Printing from lightroom to pixma pro 10

[Sbarroso]

Regarding printing in 16 bits:

I have the Pixma Pro 1 and I run LR in Windows. It is not possible to print 16 bits directly from Lightroom using windows. Even if you select the XPS driver. I've checked recently again and I still see banding in smooth (BW) gradations (e.g. the sky near the horizon in this picture).

But there is a way to do it inderectly. Install the Canon plug-in for LR "Canon Print Studio Pro". Use that plug-in for the picture you want to print. You won't find a"16 bits check box" around, just use the XPS driver and you will print in 16 bits. In this way, your smooth gradations won't show any banding.

What I don't like from the plug in is that you cannot crop the picture as you can do in LR print module. However, you have all kind of options to control the output in a nicer way than in the printer driver.

I don't think it's worth to use 16 bit printing on images that don't have smooth gradations, unless you have a very trained eye. My eyes aren't able to see differences in those pictures, although they do in the gradients. Other people I've shown examples need some time to realize it, after I point to it. Some don't still see the difference.

Best,
Santiago

[NAwlins_Contrarian]

[P]rinters have, based on settings, specific ppi that they need/expect to print and image. If they do not get that ppi, the will interpolate to that size and, usually, that interpolation is much rougher than what would be done in the computer.... Also, the computer interpolation used to get to the expected printer ppi, which is why better results are obtained be Perfect Resize or Qimage vs bicubic.

I may be missing something--if so, please tell me, I'd love to learn!--but it seems to me that some interpolation is almost unavoidable. (This post is not about how much this some interpolation affects print quality; but if the very negative statements about the quality of printer-based interpolation are correct, then the issue merits some inquiry.) One can say that the printer needs the exact ppi. But the ppi tag does not (AFAIK, in most common software) control the printing behavior; the input file's pixel dimensions and the requested output size do that. It seems to me--again, I'd love to find out that I'm wrong--that we have no reasonably-easy way to determine the exact pixel dimensions to send for a given print size for a few reasons:

(1) We do not know the printer's exact ppi resolution. We may say that Epsons are 360 or 720 ppi and that Canons are 300 or 600 ppi. But how likely is it that Japanese companies performing most if not all of their engineering in Japan are working in English units? Doesn't it seem far more likely that the exact specification is an even figure in a metric unit, like pixels per millimeter (ppmm)? Some years back by own analysis of a "250 ppi" lab concluded that it was far more likely that the printers were actually 254 ppi--which is exactly 10 ppmm. I'd bet lunch that so-called 300 ppi printers are actually 12 ppmm, which is 304.8 ppi. And what about those supposedly-360 ppi Epsons? I suspect that they are actually 14 ppmm = 355.6 ppi. But I don't know any of this. And I would be unlikely to trust information on it from any U.S.-based source. Obviously if you send a printer an image that has enough pixels for 300 ppi and the printer needs, 304.8 ppi, the printer (or its driver) has to up-interpolate to generate 1.6% more pixels. And obviously if you send a printer an image that has enough pixels for 360 ppi and the printer needs 355.6 ppi, the printer (or its driver) has to down-scale to generate 1.2% fewer pixels.

(2) We don't know what the printer thinks the media size really is. Have you ever gotten "8x10 inch" prints from a lab that appeared to be slightly less than 8 inches, and maybe slightly less than 10 inches? I have. From more than one place. I suspect that the paper rolls they were feeding into their printers were not exactly 8 inches wide, but in fact were exactly 200mm wide, which comes to 7.87 inches. So if you tell your printer 8x10 inches, does it really expect exactly 8x10 inches? Maybe, maybe not, it's hard to know. (Although I suspect that the U.S.-market printers actually do expect the exact dimensions in English units).

(3) For borderless prints, we do not know how much overspray the printer applies. The paper feed guides are not ultra-high-precision devices; they just slide into place, up to the edge of paper that might be bowing a little. Except where there's paper misaligment in the feed path, I have not seen thin white borders on borderless inkjet prints. To achieve this, the printer has to spray at least a little ink past / off the edge of the paper. How much? I don't know, but enough that some people recommend not printing borderless to avoid gunking up the printer's insides with overspray ink. If the printer oversprays by even 1/32 inch (a little under 1 mm) on each side, then for a print at (a nominal?) 300 ppi, it needs about 19 extra pixels in each dimension. So you think, '8x10 inches at 300 ppi means I send 2400x3000 pixels to avoid interpolation.' In reality the printer needs 2419x3019 pixels, to give something for overspray, and some sort of interpolation has to occur.

Is Lightroom, or Qimage, or whatever able to query the printer about exactly how many pixels it needs, to allow them to use their more sophisticated interpolation methods to send the printer the exact number of pixels? I don't know. Maybe. But if you or I can get that information from Lightroom or Qimage or whatever, please tell me how.

Thanks!

[jrsforums]

I may be missing something--if so, please tell me, I'd love to learn!--but it seems to me that some interpolation is almost unavoidable. (This post is not about how much this some interpolation affects print quality; but if the very negative statements about the quality of printer-based interpolation are correct, then the issue merits some inquiry.) One can say that the printer needs the exact ppi. But the ppi tag does not (AFAIK, in most common software) control the printing behavior; the input file's pixel dimensions and the requested output size do that. It seems to me--again, I'd love to find out that I'm wrong--that we have no reasonably-easy way to determine the exact pixel dimensions to send for a given print size for a few reasons:

(1) We do not know the printer's exact ppi resolution. We may say that Epsons are 360 or 720 ppi and that Canons are 300 or 600 ppi. But how likely is it that Japanese companies performing most if not all of their engineering in Japan are working in English units? Doesn't it seem far more likely that the exact specification is an even figure in a metric unit, like pixels per millimeter (ppmm)? Some years back by own analysis of a "250 ppi" lab concluded that it was far more likely that the printers were actually 254 ppi--which is exactly 10 ppmm. I'd bet lunch that so-called 300 ppi printers are actually 12 ppmm, which is 304.8 ppi. And what about those supposedly-360 ppi Epsons? I suspect that they are actually 14 ppmm = 355.6 ppi. But I don't know any of this. And I would be unlikely to trust information on it from any U.S.-based source. Obviously if you send a printer an image that has enough pixels for 300 ppi and the printer needs, 304.8 ppi, the printer (or its driver) has to up-interpolate to generate 1.6% more pixels. And obviously if you send a printer an image that has enough pixels for 360 ppi and the printer needs 355.6 ppi, the printer (or its driver) has to down-scale to generate 1.2% fewer pixels.

(2) We don't know what the printer thinks the media size really is. Have you ever gotten "8x10 inch" prints from a lab that appeared to be slightly less than 8 inches, and maybe slightly less than 10 inches? I have. From more than one place. I suspect that the paper rolls they were feeding into their printers were not exactly 8 inches wide, but in fact were exactly 200mm wide, which comes to 7.87 inches. So if you tell your printer 8x10 inches, does it really expect exactly 8x10 inches? Maybe, maybe not, it's hard to know. (Although I suspect that the U.S.-market printers actually do expect the exact dimensions in English units).

(3) For borderless prints, we do not know how much overspray the printer applies. The paper feed guides are not ultra-high-precision devices; they just slide into place, up to the edge of paper that might be bowing a little. Except where there's paper misaligment in the feed path, I have not seen thin white borders on borderless inkjet prints. To achieve this, the printer has to spray at least a little ink past / off the edge of the paper. How much? I don't know, but enough that some people recommend not printing borderless to avoid gunking up the printer's insides with overspray ink. If the printer oversprays by even 1/32 inch (a little under 1 mm) on each side, then for a print at (a nominal?) 300 ppi, it needs about 19 extra pixels in each dimension. So you think, '8x10 inches at 300 ppi means I send 2400x3000 pixels to avoid interpolation.' In reality the printer needs 2419x3019 pixels, to give something for overspray, and some sort of interpolation has to occur.

Is Lightroom, or Qimage, or whatever able to query the printer about exactly how many pixels it needs, to allow them to use their more sophisticated interpolation methods to send the printer the exact number of pixels? I don't know. Maybe. But if you or I can get that information from Lightroom or Qimage or whatever, please tell me how.

Thanks!

Yes.  Qimage queries the driver and gets the answer to all three of your questions.  It then uses the highest quality interpolation algorithms (not bicubic) to resize the image based on the paper and driver settings. 

While resizing it uses "smart sharpening" to maintain the sharpening for any size changes required.  This sharpening can be tweaked for specific paper/ICC type.

Here is one old video that may start your understanding...There are manymore
https://m.youtube.com/watch?feature=youtu.be&hd=1&v=gMzUxDP_JwM

[NAwlins_Contrarian]

Qimage queries the driver and gets the answer to all three of your questions.

Thanks for the link. I watched that and several other videos on Qimage (listed below). What follows may just tend to prove the limits of what I understand (a/k/a my own ignorance), so take it with appropriate grains of salt. But:

Unfortunately, the linked video actually decreased my confidence in Qimage / this procedure. The prime example is his part about when you select borderless printing on the Epson R1900 he's using, Qimage reports 725 ppi instead of the usual 720 ppi. If the guy presenting the video knows what's going on there, at least he doesn't seem to explain it. I'd bet that what's actually occurring is that the Epson driver is artificially claiming 725 ppi without really changing anything. Indeed, if the point were to compensate for overspray, then the print head run would stretch out slightly the resolution would go down, say to 715 ppi or something, not up to 725 ppi. But what I bet the Epson driver is doing is telling the program printing, 'If you're going to try to send me exactly how many pixels you think I need, increase that number by 0.69% (725/700=1.00694), because that's how much extra I need in terms of pixels for overspray. So on the shown 8.5x11 inch sheet, theoretically and presumably it is 6120 pixels wide (8.5x720=6120), but the R1900 is going to overspray by about 21 pixels, i.e., about 0.03 inch or 0.75mm on each side, so it really needs about 6162 pixels to work with, i.e., 725 ppi x 8.5 inches. But the way Qimage is shown to handle the borderless printing setting makes me doubt the sophistication / correctness of its 'querying the printer driver'. (And for the reasons previously stated, I still tend to suspect that the 300, 360, 600, and 720 ppi figures are nominal English-unit ones, and the actual figures differ slightly, and are even number in ppmm.)

Then I was under the impression that Qimage bypasses the printer driver to take direct control of the hardware. But watching the videos, that is obviously not the case. In the videos he goes into the printer properties dialog boxes for both a Canon Pro-9000 and an Epson R1900--which is clearly using the manufacturers' drivers--and in one point I think he actually discusses that. I can see pros and cons to using the manufacturers' drivers, but this tends to slot the sophistication of Qimage somewhat lower than had been my impression (and contrasting with, say, VueScan, which AFAIK does bypass the manufacturers' drivers and control the hardware directly).

From those videos and Jose Rodriguez's, I tend to suspect that Qimage's best uses are (1) making prints that are very large relative to the available input file resolution (because the up-interpolation does sound pretty sophisticated) and (2) conveniently laying out print packages of multiple prints on one sheet or roll.

videos watched
Qimage Ultimate - Paper Size and Printable Area
Qimage Ultimate - Printer Settings
Qimage Ultimate - Printing Options
Qimage Ultimate Intelligent Printer Settings
and Jose Rodriguez / Jtoolman's QIMAGE Why do so many people LOVE it and yet so many HATE it?

[Mark D Segal]

...................

Is Lightroom, or Qimage, or whatever able to query the printer about exactly how many pixels it needs, to allow them to use their more sophisticated interpolation methods to send the printer the exact number of pixels? I don't know. Maybe. But if you or I can get that information from Lightroom or Qimage or whatever, please tell me how.

Thanks!

I'm wondering whether this thread isn't going completely OT and beyond the scope of useful information. Has anyone participating here done or seen any tests, on the most recent printer models and with most recent software, whether it makes a particle of practical difference to image quality: (a) whether the application or elsewhere in the print pipeline resampling to native resolution gets done and for each option within what limits of resampling, and (b) whether the effective native resolution of the print head is give or take 5 pixels around its usually quoted native resolution? None of this discussion in and of itself would help improve my prints without a solid basis of evidence supporting it.

[jrsforums]

Thanks for the link. I watched that and several other videos on Qimage (listed below). What follows may just tend to prove the limits of what I understand (a/k/a my own ignorance), so take it with appropriate grains of salt. But:

Unfortunately, the linked video actually decreased my confidence in Qimage / this procedure. The prime example is his part about when you select borderless printing on the Epson R1900 he's using, Qimage reports 725 ppi instead of the usual 720 ppi. If the guy presenting the video knows what's going on there, at least he doesn't seem to explain it. I'd bet that what's actually occurring is that the Epson driver is artificially claiming 725 ppi without really changing anything. Indeed, if the point were to compensate for overspray, then the print head run would stretch out slightly the resolution would go down, say to 715 ppi or something, not up to 725 ppi. But what I bet the Epson driver is doing is telling the program printing, 'If you're going to try to send me exactly how many pixels you think I need, increase that number by 0.69% (725/700=1.00694), because that's how much extra I need in terms of pixels for overspray. So on the shown 8.5x11 inch sheet, theoretically and presumably it is 6120 pixels wide (8.5x720=6120), but the R1900 is going to overspray by about 21 pixels, i.e., about 0.03 inch or 0.75mm on each side, so it really needs about 6162 pixels to work with, i.e., 725 ppi x 8.5 inches. But the way Qimage is shown to handle the borderless printing setting makes me doubt the sophistication / correctness of its 'querying the printer driver'. (And for the reasons previously stated, I still tend to suspect that the 300, 360, 600, and 720 ppi figures are nominal English-unit ones, and the actual figures differ slightly, and are even number in ppmm.)

Then I was under the impression that Qimage bypasses the printer driver to take direct control of the hardware. But watching the videos, that is obviously not the case. In the videos he goes into the printer properties dialog boxes for both a Canon Pro-9000 and an Epson R1900--which is clearly using the manufacturers' drivers--and in one point I think he actually discusses that. I can see pros and cons to using the manufacturers' drivers, but this tends to slot the sophistication of Qimage somewhat lower than had been my impression (and contrasting with, say, VueScan, which AFAIK does bypass the manufacturers' drivers and control the hardware directly).

From those videos and Jose Rodriguez's, I tend to suspect that Qimage's best uses are (1) making prints that are very large relative to the available input file resolution (because the up-interpolation does sound pretty sophisticated) and (2) conveniently laying out print packages of multiple prints on one sheet or roll.

videos watched
Qimage Ultimate - Paper Size and Printable Area
Qimage Ultimate - Printer Settings
Qimage Ultimate - Printing Options
Qimage Ultimate Intelligent Printer Settings
and Jose Rodriguez / Jtoolman's QIMAGE Why do so many people LOVE it and yet so many HATE it?

http://ddisoftware.com/tech/articles/december-2007-border-patrol-all-about-borderless-printing/

P.S. You have misplaced views on sophistication.  Also, Vuescan often uses mfg. scanners drivers.  I have used for years and have had Vuescan unable to support scanners when new OS did not have driver for old scanner.  A lot depends on the scanner and driver interface.

Hamrick is pragmatic, not sophisticated....So is Chaney.  Qimage is not a RIP and does not claim to be, but it does print things that few others do...Just focus on its print capability.

[jrsforums]

I'm wondering whether this thread isn't going completely OT and beyond the scope of useful information. Has anyone participating here done or seen any tests, on the most recent printer models and with most recent software, whether it makes a particle of practical difference to image quality: (a) whether the application or elsewhere in the print pipeline resampling to native resolution gets done and for each option within what limits of resampling, and (b) whether the effective native resolution of the print head is give or take 5 pixels around its usually quoted native resolution? None of this discussion in and of itself would help improve my prints without a solid basis of evidence supporting it.

Ask Jeff.  For years he strongly proposed that there was no need to resize to "native resolution".  His recent books and statements fully support it.  I praise Jeff as it takes courage to reverse your position.

Jeff Kesson did some work on interpolation outside of lightroom, using Qimage and Perfect Resize.  Here is one article on Qimage http://blog.kasson.com/technical/resampling-for-printing-with-qimage/  You may wish to contact him to see if he did any detail work comparing resizing in computer vs printer driver.  Chaney has done work, but like many "experts" I feel you may not believe him.

[Mark D Segal]

Ask Jeff.  For years he strongly proposed that there was no need to resize to "native resolution".  His recent books and statements fully support it.  I praise Jeff as it takes courage to reverse your position.

Jeff Kesson did some work on interpolation outside of lightroom, using Qimage and Perfect Resize.  Here is one article on Qimage http://blog.kasson.com/technical/resampling-for-printing-with-qimage/  You may wish to contact him to see if he did any detail work comparing resizing in computer vs printer driver.  Chaney has done work, but like many "experts" I feel you may not believe him.

I'm appropriately asking the present discussants. But maybe it's a bit risky to "feel" what you don't "know". This isn't about "taking positions" or"reversing them" or "believing" one person or another. It's about whether there is evidence to support a hypothesis. The only expertise on call here is that which is prepared to accept or reject a hypothesis based on credible evidence, so I'm asking about the evidence.

Thanks for mentioning the sources you point to. I have not seen that evidence, as I do almost 100% of my printing straight from raw files in Lightroom and would first like to see such evidence emerging from a comparatively simple Adobe-based workflow. I say this without prejudice to what others may find preferable for their work.

[jrsforums]

I'm appropriately asking the present discussants. But maybe it's a bit risky to "feel" what you don't "know". This isn't about "taking positions" or"reversing them" or "believing" one person or another. It's about whether there is evidence to support a hypothesis. The only expertise on call here is that which is prepared to accept or reject a hypothesis based on credible evidence, so I'm asking about the evidence.

Thanks for mentioning the sources you point to. I have not seen that evidence, as I do almost 100% of my printing straight from raw files in Lightroom and would first like to see such evidence emerging from a comparatively simple Adobe-based workflow. I say this without prejudice to what others may find preferable for their work.

Mark...First too clarify my post....The first Jeff I referenced was Jeff Schewe. I would suggest you  contact him on reason he changed his advice.

Second Jeff should have been JIM Kasson

I have been using Qimage since Late 1990's or early 2000's, so always had benefit of accurate resolution and high level interpolation.  I have, for my own benefit, done multiple test comparisons from Photoshop and Lightroom....Mostly using proper resolution, sometimes without.  On every print comparison Qimage print was superior.  I have not specifically done tests of native vs non-native....Again, I would suggest you discuss with Schewe if you do not agree with is current recommendations.

Btw....Printers I used were 3800, 3880, 4900, and, currently, P-1000

You may also like to review this http://www.ddisoftware.com/qimage-u/tech-prt.htm

And this http://ddisoftware.com/tech/articles/july-2011-restless-natives/

[Mark D Segal]

Thanks for the references, and something I can easily discuss with Jeff, as we may have in the past. I'm not partial to either approach regarding resampling strategy because I think there is reason to believe both are fine. I can't test QImage because I am on OSX and won't go through the pain of installing Parallels/Windows just for that; as well, I do like my raw workflow in LR and I'm quite sure 99% of the time it's fine, because I actually have compared letting LR resample to the driver resolution or letting the system do it and frankly for the most part I think for the range I tested you'd need a magnifier to see any difference. But useful for those on Windows to know you are very pleased with what QImage is doing for you.

[jrsforums]

Thanks for the references, and something I can easily discuss with Jeff, as we may have in the past. I'm not partial to either approach regarding resampling strategy because I think there is reason to believe both are fine. I can't test QImage because I am on OSX and won't go through the pain of installing Parallels/Windows just for that; as well, I do like my raw workflow in LR and I'm quite sure 99% of the time it's fine, because I actually have compared letting LR resample to the driver resolution or letting the system do it and frankly for the most part I think for the range I tested you'd need a magnifier to see any difference. But useful for those on Windows to know you are very pleased with what QImage is doing for you.

A friend of mine went to OSX, but is running a Windows system just for Qimage printing.

Jim Kasson did, and posted, test of Perfect Resize and Qimage vs. PS and LR.  Both PR and Q were comparable and beat the Adobe systems.  Qimage is easier to use as you sharpen to what you view on display, then Q will properly do final sharpening for paper, ink, size required.  No need to guess if you have over sharpened enough or to sharpen for each size, paper, etc.