Scaling and sharpening photos in Illustrator

Question

What actually happens to a photo when it's scaled in Illustrator? Does it get resampled?
I'm asking because in Photoshop you need to have an image which will be printed to be the actual size it will print before applying an Unsharp Mask.
Scaling it afterwards will affect the sharpening.
So, what happens when a larger photo is scaled down in Illustrator? Does that photo get resampled? Does it remember its original size?
Is it safe to apply Unsharp Mask to a photo in Illustrator after it's been scaled down? Would the sharpening be applied to the photo at its new, smaller size or would the calculations be applied as if the photo were still larger?

Answer 1

I'm asking because in Photoshop you need to have an image which will be printed to be the actual size it will print before applying an Unsharp Mask.
Scaling it afterwards will affect the sharpening.
Not exactly.
Scaling a raster image that has been sharpened does not affect the sharpening. Resampling the image, of course, will.
The distinction is important. The printing process itself effectively "resamples" a raster image to create a new one (actually several new ones; one for each color separation plate) that matches the printer-spot grid of the output device. (That's what's meant when you hear people say that eventually, everything gets rasterized.)
That's why oversampled images tend to somewhat negate carefully-applied sharpening. The "resampling" that is inherent in the printing process imposes a certain amount of "blurring" as the too-many pixels are averaged together to make a single tone dot and those in turn are resampled into individual printer spots. For an over-simplified conceptual example, consider:
An image has a light pink pixel, next to another light pink pixel, next to a bright red pixel.
Sharpening the image results in a pink pixel next to a white pixel, next to a bright red pixel.
The image is scaled to too-high a ppi for the printing method being used.
The printing method effectively results in a light pink halftone dot next to a darker pink halftone dot. The white sharpened edge is lost, as is the brilliant solid red--you've effectively averaged-away your carefully-applied sharpening.
So the bottom line is, whether you do the sharpening in Photoshop (as I would) or elsewhere, there is no "quick and easy" avoidance of performing the sampling on an image that has a pixel count appropriate to the intended reproduction method. This is one of the reasons why I maintain that the conventional-wisdom "one-size-fits-all" admonition to make all printed images 300ppi is a corruption and misunderstanding of the original and legitimate rule-of-thumb, which is "one-to-two times the halftone ruling."
So...
After doing all your color correction and converting the image to CMYK:
Resample to a PPI appropriate for the intended printing process.
Then sharpen. When sharpening, do so while VIEWING the image at at least a 1:1 monitor-pixel-to-image-pixel zoom. That way, you are at least viewing your monitor's rendition of all the pixels in the image, not just a temporary on-screen downsampled image (therefore, pixels averaged together).
Thereafter, merely scaling the image does not resample the sharpening--unless, of course, you scale the image such that its ppi exceeds the rule-of-thumb of "one-to-two-times the halftone ruling."
(That's why, for the typical 150lpi sheetfed halftone ruling, I use 225ppi as my default "ideal"; not the oft-cited 300ppi-fits-all. But I size/rotate/sample/sharpen images before placing on the page, and don't do on-page transformations thereafter.)
JET