So I had in mind when photographing the Tetons and the Snake River at sunrise that a split grad ND filter might be handy. I brought my Cokin set, which I had purchased many years ago and used before I became a professional photographer. However, I found I was able to do more at higher quality with digital ND filters
In my early days I probably didn't realize that the ND filters reduced the sharpness of the image because I never liked my images enough to call them professional, and my largest prints were 8x12's from film. With film, I didn't have the option to zoom in or out of the negative to see its maximum sharpness.
But since then my eye has become more discerning and my skills 100 times sharper, and file quality much better. After this trip to the Tetons in July, after getting perfect captures and zooming into them in Lightroom, I was able to see that adding a Cokin split ND grad actually significantly reduces sharpness. Even though the filter covers the whole lens, the part covered by the dark area is less sharp than the clear part.
I've compared several images of course and the ones without the ND are much sharper than the ones with. There's just no doubt, the ND reduces image quality. This shouldn't be surprising, since Ansel Adams also noticed that filters reduced optical quality (see "The Negative" -- filters).
So, lesson learned. If you want the best quality, leave ND grad filters off the camera. There's not much a physical split grad ND can do that a software one in Lightroom or Nik cannot. The only use I could see is for motion blur, but if you wanted quality you could also just combine two exposures. Software and HDR has eliminated need for physical ND grad filters in most circumstances.
Also, in my experience, not as many scenes fall outside the dynamic range of a DSLR as you might think. It does happen, but usually in more extreme situations. Most of the time a scene falls within 6 stops, even in bright daylight. With any DSLR, you're safe with 5.5, maybe 6 stops at least, and up to 7 or 8 on the high-end cameras.
If your goal is to reduce total light in the exposure and you must use an ND, I would recommend a screw-on glass filter or polarizer. I have not seen image degradation with polarizers. I have not tested screw-on NDs for sharpness, so I can speak only for the Cokin slide filters, but I would guess that the screw-on glass filters are a step up in quality and are less likely to reduce sharpness.
If you decide to use an ND -- if for some reason you prefer a degraded image -- there's a technique for exposure you should know. The purpose of an ND is to reduce light on the exposure. Therefore, when metering manually, you do not open up exposure for an ND like you would a polarizer. My recommendation is to use an incident meter to measure the light in the foreground and set the exposure to that. Adding the split ND will automatically darken the sky relative to the ambient. If you spot meter say for a highlight and open up for the ND, you defeat the purpose of using it.
If you're using matrix metering, adding the ND will influence the total light the camera is seeing. My recommendation there is to take the reading the camera gave you before adding the ND and use that. But, by using software NDs, you avoid this hassle and gain exponential amounts of control over the finished image in ways a physical ND simply can't provide.
Comments