How to apply a 'calculated' radius in Unsharp Mask en mass?

Question

In about a week's time, when I've finished applying Curves to about 1200 B&W images, I'll be ready to apply an Unsharp Mask to each of them. Exactly how I'll do that I'm not certain yet, but I've been reading my Photoshop CS Bible (Deke McClelland) in preparation. On page 503 it states:
If you're looking for a simple formula, I recommend 0.1 of Radius for every 15 ppi of final image resolution... If you have a calculator, just divide the intended resolution by 150 to get the ideal Radius value.
For example, at 300 ppi use Radius = 2. Such a recommendation is just that: a recommendation (and not 'ideal' as he also states), but I've taken it as a good starting point; and while I've been adding Curves I have also been experimenting with Unsharp Mask and various Radii (but not yet saving the sharpened images).
McClelland's statement is not clear regarding what Radius should be used if you are NOT working on the image at final resolution. I think I know what the answer is, but I want to confirm.
Some background: I'll be printing on a Xerox iGen running a colour line screen at 175 lpi. The printer wants the images to be 300 ppi. The images at the moment range from an effective ppi of about 100 to 2000, depending on how much they have been scaled in InDesign. So I've got resolutions all over the place which at some point I will have to tidy up. Those images below about 250 ppi, I will be upsampling to 300. Those above about 450 ppi (this may change), will be downsampled to 300 when I convert to PDF.
I have some questions:
QUES 1
Say I have an image at 1200 ppi, that will be downsampled to 300 ppi. Should I do the sharpening on the 1200 ppi image, or on the downsampled 300 ppi image? i.e Will it make any difference to the final printed result if sharpening is done before or after downsampling?
QUES 2
I think the statement:
I recommend 0.1 of Radius for every 15 ppi of final image resolution
should read
I recommend 0.1 of Radius for every 15 ppi of image resolution
Surely the Radius (assuming you accept the figures McClelland gives) depends only on the resolution of the image you are working on. i.e if you have a 1500 ppi image, use a Radius of 10 (1500/150), NOT a Radius of 2 (300/150) -- the final image resolution that is sent to the printer. Is that a correct reading of McClelland's recommendation?
QUES 3
My experiments indicate that for the type of images I am using, McClelland's recommendation oversharpens, at least for my liking. I will probably use half the figure he suggests: Radius = 1 for 300 ppi.
Given that I have about 1200 images, if I was going to apply an Unsharp Mask manually this is what I would do:
1. Put all the layers in each image into a Group.
2. Calculate the Radius to be applied to each image from the formula: Radius = (IMAGE PPI) /300.
3. Apply an Unsharp Mask to the Group using the calculated Radius (plus Amount = 50, Threshhold = 0)
Of course I'm not going to do this manually -- well, I hope I'm not -- but what is the best way? Could an Action handle all of this? A Script? A combination of both? Something else entirely? Not possible in PS?
In summary: I want to automatically apply a variable Radius that depends on the image's resolution. Possible?

Answer 1

The PDF says, for one of the steps:
Command-click the duplicate channel to make the edge mask a selection.
When I do that, a selection outline appears, but I can't work out, or find out, what conditions are applied to make the selection. Any ideas? I may have to start a separate thread for this question. I don't like to blindly follow recipes. I like to know what's going on.
Guy,
I'm not sure I understand your question. With command click, the channel is the selection. White = fully selected, black = not selected. It's all the steps made to create the channel that affect the selection. To me the selection edges are visually interfering with a soft mask, I usually hide the edges.
In post 9 steps 2-6 all contribute to the final section. Find edges identifies the edges. Median has a smoothing effect. Maximum expands the selection a little, then the gaussian blur softens it.
This process (post 9) works for me when I sharpen images, but you may not like it. Sometimes no mask at all is a viable option - some folks like a grainy/noisy appearance in non-edge areas. Some people may even purposefully add noise to create that kind of look.
With USM the radius thickens the light and dark areas of edges, and amount intensifies (lightening the light, darkening the dark). Threshold is just what it's  name implies, it sets a parameter for Ps to apply the sharpening or not (but it can result in a pockmarked look in smooth areas, so I like 0)
I always look at images at actual pixels in Ps when sharpening. With sharpening there is room for opinion. Some people like thicker edges (higher radius). Some people go easy on the amount, some people are more aggressive.
It's the scaling in ID that makes things difficult, because actual pixels won't give you a true print appearance. To correct this the image either needs to be scaled to size in PS, and placed at 100% in ID - OR left as is in ID, and the resolution changed in PS (300 x % i.e. 300 x 25% = 75 PPI)
I like to start with radius 1 and 100 amount but these values are always subject to change. I wish it was an exact science but it's not. I like sharp images, but I don't want icicles in eyebrows either. I believe you said 175 LPI screening, so if it was me I would err on the sharp side...
Something else I just thought of - the print size is something to think about, too. Consider the two extremes (which may or may not apply to your project) - thumbnails, and posters. If it's a thumbnail, by the time you've got the scaling right in Ps, there really isn't much pixel information left! In his instance a lot of sharpening probably wouldn't hurt, it would create contrast (which is important in very small images). The other end of the spectrum - posters. Actual pixels may be a little misleading in this case, because when the average person looks at a poster, they don't get right up on it at a normal reading distance, they view it from several feet away. For this reason sometimes poster images are only half-resolution, to keep the file size manageable, and give a more realistic view perspective.
Hope this helps. I am not a forum expert. The others may have better advice, or may see errors in my information.