This is going to be a two faceted post. Based on the title you can assume some magic secrets pertaing to Aperture 3's sharpening tools. Well I will comment on those but to give you some context before I get to Aperture, I will discuss sharpening in general. Althought sharpening as pertains to digital imaging has been discussed ad nausium, I still find the vast majority of digital photographers really don't have a good handle on what's what in the world of sharpening based on the number of questions I get in comments and even more in emails.
Without going too far into the weeds on the history of sharpening we should all start with the rudimentary idea that sharpening in the digital world was thought of as nessesary to counteract a lot of de-sharpining inherent to the digial process. Be it in degrading effects of scanning images or AA filters on most direct digital capture devices. Over time it has morphed more into an effect that pervades all of our visual vocabularies. We don't even see it as an effect anymore. If you explore prototypical images over the last 20 years you can clearly see the massive increase in edge contrast evolve not-so-slowly in just about all of the normal images we consume in our daily lives.
I am not talking about some massive increase in detail - just edge sharpness. With that in mind understand that at the end of the day - how your images look, and how well they fit what you are trying to convey win out over any right or wrong way, or forumlatic approach. If it looks right to you then it is right. My purpose here is to give you some idea why your images might not look like you want them to and put you on the path to get there.
Fast forward - it has been said many times that the amount of sharpening depends on how big your image file is. While somewhat true, I think that this is highly misleading because it separates all of the important context from the bigness of the image file. The really important part is not really the image file size, it's the resolution of the output medium as well as some of the more minor characteristics related to that medium.
Take that in for a moment to understand why sharpening treatment should be way different for an image meant to display on a typical 75-110 DPI computer screen vs a 240-300 DPI printer or even a Durst with ColorByte RIP at 400 DPI. For the most part you don't need to worry too much about the difference of 75-100 DPI however the difference between 240 to 400 DPI on a printing device could be significant.
The reason that the target output and viewing resolution of the image is so important is that it and it alone determines the actual physical edge size that you are manipulating with sharpening. A line drawn 1 pixel wide using 75dpi pixels is clearly visable. A line drawn 1 pixel wide using 300 dpi pixels - not so much. To have the same sharpening effect that you see on a 75-100dpi output device on a 300dpi printing device the edge that you are sharpening has to have the same physical width. Get it?
Virtually all sharpening tools alow you in one way or another to control the width of the edges that you are *enhancing". Usually this control is called the radius. This is the most important parameter and what settings you may want to use are completely dependent on the output resolution of your target device to achive any given look you have in mind.
If we cut to the chase a reasonable range of values for screen out put is anywhere from 0.5 to 1.0 pixels. The value you choose has more to do with how you want the image to look vs any right or wrong standard. When it comes to printing there are many other variables but the edge width or radius that you choose again has everything to with the target mediums output resolution. As a way to speak generally about printing, lets assume that for any give size you will be sending a file that has 300 DPI to the printer. Reasonable values for printing are somewhere in the neighborhood of 2.0 - 3.0 pixels. The reason is again to make the edges have a specific physical width on paper.
The variables you have to deal with in printing have to do with the actual material that you are printing on, the target viewing distance of the audience, and to some lesser degree the characterisitcs of the printing device itself. A lot of these are more of a fine tuning exersise than critical elements and can only be understood for your own personal goals via experience.
Moving on to the other basic controls, the venerable unsharp mask type enhancements give you amount and threshold in addition to radius. The amount parameter being how much contrast you want to had to the edge width defined by the radius you chose. The threshold is for determining what level of contrast is actually and edge. Threshold is a crude way to avoid sharpening things that you may not want to emphasize fine texture, noise, and film grain. The newer, more better, sharpening tools add more sophisticated ways or algorithms to avoid sharpening things you don't want emphasised but masking the good ol' USM is just as effective if not more so.
I hope that you can extrapolate a few things from the preceeding illumination. If not, here they are. For the best possible results your work-flow should look something like this:
- Resize your image to the target output medium and display size
- Sharpen choosing the radius range based on the output resolution and the other parameters how you want your image to look.
In the case of sharpening for screen ouput you will typically choose the physical pixel demensions as the size, as in 1024, 800, 600, etc. Do the resize, Sharpen at 0.5-1.0 radius. For printing you will typically choose what print size you want and a resolution optimized for the printing device, 240dpi, 300dpi, 400 dpi, etc. Then sharpen with a radius from 2.0-3.0 prior to sending your image file.
The whole goal here is to avoid as much as possible resampling the actual sharpening. In most cases downsizing a print sharpening treatment for screen display will not look that great. On the printing end it is also preferable not to sharpen a gigantic file meant for a 20x30 print at 2.0 pixel radius downsampled for a 8x12 print. Of course those results may be good enough for quick and dirty work but generally will produce inferiour results to sharpening a file that is sized approriately for the output intent.
Okay, now we are at the Aperture 3 part. You can probably predict my feelings at this point about the tools that Aperture 3 provides in terms of sharpening. There is nothing wrong with them but in my opinion they are way too cumbersome to use for anything but quick and dirty casual work. This is purely due to the fact that there is no way to have the sharpen effect take place with the appropriate parameters after the output resize. It always happens before the resize and then is upsampled or more likely downsampled for exports and printing.
The only way to deal with sharpening in Aperture is by trial and error, messing with various prameters in the sharpening adjustment block(s), until you find settings that happen to look good at various ouput sizes. For me this is way to fiddly, error prone, and time consuming for anything remotely critical, including producing web output that looks good for clients. It's fine for crap I don't care about that I post on my blog - for clients and images that need to look their best - not so good. I do all of my real work resizing and sharpening in Photoshop.
If most of your work is casual and you really want to do it in Aperture alone there are a few reasonable ways to go about it. One is to make presets that look reasonable for various output devices and sizes by trial and error. Then use those presets on different versions of the image target for those particular outputs. Another way would be to rezise and export the image as a TIFF, import the resized image back into Aperture, then sharpen that image for its intended outupt device, finally exporting it at that same size. Last but not least - a tool like Nik Define attempts to take all these things into account for you based on the target medium and output size. Unfortunately I sill think that this plug-in is too fiddly and does not deal with most real-world work-flows in reasonable ways.
Please take all this with a grain of salt. I am being ultra-finiky here and trying to be give you as much information on the important variables for you to know how you may want to approach image production across a lot of different circumstances. The main point to take away is the criticality of the radius paramerter is with respect to the output resolution in determining the real physical edge size that is ultimately produced. This more than anything controls how the perceived sharpness of your images will actually look.
Ps. Sorry for all the wordi-ness.