Canon Printer Plug-In versus QImage for Canon IPF?

[StephenOzcomert]

I have been testing different media for my new Canon IPF 8300.  I have also been using the supplied printer plug-in which allows printing in 16 bits after folks here pointed me to the fact that I just needed to use the CS4 in 32 bit mode to access it.  Anyway, I have not yet moved on to roll media but when I do, I expect there will frequently be times when I want to printer smaller images than the roll maximum width and thus will want to layout multiple images.  QImage has this functionality which is quite nice but only allows an 8 bit work flow.  Unless I am missing something, I have not found any easy ways to do this printing from the plug-in on my windows 7 machine.  I know QImage claims there is no meaningful difference printing in 8 bit or 16 bit (QImage only supports 8 bit printing so far).  What do others here do?  Anyone using QImage?  Any loss in quality?  Go back and forth?  Any other solutions that I am overlooking (quite likely)?  Thanks for any thoughts.

Edited to correct my mangled initial post which I wrote late at night and clearly not thinking clearly!

[JeffKohn]

First to clarify a few things, since you're talking about 16-bit, 32-bit, and 64-bit.

There's 32-bit and 64-bit OS/Applications, but that really has nothing to do with print quality, just with application compatibility. Qimage is a 32-bit application, but will run just fine on 64-bit OS. Likewise, the Canon print plug-in is 32-bit, and also works on 64-bit OS just fine (but only works with the 32-bit version of Photoshop, because 32-bit plugins can't be hosted by 64-bit applications).

Now there's also the issue of bit-depth for printing. This has to do with the color precision of the image data sent to the printer for output (has nothing to do with 32-bit or 64-bit applications).

In the Windows world, printing has always been limited to 8-bits per channel color. This actually changed a while back with the XPS Printing Path, but most print manufacturers don't seem to be supporting this in their OS-level drivers, and most applications don't seem to support it either. Maybe with time it will catch on. I think part of it may be that this is a new printing subsystem with a new programming interface, so it's not necessarily just a matter of switching from 8-bit data to 16-bit data.

The Canon Print Plug-in supports 16-bit output, because it bypasses the windows print driver (or maybe it uses XPS, I don't know). The downside is that it only works with 32-bit versions of Photoshop, so Lightroom and QImage users are left out.

Apparently the there's a "free layout" utility from Canon that is accessible when printing with the OS driver on 32-bit OS. But it's not available on 64-bit OS for some reason, and I'm not sure if it was ever available from the Photoshop plugin.

IMHO Mike Cheney is being a bit disingenuous when he says there's no point to 16-bit printing; after all this is the guy who advocates interpolating images to 720 PPI for Epson desktop printers. The real reason Cheney doesn't support 16-bit printing is because he can't. He's totally dependent on the OS print drivers, which means the only way he can support 16-bit would be if the printer manufacturers release 16-bit capable XPS print drivers and he updates his code to use the XPS print path.

So getting back to your question of does it matter if you print 8-bit color from QImage versus 16-bit from the plugin.  In theory, sending a 16-bit image to the printer as 16-bit color is ideal. In reality many who have used the 16-bit print plugin will tell you there is some improvement for some images, but it's not a huge difference and you may not see any benefit depending on the image. My approach is to use QImage for smaller images  where I think the benefit of 16-bit is less likely to show in real-world photographs. Smaller prints are also where Qimage's layout capabilities are most useful. So 12x18" and larger goes through the PS plugin, and anything smaller goes though Qimage if I need layout capabilities. ICC profiles should be compatible with both print paths as long as all the relevent settings are consistent.

[Ernst Dinkla]

IMHO Mike Cheney is being a bit disingenuous when he says there's no point to 16-bit printing; after all this is the guy who advocates interpolating images to 720 PPI for Epson desktop printers. The real reason Cheney doesn't support 16-bit printing is because he can't. He's totally dependent on the OS print drivers, which means the only way he can support 16-bit would be if the printer manufacturers release 16-bit capable XPS print drivers and he updates his code to use the XPS print path.

So getting back to your question of does it matter if you print 8-bit color from QImage versus 16-bit from the plugin.  In theory, sending a 16-bit image to the printer as 16-bit color is ideal. In reality many who have used the 16-bit print plugin will tell you there is some improvement for some images, but it's not a huge difference and you may not see any benefit depending on the image. My approach is to use QImage for smaller images  where I think the benefit of 16-bit is less likely to show in real-world photographs. Smaller prints are also where Qimage's layout capabilities are most useful. So 12x18" and larger goes through the PS plugin, and anything smaller goes though Qimage if I need layout capabilities. ICC profiles should be compatible with both print paths as long as all the relevent settings are consistent.

Mike Chaney doesn't advocate the interpolation to 720 PPI for Epson desktop printer models. Qimage will interpolate image data on the fly to 360 or 720 PPI depending on the quality settings in the printer driver and with the Qimage interpolation setting at maximum. It is the driver that makes the call for either 360 or 720 PPI input resolution and even then Qimage allows the user to send a lower resolution  and let the driver do the rest of the upsampling. The same for Canon and HP printers but then on 300 or 600 PPI.

Mike Chaney didn't see much advantage in 16 bit printing (2006) but in the past (2003) suggested a way to get a 16-bit pipeline even when there were only 8-bit driver pipelines possible on the OS systems some years ago.

http://ddisoftware.com/tech/articles/decem...6-bit-printers/

http://luminous-landscape.com/forum/index....rt=#entry110529

While I certainly do not share all the opinions Mike has, I think you didn't represent his opinions correctly here.
And I think that using Qimage for all print sizes still has advantages over using other applications despite its 8bit in- and output, (there are 16bit steps internally BTW).
I wouldn't like to set up a 44"x200" print in Photoshop CS 4 even if it runs in a 64 bit environment.  Not to mention using a 16 bit driver then.


met vriendelijke groeten, Ernst Dinkla

Try: http://groups.yahoo.com/group/Wide_Inkjet_Printers/

[Scott Martin]

If Photoshop is your main application of choice, I'd use the printing plug-in for most everything and reserve Qimage for those occasions when you need to nest a bunch of images. Of course, if you're a RAW shooter, Lightroom is notable for many reasons and would be great  for printing single and multiple images on a roll complete with print sharpening all in one app.

[jerryrock]

As a side note, I have tested the selectable 16 bit print path of Photoshop CS5 on a Mac running Snow Leopard 10.6.3 and found the print quality on the Canon iPF5100 to equal that of the Canon iPF print plug in run on Photoshop CS4.

[JeffKohn]

Mike Chaney doesn't advocate the interpolation to 720 PPI for Epson desktop printer models. Qimage will interpolate image data on the fly to 360 or 720 PPI depending on the quality settings in the printer driver and with the Qimage interpolation setting at maximum. It is the driver that makes the call for either 360 or 720 PPI input resolution and even then Qimage allows the user to send a lower resolution and let the driver do the rest of the upsampling. The same for Canon and HP printers but then on 300 or 600 PPI.

The original basic premise of Qimage (along with layout capabilities), was that there is benefit to interpolating up to the print driver's native resolution. I actually agree with this up to a certain point; I think interpolating to the native resolution (or a multiple) preserves the effect of output sharpening better than sending an odd PPI to the printer. But 720ppi or 600ppi is just stupid overkill, which is why I would use 300 or 360. But not only does Qimage support the 720/600 resolutions, it's actually the default for a lot of desktop printers. I'm not criticizing the fact that this capability is there; I just think it points to a certain amount of hypocrisy when his justification for arguing against 16-bit is that it's overkill with no real benefit.

Mike Chaney didn't see much advantage in 16 bit printing (2006) but in the past (2003) suggested a way to get a 16-bit pipeline even when there were only 8-bit driver pipelines possible on the OS systems some years ago.

http://ddisoftware.com/tech/articles/decem...6-bit-printers/

Sorry but IMHO he loses credibility with that post. To me it just shows that he is no longer targeting Qimage to users wanting to ge the absolute best print quality possible. His basic argument is that Adobe RGB is "good enough" for a working space, and 8-bit printing is good enough for Adobe RGB. He even admits to a benefit when using large-gamut working spaces. And the fact is many of us do use large-gamut working spaces, so his argument doesn't hold any water unless you've been drinking the Dan Margulis koolaid.  IMHO his argument against 16-bit printing amounts to pure spin. I think he would add it if he could (probably to the Studio version since he's intent on getting people to upgrade).

http://luminous-landscape.com/forum/index....rt=#entry110529

I can't see "hacking" the 8-bit GDI path as making much sense at all, especially when there's already an alternative printing path that supports 16-bit.

[Bart_van_der_Wolf]

The original basic premise of Qimage (along with layout capabilities), was that there is benefit to interpolating up to the print driver's native resolution. I actually agree with this up to a certain point; I think interpolating to the native resolution (or a multiple) preserves the effect of output sharpening better than sending an odd PPI to the printer. But 720ppi or 600ppi is just stupid overkill, [...]

Why do you think it is overkill? Is there no benefit to interpolating edges to a better degree than most printer drivers do? Is there no benefit to interpolating intermediate colors in a gradient better than most printer drivers do? Is there no benefit to sharpening at the pixel level after interpolating to the pixel grid that the printer driver uses internally, instead of the printer interpolating the already sharpened result? In my opinion there is a benefit on all counts.

But not only does Qimage support the 720/600 resolutions, it's actually the default for a lot of desktop printers.

It's not a default, it's whatever the printer driver signals it is going to use itself, and Qimage takes out the interpolation/sharpening blackbox by supplying a better datastream (without choking the spoolfile) ready to use.

Sorry but IMHO he loses credibility with that post. To me it just shows that he is no longer targeting Qimage to users wanting to ge the absolute best print quality possible. His basic argument is that Adobe RGB is "good enough" for a working space, and 8-bit printing is good enough for Adobe RGB.

That is not his argument, but you seem to read it with a biased mind. His argument (one of them anyway) is that when you go from capture, via processing, to a print ready state, there is usually not enough data left to justify a 16-b/ch print datastream, in fact it can generate posterization if the output colorspace is much too large for the output modality.

He even admits to a benefit when using large-gamut working spaces.

What do you want to imply by "even", and "admit", or is that your bias shining through?

IMHO his argument against 16-bit printing amounts to pure spin.

I haven't heard any objective counter arguments so far, but I'll keep an open mind to be educated.

I think he would add it if he could (probably to the Studio version since he's intent on getting people to upgrade).

What makes you come to that conclusion, or is it you apparent bias speaking again?

Cheers,
Bart

[JeffKohn]

Why do you think it is overkill? Is there no benefit to interpolating edges to a better degree than most printer drivers do? Is there no benefit to interpolating intermediate colors in a gradient better than most printer drivers do? Is there no benefit to sharpening at the pixel level after interpolating to the pixel grid that the printer driver uses internally, instead of the printer interpolating the already sharpened result? In my opinion there is a benefit on all counts.

As I said, I agree on principle, up to a certain point. But I've done the tests, and printing a 720ppi doesn't improve anything, even with a loupe. All it does is increase spool times.  As for sharpening, I don't much care for Qimage's implementation when printing large, which is another reason I use the Canon plugin when printing big.

It's not a default, it's whatever the printer driver signals it is going to use itself, and Qimage takes out the interpolation/sharpening blackbox by supplying a better datastream (without choking the spoolfile) ready to use.

Qimage's default is the "high resolution" printing mode, which means 720ppi for Epson desktop printers. When you install Qimage and run it for the first time, images sent to an Epson 2400 will be sent at 720ppi. I call that default behavior, if you want to argue semantics I'm not really interested.

That is not his argument, but you seem to read it with a biased mind. His argument (one of them anyway) is that when you go from capture, via processing, to a print ready state, there is usually not enough data left to justify a 16-b/ch print datastream, in fact it can generate posterization if the output colorspace is much too large for the output modality.

It is his argument, go back and re-read. He argues that you should be using Adobe RGB as your working space and that ProPhoto is too big. I disagree with that.

What makes you come to that conclusion, or is it you apparent bias speaking again?

Of course I'm biased, everybody is biased. Anyone who says they aren't is either a liar or deluding themselves. I'm not exactly sure what your point is except that you seem to take offense that I would criticize decisions Chaney has made in developing Qimage.  I purchased Qimage about 5 years ago and was a big fan for a long time. But I think Qimage has drifted off-course from its original mission. I personally would rather see him focus on print quality, including the features of newer printers such as 16-bit printing. But instead he's implementing things like RAW conversion which I just don't see the point of.

[Ernst Dinkla]

The original basic premise of Qimage (along with layout capabilities), was that there is benefit to interpolating up to the print driver's native resolution. I actually agree with this up to a certain point; I think interpolating to the native resolution (or a multiple) preserves the effect of output sharpening better than sending an odd PPI to the printer. But 720ppi or 600ppi is just stupid overkill, which is why I would use 300 or 360. But not only does Qimage support the 720/600 resolutions, it's actually the default for a lot of desktop printers. I'm not criticizing the fact that this capability is there; I just think it points to a certain amount of hypocrisy when his justification for arguing against 16-bit is that it's overkill with no real benefit.


Sorry but IMHO he loses credibility with that post. To me it just shows that he is no longer targeting Qimage to users wanting to ge the absolute best print quality possible. His basic argument is that Adobe RGB is "good enough" for a working space, and 8-bit printing is good enough for Adobe RGB. He even admits to a benefit when using large-gamut working spaces. And the fact is many of us do use large-gamut working spaces, so his argument doesn't hold any water unless you've been drinking the Dan Margulis koolaid.  IMHO his argument against 16-bit printing amounts to pure spin. I think he would add it if he could (probably to the Studio version since he's intent on getting people to upgrade).

I can't see "hacking" the 8-bit GDI path as making much sense at all, especially when there's already an alternative printing path that supports 16-bit.

Qimage doesn't set 720, 600, 360, 300 PPI. It picks up an API call where the printer driver says what input resolution would be nice. That input resolution number is the result of the print quality settings you make in the printer driver and has nothing to do with Qimage settings. If you set a print quality level in the driver that asks for 600 PPI input resolution then Qimage obeys that request with its Maximum interpolation setting. It will deliver a lower input resolution if the interpolation setting is set to High (300 PPI), Medium (200 PPI), Low (150PPI). It will NEVER send an input resolution higher than what the printer driver asks for. I can check that easily in my HP Z drivers as they display the input resolution number directly next to the printer driver quality settings and those numbers always are identical to the Maximum interpolation settings of Qimage as shown above the Qimage preview window.

The links were to an article of 2006 and some message exchanges of 2003. Mike wrote in 2006 "Yes, sometimes it's hard to realize what you were missing until you see the new technology, but I'm not seeing any real benefit to 16 bit printing at the moment.  As technology on both ends (camera to printer) improves over time, I may have to revise my outlook in a future article.  :-)"
There have been changes like the 16 bit pipelines in today's OS systems. Printer gamuts increased, camera absolute gamuts much less though. The higher bit RAW output of cameras and better RAW workflows translated to 16 bit ProPhoto images edited on 10-12 bit driven monitor screens are matching today's printer gamuts. But there still is some value in his remark that good 8 bit data delivered correctly to a printer is hard to beat with higher bit solutions. The last are in a way more forgiving but not really delivering more quality.


met vriendelijke groeten, Ernst Dinkla

Try: http://groups.yahoo.com/group/Wide_Inkjet_Printers/

[Bart_van_der_Wolf]

He argues that you should be using Adobe RGB as your working space and that ProPhoto is too big. I disagree with that.

Maybe the axe that you apparently feel a need to grind obscures the fact that he doesn't say you should be using Adobe RGB. As a proof he even suggests to use a 47% larger than Adobe RGB space, so-called pRGB. Here is a comparison between pRGB and Adobe RGB, and between ProPhoto RGB (which is 140% larger than Adobe RGB and consists of coordinates that allow encoding non-existing colors).
[attachment=21634:pRGBvsARGB.png] Mike Chaney's pRGB (printer workspace) versus Adobe RGB
[attachment=21635:PPRGBvsARGB.png] ProPhoto RGB versus Adobe RGB

I personally would rather see him focus on print quality, including the features of newer printers such as 16-bit printing. But instead he's implementing things like RAW conversion which I just don't see the point of.

Sure, we all have our personal preferences, but I can understand that that may not necessarily correspond with what his customer base wants. It only shows that he listens to his customers..., and they have been asking for Raw conversion capabilities for years. When the benefits of printing in a 16-b/ch workflow become more frequent than occasional (e.g. because cameras have a larger gamut to begin with), he will probably be willing to accommodate that, just like my suggestion that anti-aliasing should be used for downsampling. I don't agree fully with his implementation of that, but he did allow himself to be convinced to address the issue.

So, if anyone has good arguments for the implementation of a 16-b/ch print workflow, he'll more than likely pick up on that. It's just that after all these years, such a workflow apparently is not mainstream and probably for a number of reasons already mentioned. But things may change.

Cheers,
Bart

[StephenOzcomert]

Thanks for the comments.  I had tentatively thought about printing in 16 bit from the plug-in when printing large images, but using QImage when I needed the functionality of nesting multiple smaller images on larger media.  This makes good sense to me as subtle differences, assuming they are even visible, are more likely to matter on a larger presentation than on smaller, usually more casual, images.

[marcsitkin]

Thanks for the comments.  I had tentatively thought about printing in 16 bit from the plug-in when printing large images, but using QImage when I needed the functionality of nesting multiple smaller images on larger media.  This makes good sense to me as subtle differences, assuming they are even visible, are more likely to matter on a larger presentation than on smaller, usually more casual, images.

Hi Stephen-

I use both methods. I use Q image whenever I am printing photographs because I like the sharpening I get from Q Image. I use the PS plugin when I am printing canvas or display media. I do use the 16 bit path, and have noticed a difference ONLY on images that have exhibited gradient banding when printing 8 bit. To over come this, it means going back to RAW and reworking the image as a 16 bit when moving it into PS. Printing an 8 bit image with banding in a 16bit path doesn't yield an improvement.

Both have their place.

[StephenOzcomert]

Hi Stephen-

I use both methods. I use Q image whenever I am printing photographs because I like the sharpening I get from Q Image. I use the PS plugin when I am printing canvas or display media. I do use the 16 bit path, and have noticed a difference ONLY on images that have exhibited gradient banding when printing 8 bit. To over come this, it means going back to RAW and reworking the image as a 16 bit when moving it into PS. Printing an 8 bit image with banding in a 16bit path doesn't yield an improvement.

Both have their place.

Thanks, Marc.  What color space do you use?  Do you use Profoto RGB?  If so, do you find banding only on your Profoto images or will you also find banding on images processed only using Adobe RGB?  I wonder about the benefits of anything more than Adobe RGB.  I had tentatively decided to use Profoto RGB but have changed my mind for the moment as I think Canon's documentation only seems to recommend Adobe RGB.  I guess if the printer's gamut is not large enough to distinguish between the two, then why bother with the larger color space?  Oh, and I like the sharpening and interpolation of Qimage too, but wonder about its limitation with respect to 16 bit processing.

Thanks again,

-Steve O.

[Mulis Pictus]

Hi Stephen-

I use both methods. I use Q image whenever I am printing photographs because I like the sharpening I get from Q Image. I use the PS plugin when I am printing canvas or display media. I do use the 16 bit path, and have noticed a difference ONLY on images that have exhibited gradient banding when printing 8 bit. To over come this, it means going back to RAW and reworking the image as a 16 bit when moving it into PS. Printing an 8 bit image with banding in a 16bit path doesn't yield an improvement.

Both have their place.

How does the reworked 16bit image look when printed from Qimage? I guess the banding might be introduced during processing of 8bit image and might be gone when processing in 16bit and printing through 8bit path.

Mulis

[marcsitkin]

Thanks, Marc.  What color space do you use?  Do you use Profoto RGB?  If so, do you find banding only on your Profoto images or will you also find banding on images processed only using Adobe RGB?  I wonder about the benefits of anything more than Adobe RGB.  I had tentatively decided to use Profoto RGB but have changed my mind for the moment as I think Canon's documentation only seems to recommend Adobe RGB.  I guess if the printer's gamut is not large enough to distinguish between the two, then why bother with the larger color space?  Oh, and I like the sharpening and interpolation of Qimage too, but wonder about its limitation with respect to 16 bit processing.

Thanks again,

-Steve O.

I use Adobe RGB for commercial work and processed images supplied to me, and Pro for my own images processed from Raw files. Except for a very few images, I print my photographic work via QImage, and use Canon for commercial projects. I wouldn't get too hung up on the 8bit printing. Except for images prone to banding (in gradients), the biggest advantage of 16 bit is the extra information available when applying adjustments.

[marcsitkin]

How does the reworked 16bit image look when printed from Qimage? I guess the banding might be introduced during processing of 8bit image and might be gone when processing in 16bit and printing through 8bit path.

Mulis

I can't really say. Last time I haad the issue I was printing via Canon. Haven't done a comparison.