Hot Mirror/IRND/Mirror ND Filter Matrix [Parts 1-4]

[jbeale]

This is an excellent comparative test, it is great to see this info published!

I don't know what other tests are still to come, but just FYI, this test does tell only part of the story. The IR LED source used here is probably a very narrow slice of the IR spectrum (around 880 nm? just a guess). So this is really just one data point (but a good one) on an extended curve. In most "real" scenes you will have the entire IR spectrum, so you care about all the different wavelengths of IR that the Red sensor picks up (I'm not sure how much that is). If any tests were done looking at color shifts with tungsten lights, day exterior, golden hour etc. that will tell you more about the "full spectrum" response.

[ayarbro]

Very good! Part two seems much more objective than part one, thanks for the useful info thus far.

[Brook Willard]

The Rosco filter was my first choice after performing the tests yesterday... hands-down. It cut the most IR light, it had acceptable off-angle viewing characteristics [I thought it was the best by eye... but these data don't lie] and cost the least. Then I heard that Rosco isn't planning on making a true sandwich filter in the short term. They'll be out eventually [and they'll certainly kick ass then...], but in the short term, I don't want to have to baby a filter to protect its coating.

Ultimately whichever company can get me some 4x5 and 6x6 filters first [Rosco or Tiffen... not the Schneider] will win my money. :)

[Brook Willard]

Very good! Part two seems much more objective than part one, thanks for the useful info thus far.

What wasn't objective about the first test? The off-angle viewing characteristics are massively important to the usefulness of these filters... and the data were presented clearly...

Just saying.

[Evin Grant]

Part 3: Outdoor Field test

This is where the rubber really hits the road...
The first thing I have to say is that all the mirrored filters (Hot or otherwise) do the trick as far as attenuating the IR issue with the Red One. They all approach that goal from different directions but in the end they do all deliver the goods.

First off let me thank Danna Kinsky for being the patient model and AC as well as Christian Sprenger and Brook Willard for assisting with this test.

These frames were all processed identically using the Grey card in the top control image as a color reference. Thus any deviation in color shift or exposure is consistent with the filter's properties.

So first off Danna and her lovely IR reflective nylon pouch and gray card...


I had expected this pouch to reflect a stronger purple cast than it did (Based on the EVF image) but you can still see it and the loss of black contrast in the foliage behind is considerable. This first set is all Tiffen because they have the most filter combinations. In this test the IRND seems to contribute a bit more than was suggested by the Lite Panel test. Still there is little to choose between them. The Hot Mirror alone does a fine job. The Tiffen line seem to cut about 2/3 of a stop more light than their stated density, this may be because of the lack of an IR boost in exposure. Cross reference your in camera meters if you plan on using an external one.
The Tiffens seem to exhibit a warm/green cast to the image as well.



Here we can see the Schneider, Rosco and Pancro also getting the job done. All of these eliminate the color cast of the pouch and punch up the blacks of the foliage. The Schneider filters add a slight cyan cast and the Rosco ever so slight Red. The Pancro is almost dead neutral, and the only filter that seems to be exactly the density stated. Although this may be because the Panchro is the only filter that is cutting an equal amount of both visible and IR light.

Here are the same tests done with some levels and color correction added for a peek at what a more corrected image would look like. I let photoshop auto color correct each frame thus accentuating the individual nature of each image.


This is obviously more subjective but you can clearly see the cast in the pouch now as well as the color dilemmas you might face in post.


Again the cooler Schneider's character is obvious, the Rosco is doing very well, especially in regards to the nylon and the Panchro is still the most neutral and best exposure, however a slight hint of blue is now visible in the nylon., but no worse than in the control, it's only when viewed next to the Rosco that it is apparent.

[Evin Grant]

Same as the above but with Color Charts.



And color corrected...

[Evin Grant]

One last thing before we move on to Part 4, Tungsten.

The Tiffen, Schneider and Pancro filters are all professional camera filters that use a sandwiched or otherwise hardened design to resist fine scratches and abrasions. They all seem to perform as expected. The Rosco filter I received was a single coated prototype on some 1/2 thickness plate glass and was never intended to be a permanent camera filter. That being said the Rosco production filters will be produced on standard thickness, optical quality glass but the coating will remain a single layer, un-sandwiched design. This is unfortunate because they have arguably the best formula for iR attenuation but I'm concerned that the coating will not be robust enough for heavy duty production or rentals. Although I have urged them to go to and sandwiched design this first batch will for sure be single coated. I will still recommend the filter, especially given it's great price, but all who have ordered or are considering an order should be aware that extra special care will be needed when handling and cleaning this filter. You should treat it as you would any coated camera lens, and avoid any unnecessary contact, if possible.

[Alexander Christ]

Great tests Evin. I'm in the group buy from Michael, so far a good decision. :-) Looking forward to the field tests. Thank you.

[Evin Grant]

I've added a "real world" example of the Hot Mirror filters using the Nikon D3 and a 17-35mm lens to eliminate any potholing or inconsistency in post processing.

(The marked focal lengths are Red One 4K 16:9 equivalents)

[Michele Gavazzeni]

Why Tiffen have called his filters IR-ND if they don't cut IR.... beacuse they simply don't look at the top right image ir-nd .6 on the left ND .6 on the right... no difference.
Fraud?