Boy am I glad I did this! It helped with my original question about sizing algorithms and also confirmed something I though about IP. I don't think it will be meaningful to post images of the prints since they will go through multiple gyrations from camera to web. You will have to rely on my descriptions.
I took Bart's rings and placed them on a 13x19 custom layout in LR at various sizes:
1.389" (720 ppi)
2" (500 ppi)
2.778" (360 ppi)
4" (250 ppi)
8" (125 ppi)
I printed it on my 7900 three ways:
- Through LR without "print resolution" checked so it went to the printer @ native sizes listed above.
- With "print resolution" checked at 360ppi so LR resized everything to 360.
- I imported the file into PS and converted it from png to tif without resizing (I don't think IP will print a png). Then I opened up IP and dropped the image into a 13x19 sheet and resized them "on the fly" to the same sizes, just like Kevin did in the video.
What I see:It is obvious that Schewe's recommendation in LR to check that little "print resolution" box improves results of this test dramatically, especially when upsizing the image. The two small images, 1.389" and 2", look close, but you can see a few "extra" rings in the 2" image printed natively (print #1). The 2.778" image looks exactly the same either way. But the 4" and 8" images have quite a few added rings when native resolutions are sent to the printer. By my count 24 extra rings appear on the 8" image (printer upsized from 125 to 360).
As for the IP test, the output looks exactly like #2; printing through LR with "print resolution" set to 360. The main difference is the IP version is more neutral, which is of course why I like IP in the first place for B&W printing.
So this test supports Kevin's workflow. There seems to be no reason to output images to the exact size for IP. Sharpening may be a different thing, but that will be a tougher test for me to think about.
There is a "but." I've always felt that IP's screen renderings, while very color-accurate, are not very detail-accurate. Attached is a screen shot of what these rings look like in IP. Just ridiculous. This may be because I am on a MBP retina, but it looks the same regardless of whether I move it over to my NEC PA241w or on the MBP screen.
So when I get to the sharpening question, I won't be able to judge anything from the screen. It will have to be 100% via test prints. Edit: When you go into the sharpening mode in IP, it does render a very nice accurate image.
Thanks again Bart, the image really helped.
Dave
Note: I used Canson's canned profile through LR and IP's profile through IP. I did not use my own profile through LR.
Hi Dave,
The attached file will tell you if down-sampling is done well, non-linear gamma can cause artifacts at 100% size.
Cheers,
Bart