Sunday, April 26, 2015

Review: Deep ND Filters Compared (updated)...



[Update April 2015: Added a few comparison images and at the end of this post taken with a Hoya Pro ND1000, as well as some notes about that filter.]

This article is a quick review on some of the latest and greatest IR-blocking ND filters and how they stack up against the “old standby," a standard B+W 10-stop ND which has no IR filtration. The above image is a screenshot from Lightroom, showing a side-by-side comparison with the image thumbnails conveniently colour coded to match the colour cast of the filter used! The B+W 10-stop ND gives colour images a somewhat ruddy-brownish colour cast (the red thumbnails) and it has been speculated that this is due to infrared light affecting cameras with sensors that still have some residual IR sensitivity.

In response to the IR issue, several companies have come along and made filters that purport to solve this with the promise of easily corrected colour balance. The issue is that if a scene being photographed is heavy in IR, a sunny day with foliage perhaps, or anything that reflects a lot of IR, then a deep visible light blocking filter that lets through IR, like the B+W, might affect some areas of an image differently than others, meaning that a simple WB correction in post might still yield an image with strange looking colours or colour casts. There are two image galleries that accompany this article as follows…

There are details about what adjustments were made in the second colour corrected gallery. Also, the filters being discussed in this review are as follows (with the image filename suffix indicated in brackets, also detailed in the galleries)…

- (TA) Tiffen Apex XLE-series 10-stop IRND
- (BW) B+W 10-stop ND (110 ND 3.0 - 10 BL 1000x
- (FHF) Formatt-Hitech Firecrest 16-stop IRND



Keep in mind that the amount of sunlight varied for the tests and since filtered sunlight tends to have a fair bit less infrared, the advantage the IRND filters had might be variable. I will describe the four sets of shots below…

1-4: Photos of an X-Rite ColorChecker Passport. Sunlight was slightly filtered by some thin cloud. The fabric on the camera bag probably reflects a lot of IR and you’ll see the most dramatic colour changes there. I would say the Firecrest renders the colour of the bag closest to reality and here, the B+W seems the worst, although none of them are anywhere near accurate. Looking at the patches of the ColorChecker, I would say the B+W still did surprisingly well after correction!

5-8: There was very little sun for these shots with somewhat dense cloud cover. You could barely see where in the sky the sun was, when looking at the clouds. There was likely some variation in sun from shot to shot since it took a moment to switch filters out while precariously balanced, with tripod, on top of a narrow rock. For this set I used my Fujifilm 16-55mm zoom at 21mm with the 77mm ND filters threaded directly on the front.

9-12: For these, luckily I almost had full sun! Very thin cloud cover only with strong shadows and contrast visible. I used my Fujifilm 10-24mm zoom at 10mm (77mm ND filters used with a 77-72mm stepping ring) and despite the oversized filters, I believe some mechanical vignetting can be seen in the B+W and Tiffen shots. The Firecrest is extremely slim, even having surprisingly thin glass and it seems to yield brighter corners. Do note that there were some people walking through the scenes during the exposures. For example, there is a blueish colour smear in the lower right corner of the Tiffen Apex shot.

13-16: These were shot with the sun above and to the left, just outside the frame. No lens hood was used and I just let the sun hit the front of the filters at a glancing angle. Interestingly enough, both the Firecrest and the Apex filters show colour casts through part of the frame. The Firecrest is showing a magenta cast at the top and bottom, whereas the Apex is showing a green cast. This might be due to internal reflections from the more metallic coatings on both those filters? Interestingly enough there seems no hint of any differential colour casts from the simpler, uncoated B+W filter.


A few additional comments on the photos. First off, it seems that none of these filters will give truly accurate colour rendition with a simple click-WB! I may try to do a custom calibration with the ColorChecker Passport photos... however not only did I run out of time, but since there was slight overcast I felt the resulting calibration might not be as accurate or useful as it would have been in full sunlight. Based on the different colour renditions, which of these filters do you find the most pleasing? Depending on the subject, the amount of IR as well as direction of the light, one’s preference might change too. In addition, it is also possible that a different camera might behave slightly differently as far as the degree of colour cast and how much advantage the IR filtering actually has in practice.

I must say, that Firecrest filter is impressively dark for such a thin filter! We are talking about going from exposures of several seconds with the 10-stop filters to several minutes with the 16-stop Firecrest! It’s pretty convenient not having to stack NDs and achieve really long exposures in full sunlight, especially when using ultra-wide lenses. The way that Formatt-Hitech gets such a thin filter so dark is by using a carbon-metallic coating. They call it “hyper neutral” but I’d say that’s overstating its neutrality since one can still see colour casts on not only the uncorrected shots, but on the WB corrected ones too.

Note that I have identified an issue with the circular Firecrest ND filter I tested. I was actually getting visual distortion in parts of the images, a stretching in some corners, most visible with my ultra-wide lens. To see this, click photo 9 in either gallery, then press the right-arrow key to get to the Firecrest shot, then the left-arrow key to get back to the unfiltered shot. Do this repeatedly and you can see the lower right corner stretching away and possibly some other weird distortions happening in other parts of the image. In comparison, the B+W filter is absolutely perfect with no visible distortion or degradation in image quality even when viewed at full resolution, apart from the mentioned vignetting. The Tiffen is also essentially perfect without any image quality loss and less vignetting than the B+W. It’s not really visible on the smaller web images, but the stretching distortion also causes a slight loss of sharpness in parts of the Firecrest shots.

Afterwards, Jason here noticed that when you look at a reflection of a straight line in the Firecrest filter’s glass, you can actually see a slight “rippling” in a few spots near the edge and also an overall slight “barrel distortion” warping of the reflection. Perhaps the thin glass is under some stress from the filter’s outer metal ring, causing this distortion? I am only speculating with that, but looking at one of the flat 4x4 inch sheets of Firecrest ND we sell, it is essentially perfect with no visible distortion. I have brought this issue to the rep’s attention and since the sample filter I tested was said to be a production unit, the distributor has requested that we return all our circular Firecrest ND inventory to allow the factory to carefully check them over. Perhaps they are concerned that somehow a small batch made it through QA that shouldn’t have, and they want to proactively ensure that we don’t sell any less than perfect units…



As far as mechanical build, the B+W is by far the most robust, with thick glass, perfect threads and a weighty, solid brass mounting ring. Next was the Tiffen, with accurately cut threads but a lighter aluminum mounting ring. Coming in last was the Firecrest. Its thin glass surely contributed to its somewhat cheap feeling construction but its filter threads also seemed a slight bit undersized, not engaging with as much confidence as the other two filters, feeling a bit sloppy. Then of course there is that issue of glass distortion.

Lastly, it seems a 10-stop filter may not really be a 10 stop filter! Using the PhotoPills app on my iPhone to calculate exposure times, I found that the Firecrest filter seemed quite accurate to its 16-stop claim, the Tiffen was a little bit lighter and the B+W a fair bit darker than 10 stops. In fact, even though they are both 10-stop filters, the B+W seemed about one-and-a-third stops darker than the Tiffen!

Your mileage may vary, but personally I would choose the regular B+W 10-stop ND for myself, based on these tests. On the pure landscape shots, I actually prefer the colour rendition of the B+W filter overall, and the fact that it does not suffer from any grazing light colour casts and does not exhibit any distortion or loss of sharpness would make it my pick. On the other hand, if the Firecrest did not have the distortion and sharpness issue, I would likely buy one since I do like the fact that it does not vignette, and having a single filter that does 16 stops is quite compelling. If I were photographing man-made objects where colour rendition was important, the choice would be more difficult and perhaps the B+W would come in last for then, since the IR filtering of the Firecrest and Apex filters should offer them an advantage. Finally, different cameras and sensors may also behave slightly differently and also the degree of colour cast difference may vary.

April 2015 Update - Hoya Pro ND1000 10-stop ND Filter

After doing this ND filter test, my friend Bill asked me why I had not tested a Hoya ND?! In fact, at the time of the original test, I wasn't even aware that Hoya had a really deep ND... but sure enough, they have both an ND1000 (10-stop) and an ND64 (6-stop) in their Pro series line. I emailed our Hoya rep and he kindly enough dropped off a filter for me to test out soon after.



From a build-quality standpoint, the Hoya is on par with the Tiffen, having a solid, well machined aluminum casing, just not brass like the B+W. However optically, it actually proved to be the best filter of the group and I wish I had it when I shot my original comparison. I didn't have the other filters to test side by side with the following comparison, except my own B+W 10-stop, but remarkably, there is absolutely no colour cast with the Hoya! You can see below, with all three photos processed in exactly the same way with the same daylight white-balance, that the Hoya ND shot is virtually identical to the unfiltered shot. The B+W shows the same ruddy colour cast it always does but the Hoya seems to be essentially perfect. In addition, the Hoya added absolutely no visible distortion or blurriness to the shot and seemed to have no differential colour cast either. Impressive! First up, the 10-stop Hoya test shot…



Next, the unfiltered shot. As you can see, there is virtually no difference in colour between it and the Hoya shot…



Lastly, for comparison, here is the B+W shot of the same scene with the same WB settings...



I hope you found this comparison useful! If you want to try one for yourself, we are now renting the Tiffen and the B+W ND filters (and will likely be adding a Hoya to the mix as well). Don’t forget that if you decide to buy one after renting, you can apply up to two days of the rental fees toward your purchase. Call or email us for more information...

No comments:

Post a Comment