Resizing is just that; it resizes the image keeping the number of pixels. Resampling is the added option of changing the number of pixels within that image.
Typically, you'll want to leave resampling out of the equation when printing as it's primarily used for onscreen (monitor) viewing.
I think resampling has more uses than for onscreen viewing. I just finished a job where I had nothing but a 30 kb 2 x 3 inch image to work from. It had to be resampled so I could clean jpeg artifacts to make it ready for tracing in Illustrator (it was a logo). I can think of many times where I have had to resample either using Perfect Resize, or Photoshop (CC in my opinion does a much cleaner job of resampling than previous versions). So my answer to Birdofillomen would be that resampling can actually be essential to achieving a printable image size, especially when going from 72 dpi to 300 or even 360 dpi. Certainly, as Warunicorn said, it is best to avoid resampling when possible, especially if you are starting with a high dpi orginal where you can simply reduce the dpi to enlarge the image for printing. My rule of thumb when not resampling is that you can get away with reducing dpi to 200 or even 180, depending on the nature of the print.
dang, still a little confused.
okay, so using the example i have, with re-sampling unchecked...
if i resize it to 13, the dpi changes to 70 and the pixel dimensions stay the sam.
i thought a lower resolution resulted in a lower quality print?
but if i change the resolution to 300dpi, the image size goes to about 3x4, which would be way too small if i wanted to use part of this on a poster size print (something like 11x17)
is my only option to make it larger, starting a new file at 11x17 and then dragging the 3x4 into it and then enlarge it by transforming? or, do i pretty much have to sacrifice one or the other (dpi or size)?
i guess while i'm asking, what sort of a diff does the channel size make (8bit/16bit/32bit)?
and any insight into "save as web or device"?
sorry for so many questions, i hate not having someone hands on to ask the questions to. thank you for the replies.
just to clarify the point of the example i'm using...
let's say i'm going to use part of the knight as a composite on a larger canvas. i'm basically wanting the image to be as "optimal" as it can be, and large enough so that it can be further resized larger (if need be) without seriously compromising the quality. i know that it's easier make a larger image bigger than it is a smaller image.
i guess this way of working helps me keep the number of variables down. like if i take the helmet and put it on the image of a woman in bikini. even if the images are different sizes i should be able to scale them until they are proportional, and then if there are any issues i know that it probably has something to do with the original source and not me.
does any of that make sense?
honestly, i'm not sure if my way of thinking is ill thought or not. i'm assuming if i have an image made up of several other images, all at varying resolutions, the final product will be a mess.