what resolution?

[RalphLambrecht]

my Epson printer driver offers different resolutions for inkjet printing. Does it make a difference when printing photographs or is there a preferred resolution?

 

[L Gebhardt]

With Epsons the native resolution is going to be a multiple of 360. Ideally you will resize your image to 360 or 720 ppi. That way the driver doesn't need to do any interpolation. In the driver set it to 1440 or 2880. In my testing on a 3880 I could just barely see a difference between 360 and 720ppi when printed at the 2880 setting, but I needed my nose to the print.

I usually print at 1440 and resize the image to 360ppi before sharpening simply because it's faster and no one I've shown prints to has been able to pick out a higher res version consistently.

 

[lenny]

Larry, this information has been disproven and its been directly contradicted by Epson. There is no need to do anything with multiple of 300, 360, 363, or anything else. There was just a thread about this on another forum and Andrew Rodney made this same statement...

I would suggest that anyone that wants to see this for themselves can size up an image to 40 or 60 inches, then crop to an 8x10 portion of it, print it, then size it down to a multiple of 360 and do it again. Then you can see for yourself if there is any difference.

I don't think there will be.

Lenny

 

[L Gebhardt]

Lenny, that's great to hear. Do you have a link to the other thread?

I'm not sure how sizing up an image can really help you with determining the optimal resolution for a printer. Wouldn't taking a very sharp image and printing it downsized to different resolutions so it has lots of real detail be a better choice?

 

[Doyle Thomas]

Larry, this information has been disproven and its been directly contradicted by Epson. There is no need to do anything with multiple of 300, 360, 363, or anything else. There was just a thread about this on another forum and Andrew Rodney made this same statement...

I would suggest that anyone that wants to see this for themselves can size up an image to 40 or 60 inches, then crop to an 8x10 portion of it, print it, then size it down to a multiple of 360 and do it again. Then you can see for yourself if there is any difference.

I don't think there will be.

Lenny

I have always done as Larry to avoid interpolation. It sounds as though you Lenny are saying interpolation is not an issue? You suggest resize, crop and print then set the resolution of the crop to 360 and print again as a test?

In my Epson testing I have found 360 to be best and 270 is better that 300 if you maintain the same dimensions. If you allow the Print size to go where it may, then there is no difference I can see.

 

[lenny]

Larry,

Sorry, been away...

Here's the thread: Inkjet print resolution - Photo.net Digital Darkroom Forum

I use a drum scanner. I am not interpolating up, so I am printing real pixels no matter what I send to the printer. If its downsized the pixels are just tighter together, which does make things look sharper...

Doyle, I am not saying anything about interpolation. I am saying that I do not believe that Epson printers will print any better at the actual resolution of the file, provided you have 300-360 pixels, at minimum - vs some multiple of 300 or 360. The total number of pixels is important, but not the multiples...

I doubt that 270 is better than 300. It doesn't make sense in a practical world. It might have worked for that image, there are all sorts of coating, temperature, humidity differences that could have resulted in a different look, provided they weren't printed one right after the other. It's easy enough to imagine you are seeing something. Of course, then if you take to more scientific methods, you end up printing out test patterns and then it doesn't make sense to real images either.

 

[Richard Boutwell]

I think what ralph was asking had more to do with the output resolution from the printer 720, 1440, 2880, and not the input resolution of the file. The print head only has 1440 actual nozzles per channel per inch, and, from my understanding, the 2880 resolution in the horizontal direction is achieved by the movement of the print head as it moves across the paper (you'll notice that it is 2880x1440).

Some papers might not be able to hold the actual dots when printed at the maximum resolution, and some dots might even end up in the "wrong" place on the paper, or overlap with other dots. In actuality, not all of the nozzles are firing for any given image pixel. The tone/color for a given image pixel, when printed, is a combination of multiple inks, dot sizes, the spacing of dots based on dithering algorithms, and overlaps with other inks. There are a whole lot of internal calculations that have nothing to do with the number of pixels coming into the driver.

With all that being said, I still want the printer to spit out as many dots as I can get. That is why I think digital black and white prints made with Cone inks are better than those made with UC k3 inks. There are more dots of more dilutions of grey, and therefore less visible dithering. Those prints look smoother. That might not be everyones criteria for "better" but it is certainly a part of the equation for me.

As for what resolution the input file should be: I don't know what the epson driver does when interpolating input resolution to something that it can then turn into dots on the paper. Based on what Roy Herrington has said about how QTR handles input resolution, in that all files are converted to 720x720 dots per inch, you have to assume that epson does something similar.

This is a quote from him from the QTR yahoo group:

"QTR resamples all input files to 720x720 pixels. Then dithering is done to whatever dpi you've selected -- mostly 2880x1440 or 1440x720. A couple printers also have 1440x1440.

"The resampling to 720x720 is done by the OS on the Mac so I'm not sure of the algorithm. On the PC it's a bi-linear algorithm -- basically two linear interpolations -- horizontal then vertical. As far as I know all the Epson drivers do either 360x360 or
720x720 resampling. In general the smaller printers tend to do 720x720 and the larger ones do 360x360. If the driver has a Finest Detail switch that gives the user a choice between these resamplings."

Most of the printing I do is from drum scans of 8x10 negatives—some of that is now done from 6000x4000 pixel digital capture— but in both cases, my personal preference is to keep the "master" file with layers at the original resolution, and then an additional flattened and sharpened version at the original resolution. Then, if I am making a much smaller print (8x10-16x20), I down-sample the image to the desired print size at 720ppi in photoshop (bicubic smoother) and then do some additional sharpening to compensate for any softening that may have happened in the interpolation. When the interpolation is going to be that extreme I would rather have some control over the process. If I am making a much larger print (24x30 and up) I will generally just print at whatever the resolution the file actually is and let the driver do the rest—although I tend to never need (nor want) to make a print where the resolution falls below 360-300 ppi. In most cases I'll let the 360 dpi determine the max print size and let it go.

I assume that higher-end RIPs like Studio Print or RIPs for image setters and film recorders have a much better way of handling that interpolation/dithering/profile conversion process (although I don't know anything about how they actually do it). For those RIPs I would just hand over the file to the application, define the output size, and let it do it's job.

Lenny made a point I wish more people would take to heart. Those print size/required PPI charts are completely worthless and don't take into consideration how people act and how the print actually looks. I'd like to meet a single photographer who does not put their nose up to the print—I'd also like to see someone try to keep you back at the "proper viewing distance" at an actual exhibition. Sure, a billboard can be printed at 25ppi, because you are 300 feet away and are moving at 60mph, but that is not how you look at art. If your file doesn't have enough information to print at larger sizes and retain detail then don't print larger (unless you don't care how your work actually looks). I would much rather see a good small print than a large bad one.

 

[RalphLambrecht]

I think what ralph was asking had more to do with the output resolution from the printer 720, 1440, 2880, and not the input resolution of the file. The print head only has 1440 actual nozzles per channel per inch, and, from my understanding, the 2880 resolution in the horizontal direction is achieved by the movement of the print head as it moves across the paper (you'll notice that it is 2880x1440).

Some papers might not be able to hold the actual dots when printed at the maximum resolution, and some dots might even end up in the "wrong" place on the paper, or overlap with other dots. In actuality, not all of the nozzles are firing for any given image pixel. The tone/color for a given image pixel, when printed, is a combination of multiple inks, dot sizes, the spacing of dots based on dithering algorithms, and overlaps with other inks. There are a whole lot of internal calculations that have nothing to do with the number of pixels coming into the driver.

With all that being said, I still want the printer to spit out as many dots as I can get. That is why I think digital black and white prints made with Cone inks are better than those made with UC k3 inks. There are more dots of more dilutions of grey, and therefore less visible dithering. Those prints look smoother. That might not be everyones criteria for "better" but it is certainly a part of the equation for me.

As for what resolution the input file should be: I don't know what the epson driver does when interpolating input resolution to something that it can then turn into dots on the paper. Based on what Roy Herrington has said about how QTR handles input resolution, in that all files are converted to 720x720 dots per inch, you have to assume that epson does something similar.

This is a quote from him from the QTR yahoo group:

"QTR resamples all input files to 720x720 pixels. Then dithering is done to whatever dpi you've selected -- mostly 2880x1440 or 1440x720. A couple printers also have 1440x1440.

"The resampling to 720x720 is done by the OS on the Mac so I'm not sure of the algorithm. On the PC it's a bi-linear algorithm -- basically two linear interpolations -- horizontal then vertical. As far as I know all the Epson drivers do either 360x360 or
720x720 resampling. In general the smaller printers tend to do 720x720 and the larger ones do 360x360. If the driver has a Finest Detail switch that gives the user a choice between these resamplings."

Most of the printing I do is from drum scans of 8x10 negatives—some of that is now done from 6000x4000 pixel digital capture— but in both cases, my personal preference is to keep the "master" file with layers at the original resolution, and then an additional flattened and sharpened version at the original resolution. Then, if I am making a much smaller print (8x10-16x20), I down-sample the image to the desired print size at 720ppi in photoshop (bicubic smoother) and then do some additional sharpening to compensate for any softening that may have happened in the interpolation. When the interpolation is going to be that extreme I would rather have some control over the process. If I am making a much larger print (24x30 and up) I will generally just print at whatever the resolution the file actually is and let the driver do the rest—although I tend to never need (nor want) to make a print where the resolution falls below 360-300 ppi. In most cases I'll let the 360 dpi determine the max print size and let it go.

I assume that higher-end RIPs like Studio Print or RIPs for image setters and film recorders have a much better way of handling that interpolation/dithering/profile conversion process (although I don't know anything about how they actually do it). For those RIPs I would just hand over the file to the application, define the output size, and let it do it's job.

Lenny made a point I wish more people would take to heart. Those print size/required PPI charts are completely worthless and don't take into consideration how people act and how the print actually looks. I'd like to meet a single photographer who does not put their nose up to the print—I'd also like to see someone try to keep you back at the "proper viewing distance" at an actual exhibition. Sure, a billboard can be printed at 25ppi, because you are 300 feet away and are moving at 60mph, but that is not how you look at art. If your file doesn't have enough information to print at larger sizes and retain detail then don't print larger (unless you don't care how your work actually looks). I would much rather see a good small print than a large bad one.

Richard, I'm sure Mr. Cone appreciates your free advertisement but, I just can't see how switching inks can change printer resolution.

 

[Richard Boutwell]

Richard, I'm sure Mr. Cone appreciates your free advertisement but, I just can't see how switching inks can change printer resolution.

This isn't about free advertisement for the Cone inks, it is about making comparisons from A to B. I could make the same comparisons with the Ebony-6 inks if I had a printer loaded with them and made the necessary prints. It is not that multiple grey inks "change printer resolution" You can't create nozzles that don't exist, but you can make use of more of the ink channels and total number of nozzles. Doing so then requires less dithering of the grey/black inks to produce the intermediary grey tones.

If you try printing a grey ramp with just one black ink and you will see that the lighter tones (like 5%k) are made up of clumps of visible dots (isn't that considered a loss of resolution?). But 5%k made up of a light light light black is almost perfectly uniformly covered—or "smoother". Now imagine that being split over the entire greyscale, with each point being almost uniformly covered with dilutions of ink—it will appear smoother. You can measure the ink densities of those patches all day long and they might be exactly the same for a print with 1 ink, 3 inks, or 7 inks, but when you visually inspect the print it will (or at least should) be obvious that the greyscale that is split up with more inks appears smoother.

If adding more grey inks didn't increase the apparent resolution of actual prints, Epson, Canon, and HP wouldn't have added those light dilutions of black or color inks. The 3 blacks in epson might be good enough for you and the vast majority of users, and that is fine; i'm happy that is possible. That just might not be good enough for those people who want to make the best prints they can, so I am happy that people create products that make that possible too.

 

[lenny]

I'm happy to give Jon, a friend of mine, a free endorsement. I've been using his inks for years. I've mixed my own as well, spent gobs of time measuring LAB values on endless drawdowns, etc., but now I just mix my own special mix of Jon's inks. It's a lot easier.

All you would have to do is realize two things. The first is if you have slightly less viscosity, you can have a little more resolution. The second is if you have to make b&w values using color inks, its easy enough to beat that with any b&w inks... There really is no contest.

That said, the measure of a print isn't in the resolution (provided you have a minimum ant). It's in the creative arrangement of tones across the tonal range. Usually the mid tones make up most of the story, altho' not always... Like a piano....

Lenny

 

[L Gebhardt]

Lenny, thanks for the link. I'm not sure I believe it disproves that Epsons scale the images to either 360ppi or 720ppi, but is saying that it really doesn't matter (in Epson's eyes). Thinking about how to write a driver I'm sure the image must be resized to some native value. Based on the Roy Herrington quotes provided by Richard Boutwell my guess is the Epsons are using 720ppi. I also guess that you would be very hard pressed to see a difference between 355 and 360ppi. Or even 257ppi scaled in PS vs 257ppi scaled in the driver/OS once printed.

I also use a drum scanner and when I print I usually flatten and down size the image to 360ppi or 720ppi, apply any output sharpening and save the copy with the print size in the name. That way I can run off another copy at a later time. My guess is that providing more than 720ppi just forces the driver/OS to resize each time you print. And it makes it harder to accurately apply sharpening because the resizing is going to vary based on the print size and original size.

For smaller images, like from a DSLR, I think I will still upsize to 360ppi (or rarely 720ppi) before printing. There's enough visible difference in the Photoshop resize options that it makes sense to control that rather than leaving it to an unknown algorithm. And it also lets me apply sharpening right before printing.

 

[lenny]

Larry,
I've been using StudioPrint for many years. I'm thinking of downsizing and might be moving over to Roy Harrington's Quadtone RIP next year. However, at the moment I have a RIP that is quite good at making these conversions... I edit on a Mac and flatten everything, save as tiff, before I send it over to the PC for printing. That's why I suggest that people do their own testing, on their equipment (and not take my opinion, or anyone else's as the setup can be quite different; its easy enough to do).

I can see the difference up to 720, but you are right, I couldn't distinguish between 700 and 720. I do know that, as you make the dithering pattern smaller, you can limit the effect of the ink on the paper. Depends on the RIP, type of pattern, the inkset, the paper......

I want my images to be sharp, but that's not a driving force. I'm not printing commercial images shot with flash setups. I'm more interested in atmosphere, and depth of field than I am critical sharpness... and the drum scanner provides plenty of sharpness...

Lenny