Welcome to the third part of my review of the Nikon AF-S DX 35/1.8G. The first part was about sharpness, the second part about bokeh, and today we will look into chromatic aberrations, an optical phenomenon often referred to under the abbreviation CA.
I really don’t want to explain over and again what has been explained so often. For more detailed explanations of chromatic aberrations see articles by Paul van Walree, Ken Rockwell or Cliff Smith, to only name a few.
Fact is, they come in four flavors, the combinations of the pairs primary/secondary, and longitudinal/lateral. Primary CA (red/blue fringes) is corrected in modern lenses, all that remains is secondary CA (green/magenta fringes). Longitudinal means that different wavelengths focus on different planes. That cannot be corrected in software. Lateral can, because with lateral chromatic aberrations all wavelengths focus on the same plane, but they produce differently sized images. The software trick is, to scale the red, green and blue channel until the images have the exact same sizes. That’s what you do in software when you correct secondary lateral CA.
Some new Nikon DSLRs automatically correct CA in-camera, at least for JPEGs. Nikon introduced the feature with the D3/D300, and as far as I know, all new cameras since, the D700, D90 and D3X have it as well. Correction for CA in Nikon cameras works extremely well and whenever I have cared to look closely at a JPEG, it was flawless. Whenever I have tried to do better in Adobe Camera RAW, I have failed. Up to now, that is.
And here comes a surprising point: Nikon does not base CA removal on anything the camera is told by the lens, not even for their own lenses. CA removal is totally based on image analysis. How do I know? Because I caught them failing miserably.
Obviously this lens suffers badly from secondary lateral CA. That’s certainly a problem, but none that I’d call too severe. If you’re shooting JPEG only, you could get unlucky and trigger a case where the built-in correction fails (and that is extremely rare) or you may shoot with a camera that has no built-in correction at all. If you shoot RAW, you have to correct it, but you can.
The fact that a single camera, my Nikon D300, with a single lens, my 35/1.8, corrects for CA at one time and not at another, is proof for me that this correction is based upon image analysis only. Most of the time it works flawless, sometimes it fails, and this may be due to certain patterns in the image.
The other interesting thing is, that lateral CA is not independent of focusing distance. The image of the bark was focused very near, and I had to use other correction values than for images focused to infinity. When you remember the illustrations in Paul van Walree‘s article, this really makes sense, but for some reason I had not expected it.
I had frequently used “Previous Conversion” settings in Adobe Camera RAW when converting more than one image. This is wrong. Don’t do it, or at least only do it, when you convert images that were focused to approximately the same distance.
Now, couldn’t I be wrong about Nikon’s correction method? Could the correction generally fail at close focus?
Not at all. The Image of the Day was shot from about the same distance, and the JPEG from the camera is perfectly clean of CA, just as the conversion from RAW. QED.
Well, here we are with some lessons learned: This lens has substantial CA, but that’s something you can fix in post-processing. Some cameras do it automatically for JPEG shooters, but they don’t do it all of the time. Lateral CA varies with focusing distance, thus you better do it manually and don’t rely on remembered settings.