Back in early August, I was lamenting over some “crunchy” looking photographs and what I thought was the resolution to my problems. Now I am pretty sure I was not quite right, but nevertheless close since in fact I was very much over-sharpening my JPEG capture images. So here is how I came to find out out what the real problem was.
While comparing some prints recently with some other photographers, I noted that some of the prints were looking a lot sharper in detail for one photographer than I recalled seeing in the past. Especially when I had some images that were a bit mushy and my sharpening process was not doing the trick. We then proceeded to get into a long discussion about the merits of sharpening with a high-pass filter versus using the more traditional unsharp mask to sharpen (the latter my defacto image sharpening method). So while subsequently investigating the high-pass filter, found out that this is highly recommended for out-put sharpening. Neat, something to experiment with.
But that came with a note that using the unsharp mask was like using a dull edge knife to cut steak. hmmmmm. So I decided to look for recommendations for image capture sharpening to compensate for the slight image degradation by the aliasing filer in front of the camera’s sensor. Like I said, until now, my defacto for many, many years was the unsharp mask as a layer to provide the first sharpening action. So while reading all of this stuff, there was this other note, that for JPEG capture, not only do you lose a lot of image information as compared to raw, JPEG also does an image sharpening process.
What? Had I just overlooked this aspect of JPEG for this many years?? I suspect so, as I now find other references to the fact that shooting in JPEG for image capture will also provide sharper images that already compensates for the aliasing filer. In other words, for a JPEG image, I do not need to start my image processing with adding a layer to sharpen the image and in fact that process will start me down the road to over-sharpen the image towards crunchiness.
So I have gone back to inspect a bunch of recent JPEG capture images to evaluate with and without the initial capture sharpening. It appears that I have been doing myself a little injustice with adding that initial unsharp mask layer for JPEG images. The good news is that by eliminating that duplicate background layer for the unsharp mask will make my images files a lot smaller and image processing a bit faster. Nice!
So a reminder: don’t capture sharpen those JPEG images. For raw capture image, that’s going to be another discussion as I have learned some things here as well.
Note: the image above is from my project Middle Ground and is a JPEG (capture) image that I just processed without resorting to an initial capture sharpening process of an unsharp mask layer. I think it looks pretty good ;- )