Best Print Resolution for Epson Printer - is it really 300 dpi?

Discussion in 'Digital Darkroom' started by david_williams|11, Jul 30, 2004.

  1. Hello,

    I've recently bought a Epson R800 and I'm trying to figure out which
    resolution I should print at. Here are some of the argument sI've
    seen on other posts:

    1) Print at 300 dpi by resizing the image appropriately in Photoshop.
    2) Print at 360 or 720 dpi
    3) Send an image to the Printer and let it handle the resizing and

    Anyone have any definitive answers on this?

  2. I vote for #3
  3. My Epson is an aging 870. 240 ppi going in, and printing at 720 dpi usually gives a decent result. It may be my imagination, but I think the printer responds well to 300 ppi going in, and printing at 720.
    I don't print at 1440 dpi unless I have 450 ppi going into the printer.
    There's a good reason for storing your images at 300 ppi. It seems to be a standard for print quality. Should you sell one, the buyer will likely ask for 300 ppi. An 8 x 12 at 300 is enough for most of my purposes. You can just feed your stored image into the printer (after a judicious hit of Unsharp Mask, if you haven't sharpened already). You'll get a little cropping (since paper is 11 1/2 long), but a nice print.
  4. There is a lot of math on this, and some of it is pretty convincing. That said, I did a few tests on a couple different papers, and found that I personally could see very little, if any, benefit to going over 300, and could see the difference between 240 and 300. For me, that answered the question for what I use for most of my printing on a 1280. [I did a similar test for 1440 vs. 2880 in the printer driver, and it was less conclusive on certain papers.] It is worth getting advice on this stuff from knowledgeable people, but I still recommend doing the test yourself at some point because it can be very educational. Enjoy.
  5. Spearhead

    Spearhead Moderator Staff Member

    It doesn't take too much work to do a test and find out what you like best. It takes about 30 minutes to do all that testing, and you may feel different than other people do about it. Also, it is somewhat paper and subject dependent.
  6. jbq


    I go for #3. I've been involved in writing printer drivers, so I have an idea of what goes on inside, and I know that there's no reason to not trust the printer driver.
  7. On my 1280 I print at 1440 and size my images at either 288 or 360 dpi (depending on how big the image is) since they are divisions of 1440. But having said that, in my unscientific tests I could not see a difference between images made at 288, 300 or 360 dpi. Do a test and let us know what you find out.
  8. Epson printer drivers will make use of high-contrast image information ranging all the way up to 720 ppi. Really you should never worry about downsampling or the like, if you have more than 300ppi just hand it off the the printer and it will probably make at least some use out of it.
  9. I can't answer this question, but would like to bring up a related issue. Perhaps Jean-Baptiste would be able to shed some light.

    On an inkjet printer, each color pixel contains multiple dots. With yellow, cyan, and magenta inks, and a single dot size, only 8 colors can be printed within a single dot position (yellow, cyan, and magenta from one color of ink; red, green, and blue from combinations of two inks; black, by combining all three inks; and paper white, from no ink in that dot position). Using the same technology, getting a color depth of 256 colors (8x8x8 levels of YxCxM) requires clumps of 8 dots to form each pixel. Sophisticated inkjet printers use more complex technologies, with variable-size dots and an additional two colors (light cyan and light magenta). Then 256 colors can be mixed with fewer than 8 dots per color pixel, or more than 256 colors can be mixed within an 8-dot pixel.

    The fundamental questions, then, are:

    1) What is the color depth of a JPEG file?

    2) How many printer dots are required to reproduce one colored pixel of that color depth with (a) three ink colors and one drop size; (b) three ink colors and multiple drop sizes; (c) five ink colors and multiple dot sizes?

    3) Does the printer resolution setting refer to the number of dots per inch or to the number of color pixels per inch?

    One other issue. Some printers have assymetric dpi capacities, e.g. 1200 dpi horizontally by 4800 dpi vertically. If each color pixel requires 8 dots, and those dots are laid out in an array 2 dots wide by 4 dots high, then the printer can produce up to 600x600 full-color pixels per inch. Assymetric dots per inch produce symmetric pixels per inch.

    Understanding the relationship between printer "resolution", printer dots per inch, and printer pixels per inch would imply a maximum printer resolution beyond which one would have to sacrifice color depth.
  10. "1) Print at 300 dpi by resizing the image appropriately in Photoshop."

    When you resize you image in Photoshop it will be in pixel per inch (PPI). When you set y our printer resolution setting it will be in dpi.

    So, size your image at 240ppi or more and print at the highest resolution (dpi) setting of your printer: 1440 or 2880.
  11. Use an even divisible of the output rez (then it's faster and there's no interpolation or
    scaling going ont). 180/240/360 etc. For most work, 240 is fine and dandy. I have seen
    with some images, with proper sharpening a slight advantage to 360ppi files but
    considering the larger file and speed in printing, you may or may not find it useful.
  12. There is a simple answer to this and other printing questions when resolution of the originating image is to be considered. It holds the same for newspapers, magazines and desktop printing. The print method doesn't matter if it is offset, web press, laser or ink jet...

    The key value to consider is the printer screen value. Epson, though the information is difficult to find it is available, has an inherent print screen of 150 LINES per inch (LPI). Image resolutions vs print screen for best repro results should be in the range of LPI x 1.5 to LPI x 2 >>> AND <<< this must be at the repro size... that is, if you intend to print a 4x6 inch final print the image should be pre-sized at 4x6inches @ 300PPI; for an 8x10 print then pre-size it to 8x10 @300PPI

    With this as the basis of the decision the best print you will get based on image resolution and pre-sizing (but not taking into account other controllable factors such as colour management and printer driver settings options) and the screen value itself you will get on the Epson is with the image at 300PPI. Anything more is interpolated and discarded by the printer driver; anything less, depending on the printer and driver, may be interpolated and added. If you change the print size between the file and the print itself, you alter the PPI value (resolution) that the printer "sees" and it will interpolate up or down as it sees fit. If you want to control the print process you need to be in control of the image file before printing it.
  13. jbq


    1) JPEG files are traditionally expanded to 24bpp during decoding, but could actually be decoded into a great color depth for greater accuracy. Typically, since you lose a lot more accuracy when decimating during compression, this is fairy pointless (i.e. you'd better use PNG or TIFF in such a case).

    3) The printer resolution is the size of grid on which the ink drops are laid out, i.e. the precision of the placement of any individual drop. It doesn't mean anything in terms of drop size or of number of drops that can be laid out (specifically, many printers can't actually put drops closer than 2 or 3 grid positions because of the time it takes to send a drop - that's one of the reasons why photo modes are slower).

    2) you make the fundamental error of assuming that a printer driver works by expanding each source pixel over a number of grid positions then populates those grid positions according to the color of the source pixel. This is wrong for several reasons:

    -the grid positions can't be selected independently, i.e. there is some exchange of information between neighboring pixels.

    -this would cause some minor color accuracy issues when printing dithered images (as all source pixels wouldn't be given the same weight in the final image).

    -this would very fundamentally reduce the color range of very high-resolution pictures, except if you started to use random-threshold dithering (which can't be used in a printer because the grid positions can't be selected independently).

    A printer driver typically by scanning the picture as it prints, working on a sliding window the height of the print head. It increases the source resolution to the printer resolution in the source color space, taking into accounts a blend of requiremenent in terms of color accuracy and geometric accuracy (typically applying a scaling algorithm more complex than the simple ones available in image-editing applications). It then applies color conversion and dithering simultaneously, processing all channels at the same time. The dithering is fairly complex because it needs to take into account the way multiple inks are available for the same color (e.g. how a big drop of light cyan can replace a small drop of dark cyan if the dark cyan head has just fired, but the behavior of such a replacement depends on how close a magenta or yellow drop is fired on the paper, because the light and dark version of the same ink don't blend the same way), how the inks are organized in cartridges (try to even out usage of the inks that are in a same cartridge), how you can't put too much ink in any location (it smudges). The conversion is a tricky blend of lookup tables, optimization and feedback loops. And all of that depends on the paper that is used, of course.

    As for the assymetry, it's the size of the grid. It shows that the horizontal resolution is achieved through timing, and the vertical resolution through a stepper motor. Typically the density of ink that can be put on paper is the same horizontally and vertically.
  14. From "Mastering Digital Printing" by Harald Johnson, a book that every digital printer
    should own. There would be a lot less questions on this forum if everyone did.

    "According to Epson, the input resolution--the resolution at which data is rasterized--is
    720 "dpi" for desktops and 360 "dpi" for wide formats. However, Epson now recommends
    300 ppi at the size you intend to print as their new magic number, if you get to 220 or
    less, you start to see a difference in image quality, and conversely, you won?t see much
    improvement with bitmapped images by going over 300 ppi.

    HP, on the other hand, whose print heads are based on a 600-dpi resolution instead of
    720, recommends 150-200 ppi at final size for its ink jet printers. They claim that
    scientists doing satellite photo reproduction for the government on their printers typically
    find that 125 ppi is adequate."

    I just had a 20" x 30" print made on a Epson 9600 and the operator also said 360 ppi is
    the magic number but 300 ppi will work. I couldn't argue after viewing the final print!
  15. There's no linescreen with an Epson, it's using Stochastic (FM) screening.
  16. Andrew Rodney , jul 31, 2004; 11:36 a.m. said: There's no linescreen with an Epson, it's using Stochastic (FM) screening.

    You are right in that in most modern printing processes, ink jets included, there is no real world "screen" (as there used to be in some printing)... but there is still an apparent (or computed) value and for the most part this circulates around 150LPI more or less for Epson ink jets, dictating an image feed at 300PPI (for photo quality printing) at the predetermined finished image size to achieve the best possible output quality. There is a range of LPI values available to the ink jet user by selecting output (printing) quality - but for photographic imaging, and some high detail line art as well, we should not even be considering "draft" quality (low LPI value ca50 - 75LPI in Epson) or "standard" quality (medium LPI value ca90 - 100/110 in Epson). But when it comes to photos, and especially on photo paper, why print an image at all if not at the best quality? As a matter of interest here, I have also come across specs for wet prints in mini-labs where an "apparent" LPI value is assumed - typically at 150LPI (THAT number again!).

    If there were no rule of thumb at all - in effect, as I gather you are suggesting, a completely random process where one minute the printer is performing, seemingly, at 75LPI and at the next moment seemingly at 150LPI - then there would be no repeatability or even uniformity in a particular print. Some value has to be assigned and typically that ranges from a ca75 screen value for newsprint (pulp) [typical draft quality output in ink jets] to ca100 for general printing [most ink jets at standard quality] and ca 150 [photo settings] and upwards to 200LPI, or more, for fine arts repro of the highest quality and high quality offset to the best quality coated book papers. These higher commercial printing process screen values are not available in most ink jets - BUT - the effect, however, can be realised by increasing the DPI value of ink laydown which is not an option available to an offset print house.

    Industry uses, and typically asks for, fundamentally two image (target) resolution values for standard pre-press image submissions intended for printing - for low quality papers image@150PPI and for better quality printing needs image@300. In the printing world up-sampling is frowned upon because of colour repro problems (artefacts, ghosting, aliasing, dot gain mis-matches, registration errors and colour bleeding) - but images can always be, and often are, down sampled in the pre-press workflow without introducing these problems. There is a lot of slack in LPI values as well - where image resolutions can vary from LPI x 1.5 to LPI x 2 and still produce mostly acceptable results except for the most critical printing requirements (on "slick" paper) and where the particular image resolution needs to be, absolutely must be, right on target vis a vis the printer's intended line screen values.
  17. Hi, i didnt read all the answer, so maybe i would repeat somoene say 244 dpi its more than enougth, i say 200 its more than
    enough even for your mini lab, i have done the test my self, believe me :)
  18. Jean-Baptiste, thanks for your detailed response to my questions! The relationship between printer dpi and printer ppi is *vastly* more complex than I had previously imagined. I'll look for the Harold Johnson text recommended by William Smith in a Borders bookstore in my area.
  19. -->There is a lot of slack in LPI values as well - where image resolutions can vary from LPI
    x 1.5 to LPI x 2

    That's known as quality factor, it's used to help folks figure out how many pixels per
    halftone dot to use for output to ink on paper, 4 color repro. This really isn't anything like
    the way an Epson prints. It's not using a halftone dot, you don't send it anything close to
    it's actual DPI output (say 720). In fact if someone where foolish enough to scan and
    output a true 720ppi file to an Epson, it would produce worse quality and take forever.

    If you're working with traditional halftone output, it's useful to find the true linesscreen
    and then double that value up to about a 133 linescren (after that, 1.5X is just fine). So if
    you knew you were going to print on a 85 linescreen for newspaper, you'd want to double
    that value (quality factor) as a good bases for input resolution (that is scan at 170ppi to
    produce two pixels per halftone dot). If you knew the linescreen was 150, scanning 300dpi
    would be overkill (but it wouldn't hurt other than slowing everything down). I'd scan at

    Again, this has nothing to do with out something like an Epson prints. You can send it as
    little as 180ppi or as much as 360ppi. I've seen in rare cases where that 360ppi can show
    some minor improvements over lower values but the image has to have a lot of high
    frequency detail and a lot of users would debate if the gain is worth the cost of much
    larger files and slower processing (editing and printing). 240ppi seems to be a pretty good
    compromise. And a LOT depends on the papers too! You can send less to matte papers
    than glossy. Your mileage may vary.
  20. Andrew,

    I agree that we are not too far apart, here, on most of the technical issues. Those peculiarities of view, yours and mine, that remain are undoubtedly as a result of the perhaps slightly different end user perspectives that you and I may have.

    You noted: "...And a LOT depends on the papers too! You can send less to matte papers than glossy. Your mileage may vary...." and that, in itself, is another whole discussion in file and colour management in repro. In fact, some of the imaging I do is actually on uncoated card of various weights exactly to take advantage of dot gain in the uncoated medium in rendering pseudo-continuous toning by allowing the ink to flow and saturate the fibres, sometimes to the point of actually over inking the media though my target is to stay just shy of that point. Objectively, for my purposes, that process produces a very "soft" image that would not appear so if the same image was sent to the offset printer (where it would visually appear as being somewhat more "crisp" - maybe a lot more)

    I conclude many times, however, based on questions here and in other forums that digital photographers often just don't understand the process of physical printing sufficiently to understand the parameters that need to be assigned to their images. On the other hand, I come from commercial image publishing and graphics repro not based on ink-jet methods to the question of desktop printing that is, and so then tend to start with the rule-of-thumb values that I better understand and know how to manipulate to advantage.

    Like wet printing in a darkroom... jet, laser and commercial printing becomes a speciality skill (maybe describing the "art" of it?) all it's own over and above the actual image capture. The user begins, and slowly over time hopefully improves, in the ability to see and visualise an image on screen and judge the character that it will have when translated to the printed page in specific printing environments. In wet darkrooms there were photographers that could make beautiful images but couldn't print them for s**t - OTOH there were printers who could take a s**t photo and make gorgeous prints; the same thing has happened in the digital age.

Share This Page