Three additional things need to be known before good advise can be given; The type of image content (e.g. modern archtecture, or a busy landscape, or etc.), the current file size in pixels (width x height), and on which printer is it going to be printed. The first might dictate a preferred algorithm to use
instead of a generic one, and the others will determine the amount of upsampling required which may further point in the direction of a specific upsampling algorithm or settings.
Ok image file size is 3240 x 2160
The image is of a wave breaking, pin sharp water shot, non regimented lines.
Printer is a 9880 epson.
Thanks for all the input crew.
Okay, so we need to go from 3240 pixels to 2.4 metres @ 360 PPI = 34016 pixels. Lower PPI settings will cause the printer driver to upsample to 360 PPI, with an algorithm that's probably more geared towards speed than accuracy (because the processing power isn't there in the printer's hardware). That is a magnification of 10.5x. That also means that any artifacts that the interpolation algorithms introduce will become quite visible, because the interpolation will usually not create resolution, but just smooth transitions between existing pixel values. Therefore artifacts will stick out like a sore thumb against expected smooth transitions. Human vision is quite good at spotting anomalities that do not follow the expected pattern.
'Fortunately' there will be many more interpolated pixels than original pixels, so the artifacts will also be a bit fuzzy.
Here is a crop of 300 x 300 px, hopefully with enough different types of detail and slanted edges to reveal shortcomings from resampling. It was upsampled to 3150 x 3150 px, the same magnification a full sized image would undergo, given the above criteria:
Attached are 3 upsampled versions:
1. A Photoshop 6 upsample with BiCubic resampling.
2. A Lightroom 4 upsample, presumably using BiCubic Smoother (and virtually indistinguishable from a PS Bicubic Smoother resampling).
3. A PhotoZoom Pro upsample, Sharpness dialled down to 10% instead of 100%, and an amount of 30 grain added.
No specific output sharpening was applied to any of the files.
When printed at 360 PPI and viewed from 3.5 metres, the differences (and the artifacts) will probably be hard to see, but from a bit closer there is a distinct different in acutance.