I decided to do my own testing today to simplify my workflow with the plugin and make sure I am getting the best quality.
Test material:
1) Canon 5D image taken on a tripod with mirror lockup
2) Capture sharpen with Outback Photo EasyS Sharpening Toolkit using Low Sharpening, halo control, new layer with mask
3) Set sharpening layer opacity to 50% (sharpen to taste at 100% view) and flatten the image layers.
Test setup: Start with base image in Step #3 above. Output sharpening was done with PK Sharpener. All printing was done through the iPF5000 Export Plugin to Epson Premium Luster paper.
Print 1: Resize Bicubic smoother to 14 X 21 size at 600 PPI, output sharpen at 480 Glossy, flatten, print
Prints 2-4: Resize Bicubic smoother to 6 X 9 size at 600 PPI, output sharpen at 480 Glossy, flatten (this is used as the base image for remaining test prints), then:
Print 2: Resize Bilinear in PS to 14 X 21 at 600 PPI, print
Print 3: Resize Bicubic smoother in PS to 14 X 21 at 600 PPI, print
Print 4: Resize Bilinear in Export Plugin to 14 X 21 at 600 PPI while printing
Print 5: Resize Bicubic smoother to 6 X 9 size at 600 PPI, output sharpen at 480 Glossy, set opacity of output sharpening layer to 60%, flatten, Resize Bicubic Smoother in PS to 14 X 21 at 600 PPI, print
Notes: Prints 2-4 differ only in the second upsizing method used, and whether done by PS or Export plugin. Print 5 is the same as Print #3, except that the opacity of the output sharpening layer is set to 60% before proceeding with the rest of the sequence.
Results:
--Print 1 (my previous method) was judged to be the least sharp, and probably a bit under-sharpened. Note that the difference was subtle, but noticeable in a blind test.
--Prints 2 and 4 (both bilinear resize, one done in Export Plugin, the other in PS) were judged to be identical to my eye. The advantage of doing the resizing in Photoshop is that you can specify an exact size in PS, but have to work with percentages in the plugin.
--Print 5 was a bit sharper than Prints 2 and 4.
--Print 3 was the sharpest, but to me had a subtle artificial or "digital" look.
Winner (to my eye): Print 2, although could go with Print 5. Print 2 looked more natural to me, with Print 5 having a very slight "digital" look. Either could be the best, depending on your taste.
Conclusion: I have verified that to my eye, the procedure recommended by Marcalmont produces the best result (and also takes about the least computing resources). However, I prefer to do the final bilinear up-rez in Photoshop.
--John