27

I am attempting to work out, conceptually, how the use of a neutral density filter effects the resulting captured image.

Do ND filters allow a camera to capture higher contrast scenes (without clipping the highlights) than would otherwise be possible? Or do they just decrease the amount of light uniformly so that highlight detail is gained at the expense of shadow detail being lost?

Put more technically, is the effect of a ND filter more like a linear mapping or a clipping of the low light levels? See the figure below for examples of each. Graphs of Linear Mapping and Clipping Darks

Sean
  • 3,837
  • 2
  • 30
  • 48

6 Answers6

17

It's just dark glass (or resin, of course); it reduces the amount of light entering the lens, and that's all. There may be a slight amount of nonlinearity in the colour response (causing a slight colour shift) but that is merely due to the absence of real-world, uniformly attenuative substances.

A graduated (or split) neutral density filter can be used to control contrast by selectively (by position) controlling which parts of the image are viewed through the filter, but a solid ND is just an attenuator.

  • 1
    Yes, but the part that I am not clear on is whether or not bright parts of an image are effected in the same way as shadows. In other words, do shadows get attenuated by the same amount (say, in percentage of lux allowed through) as bright areas, or do they get proportionally less dark? – Sean Sep 01 '11 at 22:00
  • 8
    The attentuation is uniform -- the entire curve drops, but the slope isn't changed. –  Sep 01 '11 at 22:03
14

Simple answer: no, ND filters don't increase dynamic range.

In zone system, an ND filter just moves exposure of scene elements n stops lower. Everything that was in the lowest n zones captured without the filter, gets clipped off as black.

I can think of two scenarios where an ND filter might increase dynamic range, but I wouldn't use it for that:

  • low quality ND filters will reduce contrast, so dynamic range increases a little (on expense of image quality)
  • film is somewhat forgiving of underexposure, so you might gain more highlights and still preserve some shadow detail by underexposing; often you could achieve the same by simply adjusting exposure parameters
Community
  • 1
Imre
  • 31,966
  • 11
  • 107
  • 177
  • Thanks for providing cases where dynamic range might be effected. The reduction of contrast is exactly what I was wondering about in the context of "linear mapping." It is interesting to know that this is an undesirable effect only found in low quality filters. – Sean Sep 11 '11 at 01:52
6

Your question references ND filters which I think most would consider to be a solid ND filter. Those are mostly used to decrease your shutter speed for long exposures or by allowing you to open up your aperture for a more shallow depth of field in bright light.

A graduated ND filter is typically used to balance the exposure. With a traditional ND filter the top part of the filter is darker (by the rated stops for the filter) and it gradually moves to transparent. You place the darker part over the brightest parts of the scene (a bright sky) in order to properly expose the foreground.

In the foreground you do not lose shadow detail because it's viewed through the transparent part of the filter. If anything you gain shadow detail as you're able to properly expose it without the bright sky causing the foreground to underexpose.

However, if you have shadow detail you need preserved in the bright area...those would most likely be lost anyway due to the bright source and just about your only choice would be to use an HDR technique. Some of the highlights would be lost in the bright area but I think if you're trying to balance the overall exposure this is a point of choice - what do you want to show in the image? Before filters or HDR you always had to choose - expose for foreground or expose for the sky/background.

There are also reverse grad ND filters which help with sunsets/sunrises - when the brightest part of the sky is along the horizon line. You are able to preserve highlight and shadow detail in the sky above the sun due to how the graduation is applied to this filter. It's darkest at the center, bottom is transparent...but from the middle dark area it gradually lightens toward the top.

Jarrod268
  • 141
  • 1
  • 4
5

Both your graphs are correct.

If you represent the intensity levels in a linear scale, then the left graph (“Linear mapping”) is the appropriate one. If you represent intensity levels in a log scale (i.e. in number of stops, or EVs/LVs...), then the right graph is appropriate.

BTW, the filter does not clip darks. The “ND filter” curve on the right graph can be extended to the left, to negative intensity levels (negative EV values are OK).

Edgar Bonet
  • 3,435
  • 20
  • 17
1

You use the ND to get to your ideal settings.

If you digital sensor has max dynamic at 800 ISO but this would overexpose the scene you use an ND to compensate.

So in this case you could say you gain dynamic using an ND opposite to reducing it in camera.

Phill
  • 11
  • 1
0

You just need to look for the equation, I found it in Wikipedia:

I / Io = 10-d (where I is the intensity after the filter, and Io is the incident intensity)

If we simplify, it's I = A * Io (A being a decimal number, ex: 0.1), meaning that an ND filter works dividing, not subtracting. The first graph is correct and yes, it theoretically increases dynamic range.

But one last thing, let's suppose that, in an scenario, what we are seeing as black it's 1000 W/m2 and what we see as brightest is 1001 W/m2. As you can see the increase of dynamic range would be negligible. Pure black doesn't exist in our everyday world so be careful, I didn't find what is the average intensity of light in our everyday dark and bright scenarios but you can try to look for it.

Edit: I guess the answers mentioning the log scales are right, and what I did is a semi-explanation about why we need log scales.

Glorfindel
  • 1,128
  • 1
  • 11
  • 21
Joaquim Ramis Martí
  • 1