Advice on graduated ND filters for landscape photos

[Alis]

Hi Everyone,

I need some advice on buying GND filters. I have never used any GND filters before. To get ready for some landscape photography I have been reading about them and as I understand, the reason my skys are always washed out is that I am not using these filters. I need to get some advice from those who use it regularly as to what brand and what different strengths (stops) I need to have to cover most situations, without spending a fortune. The only filter I currently have is a Hoya CL polarizer filter.

[Colin Southern]

Here begins a journey ...

GNDs come in 2 flavours ... ones that screw in and ones that you can put in a mount and slide up and down. Screw-in ones are a total waste of money because you can't move where the graduation start, and would be limited to referencing everything to the middle of the filter which is seldom where we want it to be.

For sliding filters you'll first need a filter holder and lens adaptor ring (to account for various lens diameters) - there are two common brands out there the first is Lee and the second is Cokin. Lee are better quality, Cokin are cheaper.

Sliding filters come in two common sizes; 84mm x 120mm and 100mm x 150mm (post card size) - my advice is to stick with the bigger ones right from the start as they give you a lot more room for error, especially when hand-holding - plus, once you go past a certain lens size they're mandatory anyway.

I use singh-ray GND filters exclusively, from www.singh-ray.com, but they're not cheap - there are other manufacturers out there, but quality varies a lot - in general, "you get what you pay for".

In terms of what to start with, I'd suggest getting 3 ... all 3-Stop.

1. A 3-Stop Hard-Edge GND

2. A 3-Stop Hard-Edge REVERSE GND

3. A 3-Stop soft edge GND.

If you can spring for a Lee foundation kit then so much the better (can give you links to it on the B&H website if you need).

Hope this is a good starting point for you

[Roxy]

Glad to see this thread; I am of a similar mind, with little experience. What do you think about HDR activity by comparison?
HDR seems to me to require skilful and time-consuming post processing, but if we already have the programme (in my case Elements 7), would avoid the expense of a filter set?

[Colin Southern]

It's difficult to explain, but in practice an HDR image and an image done with GND filters are different things. The GND approach is more closely aligned with the "Take two exposures and blend the two via a mask" technique.

[crisscross]

Do you think it is still necessary to have physical ND filters, as opposed to imposing the effect afterwards using a gradient (or gradients) with relevent enhancement(s)?

When nokinikon persuaded me to try Nik/Effex filters, the ND was the best of a bad bunch there for me, but not liking Effex in general, I soon realised it was only allowing one to be lazy and you get the same result with normal tools.

Quite often it seems that if you need a top-down grad filter, you also need it side to side and PP allows that which I imagine would be difficult with physical?

[Colin Southern]

Hi Chris,

It's one of the "great debates" of photography that rates right up alongside Mac -v- PC and to use or not to use UV filters for protection!

I use both techniques - sometimes even on the same image. As a rule I think it's accurate to say that you can get good results either way - but if you're after world-class results then you really have to know what you're doing, with either technique; I've seen people hold up "shining examples" of how easy it is to do it digitally, but upon closer inspection the images often have weak transitional areas - I've also seen examples from those who do know what they're doing and can affirm that they pass the acid test with flying colours.

With GND filters people often (wrongly) assume that they're only suitable for some fairly narrow-range types of image, but in reality due to the fact that (a) when attenuating light through the lens a typical reduction of 3 stops is barely noticeable to the human eye (and yet makes a BIG difference to the real world dynamic range that the sensor is now able to capture) and (b) due to the vagrancies of the "rules" of local contrast, in reality, it's not as limiting as people might think - and often the images survive far better than people would first think.

Movement also complicates the combining of bracketed shots - worst case scenario is if there is movement in the transitional zone - but you can also get issues when (for example) reflections don't match.

At the end of the day I feel that yes - there is still room for GND filters - my position is that - just like digital GND filters - they're just another tool in the toolbox - sometimes they're the better tool to use, sometimes not. Interestingly, most top photographers still tend to favour "getting it right in camera" - My personal experience is that if you've got the gear then it's easier to get a better result "in camera" than doing it digitally.

[Colin Southern]

When nokinikon persuaded me to try Nik/Effex filters, the ND was the best of a bad bunch there for me, but not liking Effex in general, I soon realised it was only allowing one to be lazy and you get the same result with normal tools.

On a side note Chris, I think that a big part of the problem is that the traditional teaching for doing Digital GND is based around a highly feathered layer mask - but a mask at 100% opacity none-the-less. The technique I prefer is to paint the layer mask manually, but set the brush to (a) something very soft and (b) have opacity set to something like only 5% so you can GRADUALLY transition the images as you see fit - makes blending a lot smoother and controllable.

[crisscross]

I should clarify that In NX2 one makes a 'gradient selection' with whatever values one wants for the gradient strength 1-100 at beginning end and a sliding intermediate point. The gradient need not be orthogonal to the image and 2 or more can be superimposed. They can also be contained by a lasso so as to affect, for example, only the sky; I dare say you could also do an inverted one for a water reflection...mmm must try that as it might also help to partially equate the exposure difference (just slight lack of decent water hereabouts )

[Alis]

Thanks, Colin and others for the great advice. It is a lot of information to digest. I am planning to do this when the weather warms up a little bit, so I have a lot of time do absorb the information.

I was actually thinking of buying the screw-in type, as that looked more practical to me but I agree now that it is less useful and more limited.

On a side note Chris, I think that a big part of the problem is that the traditional teaching for doing Digital GND is based around a highly feathered layer mask - but a mask at 100% opacity none-the-less. The technique I prefer is to paint the layer mask manually, but set the brush to (a) something very soft and (b) have opacity set to something like only 5% so you can GRADUALLY transition the images as you see fit - makes blending a lot smoother and controllable.

Colin, this sounds very interesting. Is it possible to elaborate on the technique a little bit more. I am asking this, because, I can imagine a situation where (let's say if I have only one type, 3 stop) there would be a need for a hybrid solution; capturing something but later do some digital processing to bring out more detail...

Also, I was thinking, is it possible to program the camera so that we can actually change the sensitivity of the pixels across the sensor, similar to what the filter does, so that different parts have different sensitivities and that it can be adjusted to fit different situations. I have a feeling it has been done already but if not, I will patent it right away!

Thanks again,

Sedali

[Colin Southern]

I was actually thinking of buying the screw-in type, as that looked more practical to me but I agree now that it is less useful and more limited.

Under-statement of the week!

Colin, this sounds very interesting. Is it possible to elaborate on the technique a little bit more. I am asking this, because, I can imagine a situation where (let's say if I have only one type, 3 stop) there would be a need for a hybrid solution; capturing something but later do some digital processing to bring out more detail...

Strangely enough, in my experience, 3 stops is usually about right - 2 is never enough, and 4 could be handy at times, but not often. It just seems to be one of those almost "constants" where the dynamic range of many (landscape) scenes always seems to exceed the dynamic range of the sensor by about 3 stops, with the "slack" being picked up by the available lattitude in a RAW capture.

Hybrid solutions are fine (at the end of the day the end justifies the means) - and often I will bracked shots even with a GND. The thing to remember is that it's NOT just a case of taking two shots (one for foregound exposure and one for background exposure) and then transitioning the hiding of the bad bits whilst revealing the good bits with a simple gradient mask or highly feathered mask - for best results you have to adjust the two seperate images with their own bound layers (particularly levels layers) - then bring the two together and see how nicely they play together and blend them using one of a number of techniques - this is the bit where I like to hand paint the mask using a very low opacity - you can easily just sit back and look at the image and "tap tap" here and there with the mouse to do the job carefully - and if you make a mistake then ctrl-Z it or press "X" to swap the colours around and start working backwards.

Probably best example I have to hand is this one:



... where the area above and below were merged using that technique for a seamless job (the bottom portion was itself an average of 3 shots, but thats another story).

Also, I was thinking, is it possible to program the camera so that we can actually change the sensitivity of the pixels across the sensor, similar to what the filter does, so that different parts have different sensitivities and that it can be adjusted to fit different situations. I have a feeling it has been done already but if not, I will patent it right away!

Not really practical - it's a physical limitation of the sensor design. High dynamic range sensors have been trialed where they have 2 photosites in each location (large and small), but to the best of my knowledge it's never really caught on.

Hope this helps!

[Dave Humphries]

Hi Sedali,

Also, I was thinking, is it possible to program the camera so that we can actually change the sensitivity of the pixels across the sensor, similar to what the filter does, so that different parts have different sensitivities and that it can be adjusted to fit different situations. I have a feeling it has been done already but if not, I will patent it right away!

That's an interesting idea.
I could imagine that if the sensor were interline type, where I believe each row of pixels is read out, a bit like a TV line, then one might think that if ISO is effected by different analogue gain, this could be increased or decreased on a line by line basis, for each row.

However, the downside is that it would only work for horizontal splits in landscape orientation. That said, I can also imagine how it would be possible to also vary gain along the row of read out pixels too, but I suspect that setting all this up via menus at time of capture is just going to be too much for 99.9% of users to cope with. At least it would be too much without a touch screen LCD or similar to get a slick user interface. That's a LOT of development for 0.01% of the customer base and it would obviously have an impact on price. Still a nice idea though ...

Regards,