The Best Approach To Printing?

Posted by ciaran on August 22, 2010 in Blog, Editing & Workflow | 9 comments

Note to the “non techies”… this post probably isn’t for you

Like most things in life, you can make things as simple or as complicated as you like. Most of us tend to stumble upon an approach that works for us; a preferred method for doing things; a solution to a daily problem… that sort of thing. Well my print flow was certainly something I stumbled upon, something which I’ve followed consistently since I bought my first printer and to a large degree something that I’ve been reasonably happy with. Ok, for sure there was the recent debacle related to colour space issues, but leaving that aside, my prints have been of pretty good quality (even if I do say so myself) and have done well in internal club competitions, national competitions and international salons. But, was my approach the best approach?

One thing that got me thinking quite a lot about printing was a visit to my camera club from Steve Crozier, who owns and runs his own printing, mounting and framing company. His talk raised more questions than it really answered and by in large, I just filed his comments and suggestions away in the back of my mind for a rainy day. But over the last few days, I’ve taken the dust cover off the printer and started to print some images for upcoming salons, most notably the Irish salon hosted by Tallaght Photographic Society, which closes for entries at the end of this month.

Anyway, I decided to do a bit (lot   ) of research, primarily by reading through the excellent Epson 3800 resource developed by Eric Chan. Taking his comments, suggestions and advice on board as well as making use of the suggestions from Steve, I decided to do a few test prints. Effectively I was printing the same image, but using differing approaches to see which is the best print process, particularly in respect to DPI and whether to let the print driver interpolate data or to resample the data prior to printing. But let’s start at the beginning…

The image above is an image I took recently of UK model, Madame Bink. It’s more or less a full sized image, with a certain amount of cropping applied, necessitated by the need to straighten some edges. Looking at the basic information, we can see that:

• The image is 2626 pixels wide by 3716 pixels high.
• The native resolution of Epson printers is 360 pixels per inch (PPI) assuming the “Finest detail” setting in the printer driver is disabled. When importing my images from RAW, I assign the image resolution to be 360 PPI to correspond with this resolution.
• At 360 PPI, an image of 2626×3716 pixels, will print at approximately 7.294 inches by 10.322 inches. These figures are obtained by dividing the pixels by the print resolution (2626/360=7.294 & 3716/360=10.322)

If I was to press print at this point, the data being sent from Photoshop to the Epson print driver matches it’s native resolution of 360ppi. The good thing about this, is that the print driver does not have to interpolate the data in any way, so it literally just prints the data as is and will produce a print which is 7.294 inches wide x 10.322 inches high. BUT… What if I want a bigger print? What if I wanted a 12 inch print on the longest side? How do I do it?

The first option is to have the print driver interpolate or up-sample the data. To achieve this, disable the “Resample Image” option in the “Image Size” Dialog box, as is shown above. This ensures that no changes are made to the physical pixel data within an image. I now change the longest side, in this case the height, to be 12 inches.

• The physical pixel dimensions remain the same at 2626 pixels wide by 3716 pixels high.
• Because the height has been changed to 12 inches, the width scales by the same amount, maintaining the original aspect ratio and changes to 8.48 images.
• There is now a drop in print resolution, from the original 360 to 309.667. This figure is obtained by dividing the pixel dimensions by the print size (2626/8.48=309.667 & 3716/12=309.667)

So if I were to print this image, I would end up with a print of only 309.667 pixels per inch, right?? WRONG In this case, even though we would or indeed should expect a print of only 310 (rounded up) pixels per inch, we in fact still get a print of 360 pixels per inch. This is because the print driver will take the 310 pixels per inch and interpolate them updwards, effectively adding pixels into the print where there are none in an attempt to create a sharp, high quality, yet larger print.

Obviously the degree to which you need to interpolate will really dictate how good the final print will be. In this case I went from 310 to 360, which whilst not insignificant, is not a massive amount. One thing to consider if you try this approach, is that the print driver is 8-bit, so any interpolation is carried out on 8-bit data rather than the 16-bit data you are probably originally printing from.

A second option is to ensure that the data is edited and prepared and printed at 360 ppi. This actually involves you interpolating the data yourself within Photoshop, prior to printing. As you can see from the image above, I have now enabled the “Resample Image” option in the “Image Size” dialog box. Once this is enabled, and with the 360 ppi resolution already set, I change the longer side to 12 inches. Unlike the previous method, where the resolution dropped, in this case the resolution stays fixed at 360 ppi and the actual pixels are changed by the image being resized up to 3053 pixels wide by 4320 pixels high.

• There are various options for resampling. For my test case (even though it is shown differently) I used Bicubic Smoother.
• The image size actually changes, the pixel data changes and once this is applied and saved, this can not be undone
• The 3053 pixels is arrived at from by the width of 8.48 inches multiplied by the print resolution of 360 and similarly the 4320 pixels is arrived at by the height of 12 inches multiplied by the 360 ppi resolution.

I guess the advantage of this approach is that you control the resampling of the data and are most likely operating on 16 bit data. You get to see what the image is like following the interpolation step and if necessary, you can try different resampling techniques. Indeed, following on from the resampling you may also decided to apply a degree of sharpening again, to some or all of the image. By doing so, you are in control of the image and the 8 bit print driver is not required to resample. However, the major disadvantage is that as you are changing pixel data, as you print to different sizes, you will either need to keep a master copy which is resized each time for each print, or you will be rezizing up and down the same image over and over (not a good idea).

• SUMMARY

I tried both methods on two different images. The image of Bink on the stairs required resampling, either by the print driver or in Photoshop,  from 310 ppi to 360 ppi in order to get a 12 inch print. The image of Bink in the doorway above, required a much smaller degree of resampling, going from just 352.9 to 360 ppi. On the second example, I wouldn’t have expected to see much difference in either approach because the degree of interpolation was so small. However for Bink on the stairs, I was expecting to see a notable difference. The truth be told, to my eye, I couldn’t perceive any difference either print, printed by either approach

So the moral of the story is this. Whilst I’m happier that I’m more aware of the print process and I feel more informed, I’m also happy to stick with the approach I stumbled into when I started printing and that’s the approach of letting the print driver handle the interpolation. I’m sure the purists may disagree, but the fact is, if I can’t see the difference in front of me, then there’s no real reason to change from what I know. Is there?