Thursday, April 10, 2008

Does size matter? Darwin Wiggett puts the 4x6 ND Grad format to the test

If you've read the previous contributions from Darwin Wiggett, you know he likes to figure things out for himself. In this story, Darwin tests some widely held assumptions about using Singh-Ray's 4x6-inch Graduated Neutral Density filters -- or any ND grad for that matter. Here's his report:

Lately there has been much discussion on photographic websites suggesting that 4x6-inch grad filters are ‘too big’ to be used effectively on APS-sized or full-frame 35mm digital cameras. The issue is that the transition zone from dark to clear on the 4x6 inch grad filters is too wide to be useful with digital SLRs whereas the transition zone on the P-sized filters (a smaller filter measuring roughly 3.25 x 4.75 inches) is more suitable to digital SLRs. Some photographers have thus ‘pronounced’ that 4x6 grads do not work with digital cameras

I thought I should find out if this assertion was true since I have both P-sized grads and 4x6 inch grads. And I know many photographers making great images using 4x6 inch grads -- are they using something that isn’t working as well as it should? Maybe my assumptions about grad filters are incorrect. It was time to do a little testing! I used my Singh-Ray P-sized grads and Singh-Ray 4x6-inch grads for all the testing below.

Assumption #1 -- The smaller the aperture used, the more defined the transition line of the grad will appear in the photo.

I have written articles for years about filters and always told people that the aperture they choose will affect the ‘hardness’ of the grad line. At f22 the grad line will be more abrupt than at f4. This assertion seems as intuitive as the law of gravity or the law that ‘all puppies are cute.'

Well... lesson number one: never “assume”! I was totally surprised by the results. With a 17-40 mm lens at 17 mm, I could see no shift or change in the grad line as I tried different apertures from f4 to f22 (photo 1 -- the subject here is my garage wall shot from 10 feet away). I tried different focal lengths of lenses, I tried the 4x6-sized Singh-Ray grad, and I even tried soft and hard Singh-Ray grads of various densities -- same results every time!

Conclusion #1 -- Aperture does not affect how the grad line appears in the photo.

Assumption #2 -- The transition patterns on the P-sized filters and the 4x6-sized filters are the essentially the same

I never really looked closely at the two types of filters to see if they differed. When I put them on a light table, it did look like the Singh-Ray hard-edge 4x6 filter was slightly ‘softer’ in the transition zone than the Singh-Ray P-sized filter (photo 2). And the Singh-Ray 4x6 soft-edge grads looked slightly harder than the Singh-Ray P-sized filters (photo 2). Hmmm, maybe there is something to the assertion that the two sizes of grads perform differently? We'll see what some real-world testing can tell us.

Conclusion #2 -- There are only slight differences between the two filter sizes in terms of their transition zones

Assumption #3 -- The two sizes of grad filters will produce the same imaging results

When the two sizes of grad filters are used with the same lens, at the same aperture, on the same camera and in the same filter slot on the filter holder (the closest one to the lens -- slot #1), we do see that the bigger Singh-Ray 4x6 hard-edge grad gives a slightly softer result than the Singh-Ray P-sized grad (photo 3).

What about the soft-step filters? When we compare the two sizes of the Singh-Ray 3-stop soft-edge grad, it is hard to see much of a difference except that possibly the P-sized filter is slightly darker at the top (photo 4). I would think that there is slight variation in the manufacturing process of grad filters and that a little variation is probably to be expected. So far there are detectable differences but not enough that I think it would matter in a practical picture-taking sense (more on that later). So I tested on...

Conclusion #3 -- The two sizes of ND grad filter give slightly different results.

Assumption #4 -- The further away the grad filter is from the front surface of the lens, the harder the transition line will appear in the final photo.

Guess what, I am finally right about one thing! The Cokin P-holder, the Cokin Z-Pro, and the Lee filter holder (the latter two are both designed to hold the bigger 4x6-inch filters) have three filter slots each. When you put the filter in the #1 slot closest to the lens, the grad line is less distinct than when the filter is in the #3 slot furthest from the lens (see photo 5).

When the Singh-Ray 4x6 hard-edge grad is placed in slot #3 of the filter holder its transition zone is very similar to that of the Singh-Ray P-sized grad in slot #1, although the P-sized filter still has a slightly harder edge (see photo 6). In a practical sense there is no difference in the behavior of these two filters if the larger one is slightly further from the lens than the smaller one. Remember this is my comparison with just two filters of the same density but of different sizes -- I suspect (read ASSUME!) there might be variation in the transition zone of the grad lines on all brands of grad filters no matter what size due to the manufacturing process. I don’t have access to many filters of the same type to test this assumption though.

Conclusion #4 -- The further the grad filter is held from the lens, the more defined the transition line becomes.

Assumption #5 -- The longer the focal length of lens, the less distinctly the grad line will appear in the photo.

This one had to be true. Whenever I used a wideangle lens, I could easily see the grad line. With a normal-focal-length lens the grad line was less distinct. Was I wrong again? Fortunately, I had this one correct. The shorter the focal length of lens used the more distinct the grad line gets (photo 7 -- score two for Darwin!).

Conclusion #5 -- The wider the lens, the more distinct the grad line.

Assumption #6 -- 4x6 inch grads won’t work on cameras with small sensors

I put a Singh-Ray 4x6-inch grad on my Canon G9 compact digital and indeed the beneficial effect of the grad filter is obvious even on this thumbnail sized sensor (photo 8).

Conclusion #6 -- Big grads work well on small sensors.



What about comparing results in the real world?

Ok, so my little photographic tests of the wall of my garage taught me a few very important things. But what really matters is how the filters function in the real world of landscape photography. Will the Singh-Ray 4x6 size grads give different results than the Singh-Ray P-sized grads when shooting the same scene? The only way to find out was to do my own testing in the field.

One of the main reasons I like to use the 4x6 filters is to prevent vignetting when shooting with wide-angle lenses. For example a Cokin P-Holder will vignette seriously when used with a 17-40mm lens on a full-frame camera. To avoid vignetting with my 17-40 lens, I had to cut off the outer two slots on the P-holder so now the holder is only capable of holding one filter (photo 9). For me, this limitation can be a problem.

Even then, at 17mm there is a tiny, tiny bit of vignetting on the corners of the frame. If I use a bigger holder with my 4x6 filters, I can have up to four slots (three regular slots and one gel slot) on the holder and still not have vignetting -- even when using a wider 16-35mm or the 17-40mm lens on my full-frame camera (photo 10). The benefits of the larger holder and 4x6 filters are obvious, more filter slots to hold more filters and more flexibility to place grads closer or further from the lens for different effects. In a perfect world photographers with wideangle zoom lenses should opt for the larger 4x6 filter size for these greater benefits. But, if in my real world testing, 4x6 sized grad filters had behaved much differently than the P-sized grads, then I might need to reconsider which system to buy.

In my tests, however, I simply could not see much, if any, difference in the photos I took with either sized grad -- no matter where I placed the grads in the filter holder. For example, in photo 11, I used a 3-stop Singh-Ray grad of each size on the same scene and in the first slot of the filter holders. Differences are so minor as to be negligible. The imaging effects of both filters are virtually the same.

In photo 12, the scene demanded a hard-edge grad, and the Singh-Ray 3-stop hard-edge P-sized grad performed beautifully (note the tiny bit of vignetting from the filter holder on each corner – photo 12a). I captured the same scene in photo 12b, but this time I was using the Singh-Ray 4x6-inch grad in slot #1 of the Lee 4x6 holder. In photo 12b the transition is a bit softer with the 4x6 filter than with the P-sized filter, but even so the end results are pretty close. If I moved the bigger 4x6 filter to the outer slot (slot #3) of the holder then the results were exactly like I got in photo 12a with the P-sized grad. So, in the real world there is very little difference in performance between the two sizes of filters.

As far as the assertion that 4x6 grads are too big for 35mm cameras, I think we can see from the results presented here that this assumption is in error. In fact, even on the relatively tiny sensor of my Canon G9, the big Singh-Ray 4x6 hard-edge grad did a great job of holding back a bright sky on a defined horizon (photo 13).

Conclusion -- In the real world, with typical landscape scenes, the P-sized ND grad and the 4 x 6-inch ND grad perform essentially the same way. The 4 x 6 grad, however, offers greater flexibility with wideangle lenses and those lenses with large front elements. For me, I plan now to carry just one size of filters in my bag – I am moving up to the 4x6-inch filter format because it works with all my lenses and with all my digital cameras.

For more information about Darwin, visit his website, http://www.darwinwiggett.com/main.html

7 comments:

Anonymous said...

Why are all the images X'd out in this post? All other posts are fine.

Andrea Moro said...

Very useful article. Thanks!

What a pity that Singh-Ray aren't available in Italy!

Andrea (Italy)

Singh-Ray Filters said...

Hi Anonymous,
Yes, seems there was a glitch on Google's photo servers over the weekend, so we had to re-upload all the images this morning. Everything should be OK now.

Tim Parkin said...

Firstly, thank you for an excellent article on grads and their relative behaviour. I would like to make an observation about the 17-40 lens that may have an effect on the results of the focal length test. Firstly the 17-40 has a nodal point so close to the front of the lens it's almost not believable but an aperture that is towards the rear of the lens - this will tend to minimise the effect of aperture and also make the grads a lot softer (the closer the nodal point to the grad, the softer the result).. The retrofocus design also makes the light rays more parralel and hence there is little softening effect at smaller apertures. A non-retrofocus design (or at least one not as extreme as the 17-40) would possibly be a better choice for this test which would also probably have nodal point and aperture coincident. Perhaps the 35mm f/1.4 or f/2 would be a good choice? Again, please don't take this as a criticism, the article is very topical and very informative.

Tim Parkin said...

Great article and I really appreciate the work put in. I've done a little theoretical work and posted an article on my blog to back up some findings and to raise a question about a couple of things. http://blog.timparkin.co.uk/2008/04/black-art-of-graddage.html

fiveonenine-ca said...

Thank you, this is so useful! I'm looking at upgrading my filters from Cokin filters/Cokin "P" holder and trying to decide if I need to move to 4x6 instead. After weeks of research, I think I can stay with the "P" holder until the day I move to a full-frame camera.

Jay Gould said...

Darwin, for me two things were important: 1) you conclusions about the 4x6 grads, and 2) you used the 17-40. I am moving up (?) from the 16-35 to the 17-40 to continue my goal of lightening what I carry. Thanks again and again for the useful information.