How come Gradient ND filters actually work?

Infinite_Day · Mar 5, 2012

Uh, you're bringing focus into question when it really has no effect on the entire principle. You're not focusing on the filter. All it's doing is a graduated attenuation of the light over the front plane of the lens. I understand why it's interesting to look into this but you're really, really overthinking it without trying to be objective. You're sticking with your base assumptions without actually taking into consideration what people (mostly Helen) are trying to explain to you.

Helen B · Mar 5, 2012

Exactly. The focus and whether or not the lens is stopped down do not affect the fundamentals of the case.

I'll assume that you do indeed understand the concept and properties of the entrance pupil. Here are two snaps that illustrate the experiment I wanted you to do. The lens is a 24 mm f/3.5, wide open, with a filter fitted to the front. The first picture is taken on axis and the second off axis. Hopefully you can see that the entrance pupil (and hence the zone through which the light rays that form the image pass) is in different locations at the plane of the filter. If you do understand what the entrance pupil is, you should now understand that the light from an object point that will subsequently form the image of that point does not pass through the whole of the filter, only a part of it (because the filter and the entrance pupil are not in the same plane).

hartz · Mar 5, 2012

Helen B said:
Exactly. The focus and whether or not the lens is stopped down do not affect the fundamentals of the case.

I'll assume that you do indeed understand the concept and properties of the entrance pupil. Here are two snaps that illustrate the experiment I wanted you to do. The lens is a 24 mm f/3.5, wide open, with a filter fitted to the front. The first picture is taken on axis and the second off axis. Hopefully you can see that the entrance pupil (and hence the zone through which the light rays that form the image pass) is in different locations at the plane of the filter. If you do understand what the entrance pupil is, you should now understand that the light from an object point that will subsequently form the image of that point does not pass through the whole of the filter, only a part of it (because the filter and the entrance pupil are not in the same plane).

I can see the bright opening do not appear to align with center of the front lens element in the second picture. Interesting, but is it relevant? Regardless the reflections of the lights behind me in the room looked pretty.

When stopping the lens down, the entire image becomes dimmer, not only the outer edge. Stopping the lens down is an artificial way of eliminating part of the lens surface from the equation, thereby reducing it's light gathering potential.

I have just done a different experiment and not gotten the result I wanted/expected. I am researching the topic and will find a scientific answer yet.

hartz · Mar 5, 2012

Converging Lenses - Ray Diagrams

hartz · Mar 5, 2012

Helen B · Mar 5, 2012

hartz said:
I can see the bright opening do not appear to align with center of the front lens element in the second picture. Interesting, but is it relevant?

Of course it is entirely relevant. You don't understand the concept of the entrance pupil, do you?

I am researching the topic and will find a scientific answer yet.

You have been given a scientific answer already, but you aren't getting it because you don't understand the whole entrance pupil issue yet.

hartz said:
Converging Lenses - Ray Diagrams

That page tells you nothing you need to know about this - it is totally misleading. I've already explained why (see post #19). This is getting very frustrating.

The filter is not in the plane of the iris (which appears as the entrance pupil). The entrance pupil of a compound (ie multi-element) lens is not usually at the front element as it is with a simple single-element thin lens. Forget the thin lens diagrams and look at the ray diagram of the compound lens I posted earlier.

Helen B · Mar 5, 2012

Here's the amended ray diagram. Sorry I haven't got a scanner handy - this is a quick picture with a P&S.

The entrance pupil is the image of the iris (wide open here) as seen from the front of the lens. In my pictures of the actual lens it is the bright white dot. Only rays that are heading towards the entrance pupil get through the lens to form the image. If a ray is not headed towards the entrance pupil, it will not reach the image. Therefore this diagram shows that not all the rays from an object point that reach the filter will then reach the image.

Helen B · Mar 6, 2012

gerardo2068 said:
I don't think light get mix up at any point if it got mix up at any point before it hit the sensor there would be weird photo. How would the light un-mix after being mix up?

Even if it meet at one small point to be inverted by the lens I don't thing it actually mix. By mix I mean every single ray of light mix to one single ray point.

The light doesn't get mixed up to a single ray (of course) but in a properly designed lens all the light that forms the image is 'mixed up' evenly at the iris. This is a necessary condition for the iris action to affect the brightness of all parts of the image evenly. (The effect may not be perfectly even when the iris is wide open or close to wide open.) By 'mixed up' I mean that the light from every object point that is imaged is spread evenly across the whole of the iris. Therefore an ND grad at that location, or close to it, would have no graduated effect, exactly as hertz points out (it's very clear that this is the case that hertz is locked into thinking about, even though hertz might not yet realise it). For a simple, thin lens the aperture stop (iris, diaphragm) can be very close to the location of a filter.

hartz · Mar 6, 2012

Helen B said:
Here's the amended ray diagram. Sorry I haven't got a scanner handy - this is a quick picture with a P&S.

The entrance pupil is the image of the iris (wide open here) as seen from the front of the lens. In my pictures of the actual lens it is the bright white dot. Only rays that are heading towards the entrance pupil get through the lens to form the image. If a ray is not headed towards the entrance pupil, it will not reach the image. Therefore this diagram shows that not all the rays from an object point that reach the filter will then reach the image.

Just so you know: I do believe you (about the fact that only those rays that are heading towards the pupil will make it all the way through the lens).

I am very surprised at that though. I don't WANT to believe it. The scientist in me does not want to accept things at face value, even when they appear to be true based on experience, I still want to know : "But WHY".

Helen B · Mar 6, 2012

hartz said:
I am very surprised at that though. I don't WANT to believe it. The scientist in me does not want to accept things at face value, even when they appear to be true based on experience, I still want to know : "But WHY".

Because if they aren't heading towards the entrance pupil (the image of the aperture stop as seen from the front of the lens) before they enter the lens they will not make it through the aperture stop itself. The WHY is very simple once you understand what the entrance pupil is, which you don't appear to have done yet.

Please think about what it means for the entrance pupil to be the image of the aperture stop. A ray headed towards the edge of the entrance pupil will be bent once it enters the lens, so that it then grazes the edge of the aperture stop. We see the ray that comes the other way - a ray coming from the edge of the aperture stop comes out of the lens towards us along the same path, then emerges from the lens towards our eye. It appears to come from the image of the edge of the aperture stop, ie the edge of the entrance pupil.

hartz · Mar 6, 2012

OK, I get it now.

hartz · Mar 6, 2012

P.S. Thank you.

hartz · Apr 3, 2012

This Telephoto zoom lenses shows that light from a point to one point, eg above the ND grad horizon, do hit and pass through the top and bottom halfs of the lens (and filter), and do get focused back to a single point.

Helen B · Apr 3, 2012

hartz said:
This Telephoto zoom lenses shows that light from a point to one point, eg above the ND grad horizon, do hit and pass through the top and bottom halfs of the lens (and filter), and do get focused back to a single point.

No it doesn't.

First, it is a highly simplified representation of the action of a zoom lens, not a proper ray diagram; second, it isn't anywhere near correct - in particular it doesn't show the iris, which is a huge reason not to trust it as an accurate depiction of the rays that make up the image; and third, even with its inaccuracies, it does not show that the rays that form the image of a point would all pass equally through a grad - in fact if you believed this diagram a grad would work, but upside-down. Any one of those reasons is enough to discount it as proof of anything that is relevant to this discussion.

As a general principle you should never use a diagram that is intended for one purpose for any other purpose, unless you understand the simplifications and assumptions used to make the diagram and also know that they do not influence the other purpose you wish to use it for.