Why is it best to sharpen at 100%?

Everything I read suggests that I view an image somewhere between 50-100% to determine the amount of sharpening needed. I've never understood this because when I shrink the image back down to size, I can't see any sharpening. Why wouldn't it be best to sharpen an image at the final output size, whether print or screen?

I've often found that in Photoshop, if I view at 25%, 50%, 75% or 100%, the image quality is better than if I zoom in at 33% or 66%. It's at these "sharper" zoom levels that I judge and adjust sharpness.

Pete wrote in post #8479523
I've often found that in Photoshop, if I view at 25%, 50%, 75% or 100%, the image quality is better than if I zoom in at 33% or 66%. It's at these "sharper" zoom levels that I judge and adjust sharpness.

That's no longer so for PSCS4 (with Open GL). You can use any zoom level there.

IMO, capture sharpening should be judged at 100% ((or greater) In fact, that's all that LR or ACR will allow)

Output sharpening should be judged at / closer to output size. So 25% or 50% in PSCS3 and older, "Print sizew" in PSCS4.

Yeah. I'm using CS3. I'm so behind the times...

Fifty percent maybe. One hundred percent, never--at least not with a big image file. With 20 megapixel images, you're not seeing enough of the image to make a pictorial judgment.

Got you beat, Pete. CS2 here.

Sharpen at 50%, then check the image at 100 by scrolling around. Reason being that haloing can pop up on some items in the pic, while not on others and 100% reveals haloing better.

CannedHeat wrote in post #8485701
Got you beat, Pete. CS2 here.

Sharpen at 50%, then check the image at 100 by scrolling around. Reason being that haloing can pop up on some items in the pic, while not on others and 100% reveals haloing better.

If you start with a 21 megapixel image, enlarge it to a final size of 30x40, then sharpen it, it's going to take a long doggone time to cruise the image at 100% to find haloing that isn't even going to be visible at printed size.

RDKirk wrote in post #8485838
If you start with a 21 megapixel image, enlarge it to a final size of 30x40, then sharpen it, it's going to take a long doggone time to cruise the image at 100% to find haloing that isn't even going to be visible at printed size.

You don't need to cruise the entire image. Just sections of similar subject. In other words, if it's a landscape with bare tree limbs against the sky (high contrast), if one part is haloing, another will be also. You don't need to examine every branch.

Also, I guess it's monitor dependant. I run a 24" at 1900 x 1200. Even at 100%, the 5d MKII images only take a few seconds to examine. I grab the red frame on the navigator tool and easily move around various sections.

I would rather take the time to examine an image than to output one of those obviously over-sharpened images you see now and then.

RDKirk wrote in post #8485838
If you start with a 21 megapixel image, enlarge it to a final size of 30x40, then sharpen it, it's going to take a long doggone time to cruise the image at 100% to find haloing that isn't even going to be visible at printed size.

You are talking about output sharpening, It's input sharpening that should be viewed at 1:1.

René Damkot wrote in post #8492473
You are talking about output sharpening, It's input sharpening that should be viewed at 1:1.

What's, er, the difference?

Capture sharpening is used to neutralize the blurring caused by the Anti Aliasing filter in your camera.
Output sharpening is dependent on output (print or screen, what size) and meant to overcome the softening caused by resizing or happening when printing.

Lightroom for instance offers both. Just posted about LR sharpening on my blog.

Oh sweet! I'm going to have to read through the entire blog now. I've always been confused about sharpening (and LR in general haha).

"Input sharpening", or "capture sharpening" is a term that is used to refer to the fact that digital cameras have an anti-aliasing filter that removes certain artifacts but somewhat softens an image (don't ask me for details), and so digital images can benefit from a certain amount of low-level sharpening, no matter what the intended output. Different cameras have filters of different "strength" and so some put out images that are "softer" than others, but all of our Canon cameras have this filter put in.

This input or capture sharpening is either put in by the camera if you shoot in jpeg or is often put in by you in a Raw converter when you shoot Raw. Some shooters, though, don't apply sharpening at the Raw level if they intend to sharpen in, say, Photoshop -- they prefer sharpening as part of the fuller process. A lot of us, though, use a Raw processor such as Lightroom or DPP as our main processor, with Raw sharpening as our basic sharpening method until a special need occurs for printing or another required output.

This is where "output sharpening" comes in -- a specific demand for an image can call for special processing, and sharpness is a big consideration. If, say, you want to print an image at 20" x 30", all the detail of that image will be enlarged, and you will likely want to ensure that sharpness of that image is optimized for a large print. If, though, you prepare an image for displaying on the Web, you may want another level of sharpening. Also, different types of images can benefit from different approaches to sharpening that can be considered "output sharpening". Portraits take a special approach to fine tune them, for example. A lot of images benefit from sharpening the central subject while not sharpening the background. These are tasks that may not be done well with a Raw processor, and certainly are not considered if shooting jpeg.

Whether you think and act with different concepts of input processing and output processing, or even throw in intermediate processing is a matter of personal style, but can make a difference to how you approach things.

tonylong wrote in post #8501151
"Input sharpening", or "capture sharpening" is a term that is used to refer to the fact that digital cameras have an anti-aliasing filter that removes certain artifacts but somewhat softens an image (don't ask me for details)

A camera without AA filter would produce very sharp files, with a ton of moiré.
Some nice examples Here. (Scroll down a bit. Whole article is a fun read)

René Damkot wrote in post #8501288
A camera without AA filter would produce very sharp files, with a ton of moiré.
Some nice examples Here. (Scroll down a bit. Whole article is a fun read)

Heh! That's what I was referring to when I said "don't ask for details"! Try explaining this stuff on a public forum! I can't even understand half of it myself! And, let's see, let's talk about Bayer interpolation -- shall we have some fun!?