One common question I see from photographers about lens filters is whether to get a hard or soft graduated neutral density filter. Which one is better? You’ve probably seen a lot of photographers recommend soft GNDs (if you can only choose one), since they’re more versatile when photographing uneven horizons. But that doesn’t capture the whole story. Instead, the optimal GND filter depends as much on your lens – wide vs telephoto – as the scene in front of you.
Table of Contents
Sample Photos
The easiest way to demonstrate my point is to show you a few sample photos. All of these are gray field images with a two stop graduated neutral density filter. In every photo, the filter’s transition point is in the center of the image.
In terms of post-processing, I corrected vignetting and made minor adjustments to equalize brightness. Otherwise, these are straight out of camera RAW files:
Do you notice anything interesting? When you’re zoomed in to 200mm, the soft GND has only the most minimal of gradients. What was once a specialty graduated filter is now acting as a watered-down standard ND filter! It provides little benefit aside from cutting out some light, which is usually more of a problem than a feature.
Note that I took all four photos above at the same aperture, f/11. But the appearance of the gradient also changes with aperture. Take a look at the two shots below, taken with exactly the same filter (2 stop hard grad) and focal length (70mm), only changing aperture:
As you can see, the optimal graduated ND filter depends on far more than just the scene in front of you. Your focal length and aperture both have important effects as well.
Focal Length Comparison
In addition to the extremes of 14mm and 200mm, I thought it would be useful to show how soft and hard GNDs look at some of the most common focal lengths in between. This way, you have a better idea of when each graduated filter is more useful than the other. Note that all of these shots are taken at f/11:
20mm
35mm
50mm
70mm
100mm
135mm
Note, of course, that a real-world scene will show these effects far more subtly. Even at the wider focal lengths, a hard GND is unlikely to have a noticeable line in the images, except for cases where the transition is over an empty area of sky.
When Should You Use a Soft or Hard GND?
If you’re shooting with an ultra-wide lens (anything wider than 20mm full-frame equivalent), a soft graduated neutral density filter is usually the way to go. Hard GNDs just transition too suddenly at such wide focal lengths, and they only work well for very flat horizons.
From 35mm to about 50mm, hard and soft GNDs both have a place. Something between the two would really be ideal; hard grads are a bit too harsh, and soft grads don’t have a massive effect. I would tend toward a soft GND when the horizon is broken by an obvious mountain, tree, or something similar. In most other cases, a hard GND will be preferable.
With telephoto lenses beyond 70mm, I really wouldn’t use a soft grad filter in most cases. They just don’t have enough of an effect at those focal lengths; it’s like not using a filter at all. Although hard GNDs are still fairly strong in the 70mm to 100mm range, real-world scenes will be much more accommodating than this torture-test of a gray field. For example, the photo below (taken at 100mm and f/5.6) is a good example of when a soft GND will have practically no effect, but a hard GND is quite valuable in equalizing the sky and foreground:
Conclusion
Hopefully, this article gave you a better understanding of how to use filters more effectively, specifically graduated neutral density filters. The typical logic of “flat horizon = hard GND, broken horizon = soft GND” is not totally wrong, but it’s based on the assumption that you are using a wide angle lens. Once you get to about 50mm, and especially at 100mm and beyond, the equation changes. Suddenly, even for mountain landscapes and broken horizons, a hard graduated filter is the way to go.
So, if you’re tempted just to buy one type of GND and not the other, make sure you consider more than just the types of subjects you plan to shoot. Your expected focal length range is an equally important factor, if not more so.
Thanks, just the info I was looking for!
I guess it would of course also have different impact if you are working with a aps-c sensor or your phase one 4X4 cm…
What an absolutely cracking article, thank you. It takes great skill to make what could be complex into easily understandable information. I’m still finding my way around the fascinating world of photography and this kind of article is incredibly helpful.
Will now start looking at medium grad filters. Excellent work.
Thanks Heaps. The best of its kind. I never found such info anywhere else on the Web. I have SW150 and 3 stop very hard grad which is too harsh for my Tamron 15-30 G2 so will get a 3 stop soft grad to complete the collection. Appreciate your help and time on this matter. Thanks from NZ.
Do the soft VS hard affects only apply to zoom lenses when comparing focal lengths? I can imagine when you zoom out in Teléphoto we are only looking through the center part of the lens so it makes sense. But it does not seem to apply to prime lenses, does it?
It does apply to prime lenses just the same as zooms. What you’re doing with either a prime or a zoom is the same – essentially, including a larger or smaller portion of the filter in your photo. With a wide angle, you’re including a larger portion of the filter in your image, which means that the gradient appears smaller (and therefore sharper/harsher) by comparison.
Hi. Im looking to buy gnd filter. I have an older model haida pro 100 series and want to buy filter for it. I have a canon 7d with kit lens, 18-55mm and the 75-300mm will be ok with a 100X100 filter size? Or should i get the cokin p series holder, coz i plan to buy a wide angle lens later on. Thx
Very interesting, and not something I expected. I haven’t used GND filters much yet but now I have a much better idea of what I should try.
I wonder how it looks if you keep the same physical aperture at various focal lengths, rather than the f/stop aperture ratio? For example, if you compared the hard grad filter at 24mm f/2, 35/2.8, 50/4, 105/8. I’m expecting these to look very similar but maybe I’m wrong.
I bet the effect also changes with the placement of the filters, i.e. if you could position the GND closer or further from the lens (not possible with wide angles I know) or if different lens designs at the same focal length effectively change the filter position.
I’d be curious to know if shutter speed has an affect as well. Sensors do tend to bleed some light at slower speeds. May have to test it myself!
A very helpful article, Spencer, thank you very much. Would you elaborate why you’d use a physical filter rather than a software filter?
As the argument pro physical filters I see the possibility to acquire a wider dynamic range of the scenery than it would be possible without filter. An argument against physical filters is the general loss of light and thus higher ISO setting if keeping time and aperture.
But today’s cameras with their ISO invariance of about 4-5 stops allow you to expose for the highlights and then apply a digital ND filter(s) (lower sky, brighten foreground). You can later decide the exact position and the gradient and the falloff. This is a big pro for using digital filters. So what do you think?
In flash light or stage photography you sometimes can’t avoid disturbing exposure gradients. One face is bright, the next one not that much. But there is not really enough time to set up physical filters before each shot. So digital filters in post processing do a fine job. Couldn’t they generally replace the graduated ND filters?
Cheers! -jan
In regards to using grads with crop sensors. Should I just apply the crop to my lens focal length and work with the equivalent focal length, or is it more involved than that.
Great info though.
Yes, just apply your crop factor if you want to mimic the look of each photo – both to focal length and aperture. (I used 14-200mm at f/11 here, but with APS-C the equivalent is 9-130mm at f/7.1).
More important than exact values, though, is the general trend – and making sure that when you use a grad filter, its effect is neither too strong nor too weak. That’s usually easy enough to figure out in the field.
I’ve learned something new today, thank you for the lesson! :)
Glad you found it useful, Justin!
Thanks a lot, Spencer,
I’ve never seen a better article regarding this topic.
You’re being most helpful again!
You are quite welcome, Hans!