Re: sensors, megapixels and such part 3

Started 1 month ago | Questions thread
Mark Scott Abeln
Mark Scott Abeln Forum Pro • Posts: 15,307
General guidelines

Macro guy wrote:

Is there a formula or a general guideline for how much sharpening one has to do and what type of sharpening to use i.e. high pass sharpen, unsharp mask, etc.?

A general guideline is to make images pleasantly crisp without visible artifacts. So you can—and ought to—evaluate sharpness by eye.

But this evaluation depends on viewing distance and viewing conditions. For large scale detail that takes up a significant portion of your view, very slight sharpening will have a strong effect, but very fine detail at the limit of your perception will need a large amount of sharpening to get a good visual effect. The trouble is, if you zoom way into an image, otherwise good sharpening will show lots of artifacts, and this artifacting is clearly seen in some out-of-camera JPEGs (particularly with older cameras). If you have a low-resolution monitor, you may find that doing sharpening when zoomed way into an image may lead to disappointingly soft images, when they are viewed from a greater distance.

Many sharpening algorithms do a bad job on images with a gamma applied, such as JPEGs, or any image in an standard color space in 8 or 16 bit mode Photoshop; what happens is that the bright haloes generated in sharpening are more intense than the dark haloes, and so it is difficult to get enough crispness without showing artifacts. Back in the old days, I had to do two sharpening steps, one which would only lighten and another which would only darken and then adjust the opacity accordingly. The solution is to sharpen in a linear color space: Photoshop's 32 bit mode is good for this, and the dark and light haloes are balanced; this allows for much stronger sharpening without visual artifacts. 32 bit mode is also good for downsampling, as this process will otherwise change the tonality of an image with gamma (and this is most noticeable with dark images), and in my typically workflow, I'll downsample and sharpen as my last two steps.

Having a 5K monitor helps doing sharpening considerably, as I can view my entire image, or at least most of the image, when viewing at 100%.

Prints need more sharpening than images displayed on a typical monitor. Usually, I'll increase the sharpening amount by 50%, but this is purely empirical, and it helps to review the appearance in prints.

The sharpening radius used depends on the viewing conditions. A large print may use a larger radius than an image on a monitor.  I've had to calculate the radius according to my print size, where it is proportionally larger for bigger prints.

There is an advantage to upsampling an image before printing, in order to match the native resolution of the printer (pixels per inch and *not* dots per inch), and in this circumstance, it may be advantageous to not do capture or final sharpening before upsampling, but rather do final sharpening afterwards. Aggressive in-camera JPEG sharpening artifacts can significantly harm fine details in such a large print.

 Mark Scott Abeln's gear list:Mark Scott Abeln's gear list
Nikon D200 Nikon D7000 Nikon D750 Nikon AF-S DX Nikkor 35mm F1.8G Nikon AF Nikkor 50mm f/1.8D +2 more
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow