howeve, to get the density I need without the Pizza wheel marks.

Sandy

Anonymous · Mar 20, 2007

sanking said:
Well, based on Kerik's reports on how smooth the results are I have decided that the 3800 is the one.

BTW, it is available on some sites now for about $1100, delivered.

Sandy

Sandy, Where did you find that price?

Michael Mutmansky · Mar 20, 2007

Kerik is going to have to rebuild that 7800 profile to get properly adjusted curves from the 3800.

Even though the inks are the same, the method through which the printer lays down the ink is apparently very different with the 3800 compared to the other x800 printers. The result is that the profiles that have been set for a 4800 or 7800, will probably not work in a linear manner. They'll work, though, but not in an ideal manner.

In particular, it will be necessary to determine the ratio of UV densities of the various gray inks relative to each other to confirm whether or not they have the same laydown characteristics. I don't believe they do.

If they are the same, it my be possible to simply linearize the profile and move on, but otherwise, it will require adjusting the crossover points first, and then doing the linearization.

Regardless, it's a simple procedure with a UV densitometer, and a reflection densitometer for the linearization portion.

---Michael

Kerik · Mar 20, 2007

Michael,

I've just started using QTR, so I'm not tracking with your lingo yet. I have a UV transmission densitometer, but no reflection densitometer. Can I get along without it, or should I start trolling ebay for a cheap reflection unit?

Michael Mutmansky · Mar 20, 2007

Kerik,

You can build one to your liking without a reflection densitometer, but I suspect it will be a good bit easier if you used one. The main reason is that the print density is supposed to correlate to your monitor, and if you build one that you like, it may not really correlate to your monitor close enough to always have the confidence that the negative is going to produce the desired result.

Some people are using a scanner for this, but there are inherent problems with this approach that make it good enough for some, but won't be an acceptable result for others. A densitometer should eliminate that variable.

I picked up an X_Rite 400 a few years ago for about $80. Works great for this application.

---Michael

sanking · Mar 20, 2007

Donald Miller said:
Sandy, Where did you find that price?

Donald,

The place I bought mine no longer has them in stock, but buy.com has the 3800 for about $1150, including shipping.

Sandy

Anonymous · Mar 20, 2007

Thanks Sandy, that is a very good price.

sanking · Mar 20, 2007

clay said:
The reason that Kerik's getting such smooth results is because he is converting a monochrome file to RGB, and the Epson driver is throwing the whole ink range into the mix. I got one of Kerik's prints in the mail yesterday, and it is SMOOOOOOOOVE. As are the ones I'm getting from the QTR approach.

If you get smooth results from digital negatives with the 3800 printing a monochrome file in RGB then there should be little else one need to do if the goal is printing with pure palladium except adjust the process curve. My measurment of a 101 step wedge printed this way shows that the DR is about 2.0, almost perfect for a pure palladium mix, and very close to what I also use in carbon and kallitype. So if the results are already very smooth and you have enough DR why dicker around any more with the colors?

Is there any reason in theory why a QTR should print any smoother? I would think that the way the inks are laid down is pretty similar with a QTR as when printing a desaturated RGB file in color.

Sandy King

clay · Mar 20, 2007

It is probably splitting hairs, agreed. The main reason is that you have access to another dithering algorithm (four of them in fact) that to at 20x may just possibly be a bit smoother, and you can dispense with curves altogether once you get it dialed in. Just invert, flip and print.

sanking said:
If you get smooth results from digital negatives with the 3800 printing a monochrome file in RGB then there should be little else one need to do if the goal is printing with pure palladium except adjust the process curve. My measurment of a 101 step wedge printed this way shows that the DR is about 2.0, almost perfect for a pure palladium mix, and very close to what I also use in carbon and kallitype. So if the results are already very smooth and you have enough DR why dicker around any more with the colors?

Is there any reason in theory why a QTR should print any smoother? I would think that the way the inks are laid down is pretty similar with a QTR as when printing a desaturated RGB file in color.

Sandy King

Michael Mutmansky · Mar 20, 2007

sanking said:
Is there any reason in theory why a QTR should print any smoother? I would think that the way the inks are laid down is pretty similar with a QTR as when printing a desaturated RGB file in color.

Yes, because the ABW mode uses yellow ink. QTR permits you to avoid yellow until high density.

---Michael

Michael Mutmansky · Mar 20, 2007

I would also suggest that you need about 2.3 or so to ensure you have clean highlights with pure palladium and no restrainer. 2.0 is simply too low for that.

--Michael

sanking · Mar 20, 2007

clay said:
It is probably splitting hairs, agreed. The main reason is that you have access to another dithering algorithm (four of them in fact) that to at 20x may just possibly be a bit smoother, and you can dispense with curves altogether once you get it dialed in. Just invert, flip and print.

I can see that an extra dithering pattern may help to smooth things out. But how do you dial it in without at some point doing reflection readings? Once I get to that point adjusting the curve is a piece of cake.

Sandy

sanking · Mar 20, 2007

michael mutmansky said:
I would also suggest that you need about 2.3 or so to ensure you have clean highlights with pure palladium and no restrainer. 2.0 is simply too low for that.

--Michael

I agree with that since I always use a tad of dichromate in the developer with a DR of 1.9.

But if you push the DR to 2.3, is the result still smooth?

Just interested if anyone has information as to maximum UV transmission with normal lay down of each of the inks in the 3800? From previous messages I understand that the QTR can plot that information?

Sandy

Michael Mutmansky · Mar 20, 2007

Sandy,

It's not a push to go up to 2.3. It's possible to get 3.0 out of QTR with the right blends of inks, if I remember correctly. The higher you go, the more yellow you have to use, in general, but by that time, you are using other comparable high density inks as well, so there isn't a grit problem.

As I see it, the one shortcoming of the PDN approach is the uncontrolled mixing of high density inks with low density inks that occurs in the basic Epson driver. Even though the inks visually look like a wash of color, they act like film grain in that each droplet has an inherent density that as a composite makes the desired density, but individually, they may interact in a visible way if care isn't taken to avoid undesirable interactions.

QTR (or any RIP, for that matter, this isn't a new idea, really) allows you to do just that kind of manipulation.

---Michael

chrisf · Mar 20, 2007

Sandy,

Will you be testing negatives made with this printer for contact silver process?

chris

sanking · Mar 20, 2007

chrisf said:
Sandy,

Will you be testing negatives made with this printer for contact silver process?

chris

If I can get tones with AZO as smooth as with contact printing ULF silver negatives I will eventually use the 3800 for an AZO printing project. As smooth as ULF negative would mean that there is no visible dithering pattern on the smooth paper after it dries. I will probably test the premise soon after I get the printer.

I don't think I will be doing any testing with VC silver papers. This is not a direction I want to go with my own work and I think that kind of testing could be better done by someone with more interest in silver.

Sandy

chrisf · Mar 20, 2007

sanking said:
If I can get tones with AZO as smooth as with contact printing ULF silver negatives I will eventually use the 3800 for an AZO printing project. As smooth as ULF negative would mean that there is no visible dithering pattern on the smooth paper after it dries. I will probably test the premise soon after I get the printer.

Sandy

This is what I'm interested in too.

Thanks

clay · Mar 20, 2007

When I say 'dial it in', I mean that you have the curve calibrated properly. You still have to do some of the same type of reflective density step tablet measurements that you do with any system.

sanking said:
I can see that an extra dithering pattern may help to smooth things out. But how do you dial it in without at some point doing reflection readings? Once I get to that point adjusting the curve is a piece of cake.

Sandy

bobherbst · Mar 29, 2007

Opacity for color layer

Kerik, Clay, Michael, et all,

What opacity percentage are you using in the colorizing layer in your b&w RGB negs? I've found 50% opacity using green inks yields extremely high densities on the 2400.

As an aside, I had the opportunity to work with the 9800 and 4800 last June while teaching at the Cleveland Institute of Art. I purchased the 2400 because a 4800 wasn't in the budget and the 3800 was not yet out. I found that adjustment curves only require very minor tweaks between the 4800 and 9800 model printers and even the 2400. Now moving from a MAC to a PC environment is a whole different story because of differences in the drivers I assume. But that's my own burden in teaching at different locations.

I have also found that visual grain in the print can be as much a function of the scan and photoshop manipulations for a given color used for the digital negative. Some files can be sharpened with no effect on the image grain. Some require a small amount of gaussian blur to minimize the grain appearance, so I don't think grain is purely a function of the color used for the digital negative. But like everyone else, I'm still experimenting with mixed results when compared next to a contact print from an in-camera negative.

Kerik · Mar 29, 2007

Bob-

A few things... I'm no longer using a color layer. The very slight 4 points of green aren't really necessary for my RGB B&W negs.

The grain I was experiencing was absolutely the result of the green/red negs I started with. There was no scan involved. I was printing a step tablet created from scratch in Photoshop. I agree with what you're saying about sharpening increasing apparent grain, etc., but that variable was eliminated from my initial step tablet testing that showed grain. When I started printing from an actual neg, I used the same file repeatedly and only changed colorization, curve, etc. but no changes to size, sharpening, etc.

BillSchwab · Jun 7, 2007

Just placed an Epson 9800 into service today. What a sweet machine! Just ran my first digi-neg using PDN and a sample sheet of Inkpress Clear Film. What a joy. I can already tell it is far smoother than those done with my 2200. Also made the first 44 inch square print today. What a monster... beautiful..Bill

Epson 3800 Digital Neg Success

Privacy & Transparency

Privacy & Transparency