Uprezzing - is it really worth it?

JoeSchmoe007

Leading Member
Messages
568
Solutions
1
Reaction score
213
Location
US
I recently sent my 8 Mpixel photo from Rebel XT in native resolution to be printed at 20"x30" at Elcocolor. I didn't do any uprezzing myself - it was done by their printer driver, and I was completely satisfied.

The problem I see that everywhere on the websites of the companies that make/sell these plugins they compare one uprezzer plugin with the other. But this is not what they should compare IMO.

Why would we possibly want to uprez? To improve print quality. So one should print the oringinal image with printer driver as uprezzer, then uprez the image with plugin and print it at the same size at the same printer, then scan prints and post them for us to pixel peep.

Is there a scientific review of the uprezzers with resulting prints scanned and compared?
 
Have a look at this thread:

http://forums.dpreview.com/forums/read.asp?forum=1003&message=7642362

The reason you do it of course is that there are more refined methods of interpolating than the printer driver uses. Ideally you want to use good interpolation software such as Genuine Fractals or Qimage and use this software to get the final resolution the actual printer will use.

Ron
 
Is there a scientific review of the uprezzers with resulting prints
scanned and compared?
At a quick glance, the only thing I found was a comparison of interpolation methods at...
http://www.ddisoftware.com/testpics/resample.jpg

Save that JPG and then go into a photo editor to examine the test results at 100% or more. You could even try printing that JPG as then you would have the tests all side by side on a bit of paper, but I do think the JPG pretty much says it all. The Pyramid and Vector interpolation methods are apparently Mike Chaney's creation.

That link above was found from the Mike Chaney article at Steve's Digicams...
http://www.steves-digicams.com/techcorner/January_2005.html

where Mike makes more sense than most when he writes about interpolation. His Qimage program is hard to beat.

The summary of his advice is never to interpolate your original image, only do it once at print time to suit the print size, and that's where Qimage does it automatically for you.

I've experimented with using the "remove jaggies" option (Effects- Image Optimiser) within the Canon printer driver and it does quite a good job, pretty much as good as Qimage for normal images and not too high a degree of interpolation. Qimage shines when there is a large degree of interpolation but you may need to choose which interpolation method and fine-tune the individual interpolation method to suit the image type.

Think in terms of trying to print a big print that would yield maybe 25 or 50 original camera pixels per inch of print and Qimage does a good job. Of course close examination shows the image to be soft but the big print viewed at the correct distance would be fine.

For home photo printing, remember that whatever dpi file you send the printer then the printer driver will always re-interpolate (if necessary) to what it requires internally. Epson uses 720 dpi and Canon and HP 600 dpi. Qimage takes care of that fact automatically and feeds the right dpi file to the printer driver.

In fact it's a good idea to sniff around http://www.ddisoftware.com to read all you can about how Qimage works. You can learn a lot about printing that way. The growing series of articles at Steve's Digicams is probably an up-to-date summary of Mike's thinking about this subject.

Sending to an outside printing lab you would have to talk to their operators to find out in what form to send the files, what dpi etc.

I must admit I was tempted once to do all that printing testing and then rescan the prints for comparison purposes, but I now think it's a matter of wasting time and materials over something that often doesn't matter. I use Qimage and I NEVER have to think about interpolation at all.

Regards......... Guy
 
Artistic taste is important. It is much easier to judge if you will be happy with something in the print rather than on the screen (until sufficiently experienced).

The quickest way to compare would be to choose just one of the higher order methods, shown in the very useful page linked in the previous post, and compare prints made using that method to those printed with interpolation done in the printer driver. Avoid sending "bad" sizes during the test, or test that first.

If you see no significant improvement (in your view), with one one of the higher order methods, you will probably not see much difference with any of them. The one thing to look out for here is that some of the methods include significant sharpening, and that can confuse the issue. So you could also send a slightly sharpened, non-interpolated version to compare.

Do this with a couple of normal prints, view at normal distance. It is a mistake to optimise your prints for unrealistic viewing conditions. If you make a range of print sizes, try large first (but not larger than you normally make for real).

If you do it yourself you will have certainty of the outcome for your style.

Ken
 
Joe, I see you have an HP printer. At Maximum DPI it will uprez your image to 1200 dpi, so your best quality way of printing with the HP is to use Qimage and it will take care of this for you - at the cost of relativelys slow printing. 1200 dpi makes for a very large print file.

Ron
 
Joe, I see you have an HP printer. At Maximum DPI it will uprez
your image to 1200 dpi, so your best quality way of printing with
the HP is to use Qimage and it will take care of this for you - at
the cost of relativelys slow printing. 1200 dpi makes for a very
large print file.
Does the HP printer driver really interpolate the input file to 1200 dpi? Surely that's just the print engine resolution setting that has nothing much to do with the input file.

I've never played with HP but my Canon experience (with Qimage) shows that the input file is interpolated only at two different dpi, 300 for lower quality settings and 600 dpi for any photo type setting - and that is when the printer is claimed to be outputting at 2400 or 4800 splats per inch (or whatever, I don't take too much notice of the printer claims, I only select the slowest and best quality settings). Of course when doing borderless the 600 dpi changes to funny values like 614 dpi.

Those dpi input figures are seen on the Qimage screen when you have selected printer type, mode etc.

I would have thought that HP would never need a 1200 dpi file, only a 600 dpi file as the 1200 would make for some big temporary files and possibly slow printing. Do you see that 1200 figure with an HP when using Qimage, Ron? Anybody?

Regards.......... Guy
 
HP is a little unique in the way it handles resolution. They have what they call Photo REt which is now up to level IV and Pro. In this mode it only uses 600 dpi but I think up to 32 droplets per dot. This is their higher speed mode. The other mode I believe uses a more traditional matrix of up to 1200x4800 or so. The Photo REt mode is called Best, and the 1200+ called Maximum. In Photo REt Qimage interpolates to 600 and in the Maximum mode it does 1200. While Photo REt is much faster, in side by side comparison I see a better print with the Maximum mode, and use that unless I am in a hurry. See link below which is a little dated, as it still only talke about level III.

http://h10025.www1.hp.com/ewfrf/wc/genericDocument?lc=en&cc=us&docname=bpu03070&dlc=en&lang=en

http://h10025.www1.hp.com/ewfrf/wc/genericDocument?lc=en&cc=us&docname=bpd08770&dlc=en&lang=en

Ron
 
Thanks for that information Ron, I learnt something new. So the 1200 dpi interpolation is done and you obviously need a decent PC with an HP printer.

Regards............ Guy
 

Keyboard shortcuts

Back
Top