[Joseph464]

I recently purchased a graduated ND filter for sunrise and sunset photography - i.e. to hold back bright highlights. My K-r's internal highlight correction feature is fine, but has its own shortcomings. The filter in question is 0.6, or 2 stops. But here's my question: isn't that 2 stops achieved only on full frame? The K-r sensor is APS-C, so doesn't that make the filter's effect only about 1 stop?

Here is a visual comparison of FF and APS-C sensor size. If we place an imaginary horizon through the center, the upper part of the filter (where it is most dense) isn't seen by the APS-C sensor.

This tells me that if I want 2 stops with a graduated ND filter, I have to buy a 1.2, or a even better, a 1.5.

Is my assessment correct?

Attached Images

 

[amoringello]

I don't think the effect is completely linear from the center to the edge.
There is a good portion of the filter that is the full 2-stops. Then a small fall-off for some limited portion near the center.

That graduated area will take up a larger portion of the total scene to it will probably seem softer for the visible area, but I believe the 2-stop portion will still be in frame.

I guess it depends on the filter?? I've seen some with really soft graduation, some that seemed pretty quick falloff.
The ones I just bought (el-cheap-o $20 set from no-name brand) appear fairly linear almost to the edge.
In that case, yeah. You'd be losing some f-stoppage if the filter remained centered on a crop frame.

[JohnX]

You're overlooking the fact that you have to align the filter in the filter holder in order to 'set' the location of the transition zone.

You'll be able to see where the densest part of the filter is, ie. at the top of the frame, through the viewfinder. You'll align the transition with the horizon.

You might adjust it upwards in order to lessen the effect. In any event, the camera's metering will sort it out.

[Just1MoreDave]

I think you can minimize/eliminate the issue by using a lens designed for APS-C. Those lenses are designed to see and send an APS-C sized image to the sensor. Interesting question.

[acoufap]

But here's my question: isn't that 2 stops achieved only on full frame?

A ND filter reduces the light in a defined way. Every step halves the passing light. Every such step is expressed by 0.3, the whole value is incremented and expresses the value of the ND filter. 0.3 stands for one exposure value (EV): 0.3 => 1 EV, 0.6 => 2 EV, 0.9 => 3 EV, ...

Based on the designed EV system you can compute within the ISO-Time-Aperture equation. One (aperture) full stop (by example 5.6 => 8.0) takes 1 EV. So, often people say one stop. In reality the equivalent 1 EV is meant - 1 stop takes the half of light that hits the sensor.

ND filter strength has nothing to do with a camera sensor format.

[Joseph464]

Thank you Amoringello, John X and Just1MoreDave.

I use M42 full frame lenses exclusively, so my filters are all of the screw mount type. Until I win a lottery, my retirement cash flow won't allow me to invest in new lenses or other filter systems. But your comments are much appreciated.

---------- Post added 02-05-17 at 12:25 PM ----------

A ND filter reduces the light in a defined way. Every step halves the passing light. Every such step is expressed by 0.3, the whole value is incremented and expresses the value of the ND filter. 0.3 stands for one exposure value (EV): 0.3 => 1 EV, 0.6 => 2 EV, 0.9 => 3 EV, ...

Based on the designed EV system you can compute within the ISO-Time-Aperture equation. One (aperture) full stop (by example 5.6 => 8.0) takes 1 EV. So, often people say one stop. In reality the equivalent 1 EV is meant - 1 stop takes the half of light that hits the sensor.

ND filter strength has nothing to do with a camera sensor format.

Hello acoufap.

You missed the word "graduated." Since there is a change in density from the outside of the filter towards the centre, sensor size does matter.

[AstroDave]

From your description, I presume you are using a radial density variation filter, rather than the more common half-and-half ND filter (density gradient pretty much a line across the middle of the filter).

Then, it depends on how the variation depends on radius. If it is relatively uniform, then with an APS-C sensor, and a full-frame lens, the density range across the sensor will be less than advertised, as depicted in your cartoon.

If you want the density (more filtration) to be higher in the middle (what I guess you want, to cut down on the brightness of the sun in the ~middle of your images) and you can not use an APS-C lens that matches your filter size, you will need filters with either larger central density or a more concentrated effect.

If you are, in fact, using the more conventional ND filter (the gradient across the middle), it shouldn't matter what sensor size you are using. The gradient still goes across the middle of the sensor at the horizon. The top half of your image will get the full effect of the filter (2 stops, say) while the bottom of the image is normally exposed.

[Just1MoreDave]

The lens design has to matter also. For example, the Takumar 50mm f1.4 takes a 49mm filter and the front element is something like 40-41mm. The recent Sigma or Samyang 50mm f1.4 lenses are also intended for FF and use a 77mm filter. I can't tell the exact size of the front elements of those lenses but they are much bigger than ~41mm. The Tamron Adaptall-2 28mm f2.5 has a very small front element, much smaller than any of the Pentax or Takumar 28mm f2.8 or f3.5 lenses. If it was just about an APSC-sized rectangle on the lens element, the Tamron would see much less of a 49mm filter than a Pentax would. I don't know if the filter designers consider any of that.

Rather than get a degree in optics to figure it out on paper, I'd get a couple of L-shaped pieces of cardboard and some tape to make a variable-sized rectangular mask for the lens I wanted to use. Use Live View and close up the Ls until they start blocking the frame. That's the part of the filter the camera sensor will see. Then see if the filter maker will tell you anything about the density for that location.

[Joseph464]

Thanks guys - this a fascinating discussion.

AstroDave:
I may have led you astray. My 49mm screw-on filter is the half-and-half variety. I should have said the change in density is from the outside starting at the top towards the centre. The level of density is the same left to right. The filter in question is a Tiffen which the manufacturer says has a "smoothly graded transition"...in other words meaning continuously decreasing from top edge to the centre. If you look at the illustration I first posted, you can see the portion of the filter's coverage area which is outside the APS-C sensor rectangle.

Just1More Dave:
Your point that the size (diameter) of the filter matters is right on! It's another variable for sure. Although the 49mm filter fits two of my SMC Takumars (28mm and 55mm) ... it's obvious just by looking that the 28mm has a smaller front element. If I'm only getting about 1 stop with the 55mm, maybe I'm getting even less with the 28mm. Over the last hour I've been searching the internet for a 49mm threaded graduated ND that's stronger than 0.6, and haven't found one, so maybe I'm plumb out of luck.

[acoufap]

]
Hello acoufap.

You missed the word "graduated." Since there is a change in density from the outside of the filter towards the centre, sensor size does matter.

Ok - I do understand what you mean.

I’d say there are different aspects we have to consider especially when using a graduated screw-in ND-filter.

• different vendors deliver different gradient „flows“.
• there may be hard and soft grad variants. These variants have different effect in strength when using with different sensor formats.
• if you use StepUp rings to support a bigger ND filter diameter on a lens with smaller filter thread this also results in a different effect.
• it’s even a difference if you use the camera/lens with filter in landscape or portrait mode.

With a not-screw-in graduated ND filter system we are much more flexible. I’d always prefer this type.

[Joseph464]

Thank you, acoufap.
You've provided a great observation: "it’s even a difference if you use the camera/lens with filter in landscape or portrait mode."
I've rotated the illustration so we can see difference that orientation makes.
The area covered by the filter at the top of the sensor has more than doubled!