Friday, July 17, 2020

"Capture Sharpen" - AA filter and optical comparison

After having experienced "Capture Sharpen" in RawTherapee and seeing how it can clean up an image, I wanted to take a look at how it works on images that come from a system that have a strong anti-aliasing (AA) filter (for which "Capture Sharpen" was made for) and a system with a weak AA filter (for which "Capture Sharpen" might not have been designed for, but might have an influence on).

Setup -
  • One image from
    • Canon 5D MkII
      • strong AA filter
      • 24-105mm f/4 L IS
      • shot wide open
    • Sony A6000
      • weak AA filter
      • Sigma 60mm f/2.8 Art DN
      • shot wide open
  • Process in RawTherapee
    • Lens Corrections
      • chromatic aberrations
      • field distortions
    • "Auto Levels"
    • Set "Curves" black to the bottom end of the image's histogram
    • "Capture Sharpen"
  • Pull 100percent resolution 500x500pixel sections from the image and display them along with a down-rez'd copy of the original image
Comparison -

[If you click on the following images and then select full-resolution versions of these images you will be able to see differences between the photos]

Starting with a Canon 5D image...

RawTherapee Capture Sharpen ~Canon 5D 24 to 105mm f/4 L at 50mm


Following up with a Sony A6000 image...


RawTherapee Capture Sharpen ~ Sigma 60mm Art Example


Comments -

Starting with the Canon image we see that the photo looks nice and seems to have fairly decent resolution.  Without the "Capture Sharpen" processing step, however, the image is rather soft. Knowing the Canon system as I do I can say that there are two things to take into account.

First, the AA filter Canon uses is strong.  It deliberately softens an image so as to get around any moire effects that might crop up.  Looking at a Canon image all by itself I doubt people would be displeased with their results, but...

Second, the 24-105mm f/4 L IS (version 1) was highly touted at the time I purchased it.  I thought nothing of its performance and assumed it was as good as I could get, regardless of price (at the time the 5D MkII was introduced a kit was rather expensive at 3500USD).  Over time and with many many photoshoots under my belt I started to uncover subtle performance degradations. 

When I shot it side by side with a Sony mirrorless system I was instantly struck by how soft images were with my Canon gear.  The following is a hint of how I came to the decision to sell my Canon equipment. 

Take a look at the second image and the differences between the two systems (Canon and Sony) should be readily apparent.  The Sony A6000 + Sigma 60mm f/2.8 Art DN is fabulously sharp.  And under the treatment of "Capture Sharpen" the overall image "sharpness" is over the top incredible.

I bought a Sony A6000 body new at the local Salon de la Photo some years ago and paid 450Euro (less than 500USD at the time).  Shortly after I bought the body I picked up a new Sigma 60mm f/2.8 Art Dn for less than 200Euro (around 220USD).

You can see where I'm going with this, right?

Before anyone gets upset about me comparing a Canon zoom against an aftermarket prime, my overall experience of using the Canon DSLR was very consistent.  It didn't seem to matter if I used a prime lens or a zoom on any Canon I ever owned (and I owned a lot of them).  The strength of the Canon AA filters over-ride pure optical resolution.  The 24-105L proved to be less than satisfying in addition to the AA effects.

What I see here is that taking a camera with a strong AA filter and passing it through "Capture Sharpen" can bring an image the appearance of "sharpness."  It is by comparison with other systems that one might question the use of strong AA filters in general photography.

What I see here in taking a camera with a weak or non-existent AA filter and mating it with a sharp lens (remembering, of course, that it is difficult to find a lens that is _not_ sharp) and passing it through the "Capture Sharpen" process can yield images that are sharper than I could ever have imagined.


No comments: