Last month in DV101, I examined the issue of infrared contamination. I discussed the problems of IR radiation with digital cameras, especially when light is severely attenuated with neutral density filters. A couple months ago I spent a day at CamTec Motion Picture Cameras with cinematographers Christopher Probst, Phil Holland and Jesse Brunt to test a number of infrared-cutting filters on both the RED EPIC and the ARRI Alexa. Last month I shared the results of tests with the RED EPIC; this month we’ll revisit that test and see how our filters worked on the Alexa.
- Tiffen ND 1.8
Traditional neutral density filter in 1.8 (-6 stop) density with no additional infrared filtration. Used as a control to provide a basis of comparison.
- Tiffen Full Spectrum IRND 1.8
Combining an infrared filter and 1.8 ND filter, this offering cuts down wavelengths equally until 680 nm, at which point it cuts off sharply to cut IR contamination.
- Tiffen Hot Mirror
Introduced in 2008, the Hot Mirror filter cuts only IR radiation and does not affect light. These filters start at 700 nm and continue to block IR all the way through the 1,000 nm range.
- Tiffen T1
An IR-cutting filter that incorporates a slight green hue that helps to reduce oversaturated reds that can bleed and cause trouble in standard definition signals.
- Formatt Hitech Prostop IRND 1.8
Introduces a slight blue cast to a neutral density and infrared-cutting filter. Formatt states that this design is intended for outdoor photographers to optimize the image while reducing IR contamination.
- Formatt Hitech Hot Mirror ND 1.8
A neutral density and Hot Mirror combined in a single filter.
- Schneider Optics ND 1.5
Classic ND without IR filtration. Used as a control.
- Schneider Optics True-Cut IR-750
Schneider’s version of a Hot Mirror, this filter lets light through but blocks wavelengths starting at 750 nm, allowing more red wavelengths through before cutting infrared.
- Schneider Optics Platinum IRND 1.5
With a cut starting at 700 nm, Schneider’s premium line of IRND filters incorporates IR filtration into standard ND.
Although our tests show that some of these filters are more effective than others on either the EPIC or the Alexa, it’s important to note that every sensor reacts to IR differently—you need a different filter or filter combination for each camera in order to best eliminate the contamination in your image. While one filter might not work for one camera, it’s probable it will work well for another.
We shot the Alexa to SxS cards in ProRes 422 to simplify our workflow.
The test scene was lit with a tungsten 1K Fresnel to a stop of f/16 at 500 ISO. We then incorporated various ND and NDIR filters, cutting the exposure down to T2 or T2.8 (depending on the strength of the ND available) to find the best combination for this camera.
The control image. This is how correction should look if it’s working properly and eliminating IR contamination. Refer to this reference when interpreting the quality of the filtration being used.
As I discussed last month, the key for the EPIC was to incorporate a Hot Mirror filter. Any manufacturer’s IRND filter in combination with a Hot Mirror or a True-Cut 750 produced amazing results, although we found that the Tiffen IRND + Hot Mirror combination yielded the best results.
ARRI has incorporated a very effective IR cut filter on the Alexa that acts the same as a Hot Mirror on the sensor’s filter stack, so the Alexa already deals with IR contamination extremely well. If you look at the EPIC with ND only compared to the Alexa with ND only, you see that the EPIC’s image is severely compromised, while only the reds in the Alexa image are biased; the rest of Alexa’s image retains its color fidelity and contrast. Because the Alexa already incorporates an effective Hot Mirror, adding a second Hot Mirror in front of the lens has no effect. If anything, it slightly desaturates the reds, but it is an extremely subtle result.
As with the EPIC test results, I’ll simplify here and just look at the bottom left of the test image, which includes the black nylon material and the X-Rite ColorChecker Classic color check target.
Using traditional ND filters, without additional IR control, we easily see the effects of IR contamination in the image from the EPIC camera (top left and right) and the considerably less, yet still noticeable, contamination with the Alexa.
The next two frames show the results of using only basic ND, which stops the light but allows the infrared radiation through. You can see the ugly bias not only in the blacks but in all the colors in the image. The IR cut filter inside the Alexa does a considerably better job at dealing with infrared contamination even without additional filtration. The blacks in the image are still compromised, but the overall contrast and color fidelity are much more sound.
We’ll start with select combinations from the Formatt line of filters. All three of the Formatt Prostop filter iterations had excellent results on the Alexa. The T1 adds green to the image, deliberately, and I don’t see an immediate benefit to that. Perhaps in an environment with a lot of red, the additional green can help balance out the image, but that’s purely supposition and not evidenced in our test here.
The Prostop IRND alone does very good work with the IR, and the addition of the True-Cut has no discernible effect on the IR or overall color. The clear winner here is the Formatt Prostop IRND filter.
Four Formatt filter combinations.
Interestingly, the T1 plays a strong role in the first iteration. With just straight Schneider ND and the T1, we see there’s a deeper loss of light (we should have opened up another 1/3 stop), but the T1 cleans up the red in the blacks pretty well. The Schneider straight ND with the True-Cut 750 filter doesn’t get the job done. We still see quite a bit of red contamination in the image. Basically the True-Cut is cutting the same range as the Alexa’s internal IR filter, so it’s ineffective as an additional tool here. We get a significantly better result with Schneider’s Platinum IRND filter, which clears up the red, trues up the colors and is an exceptional filter for use with the Alexa. Finally, combining the T1 with the Platinum IRND puts too much green in the image for my taste. Although this is correctable, it’s not necessary. The Platinum IRND does the job fine by itself and doesn’t need additional help.
The various Schneider filters on the ARRI Alexa.
When it comes to Tiffen, the results are pretty clear. The Hot Mirror and True-Cut 750 have no effect. The T1 puts green into the image, and I can’t see any benefit to that with Alexa. The clear winner with Tiffen is the IRND filter, which cleans up the contamination beautifully.
The clear winner from Tiffen is the IRND filter.
The key to EPIC filtration was to use an IRND in combination with a Hot Mirror or a True-Cut 750. With the ARRI Alexa, an additional Hot Mirror is of no benefit, nor is a filter like the True-Cut 750; it needs a wider range of IR/red cutting to eliminate the contamination.
The internal IR filter in the Alexa does a great job all by itself; there’s little to no need for additional IR cutting if you’re using ND of 0.9 or less. Above 0.9, you need an additional IRND filter with a wider cut range than the internal one, which cuts off somewhere above 700 nm. ARRI specifically left that cut higher, just outside the visible range, because the higher wavelengths of red light are necessary to obtain pleasing skin tones.
The filters that worked well for the Alexa were the Tiffen IRND, Formatt Prostop IRND and Schneider Platinum IRND. I saw no real benefit in incorporating the T1 with any of those filters.
Across the board, the Alexa handles infrared radiation better than the EPIC because it already has a good IR filter built into the sensor’s filter stack. Only wider-range IRND filters add to that cut to clean up contamination from the use of heavy ND filters.
It’s important to reiterate that the filter combinations that worked well for the ARRI Alexa or RED EPIC may not work as well for other cameras and sensors. Every camera requires its own filtration solutions to combat IR contamination, and only by testing can tell you what the right combination is. Although some of these filters were less than effective for the Alexa or the EPIC, that doesn’t mean the filters don’t work—they may work wonderfully for another camera system.