I cannot answer what method of upsampling PDCU 5 employs so failed at the first hurdle :-). But some of this was covered in this post, so perhaps the question germinated there
New K1 owner wants to print LARGE - PentaxForums.com.
Not to try to teach granny to suck eggs but some information that might be of use
Conventional wisdom on this subject is that if you want to retain optimum IQ then you do not let the printer manage either colour or upsampling.
Dealing with the case of upsampling the reason is damn simple. The printer driver will upsample to its required PPI either 300 or 360 Canon Epson respectively regardless of what you may think you are sending it and therefore what you think it is printing.
Send it an image at any PPI you like for a particular print size and it will upsample or downsample as required to 300 or 360 PPI and produce a proprietary bitmap to send to print. It does this by using nearest neighbour algorithm. Nearest neighbour is the simplest and fastest algorithm but the results are very poor compared to what a dedicated application can achieve including LR and PS.
Photoshop offers 5 methods of resampling:
Nearest Neighbour Basic, and very fast: To create a new pixel the value of the pixel next to it is copied.
Bilinear is more complex and generally produces better quality. Colour or grey values are set according to the colour of each pixel surrounding
Bicubic creates better effects than both the above but takes longer. It looks at surrounding pixel but the calculations is much more complex and intesive and produces smoother tonal gradations.
Bicubic Smoother is a newer interpolation method that has been designed for upsampling giving a smoother result which handles sharpening better than the others. Its alter ego being
Bicubic Sharper but this one designed for downsampling and preserving detail better than Bicubic
There is one new method in later versions of PS Preserve detail I am not 100% on this so not covered.
So the question is can you really see the difference between these algorithms in print. The answer being an absolutely, definitely maybe.
We know that Bicubic Smoother will give a better starting point and allow an image to handle subsequent sharpening the best certainly better than Nearest Neighbour which the print driver uses.
But we need to be presenting an image resolving great detail to maximise the benefits. If we then throw in theoretically correct viewing distances then the added quality is likely not to be seen (although it would be there in the print if looking closely)
Quick example viewing distance calculated on 24x36 print = 43"
EDIT: This not really correct as the viewing distance is normally 1.5x -2.0x diagonal therefore viewing distance actually around 65"
PPI required for a viewer with 20/20 vision = 80 ppi:.
EDIT: Based on the revised viewing distance above PPI needed = 53
Therefore we could easily get away with just sending image data for most DSLR's and get an OK print, but the image data at 80 ppi is going to be upsampled for a Fuji 570 printer by the driver to 300 ppi using nearest neighbour. So why not upsample first to 300 using the superior Bicubic Smoother or better?
Suppose that instead of viewing this print at 43" a photographer calls and wants a look at 12", what pixel density do we now require? The answer would be around 286 ppi our upsampling should score over the printers own for the reasons highlighted.