next thread »

[Bob]

Now that Schneider Optics new Platinum Series IRND are becoming more available, it felt good to revise my ARRI Alexa Infra Red Filter Test. For my updated test I started to involve multiple light sources, since this also can influence colour reproduction. Unfortunately I am not finished with the whole test, and to date I have only used HMI and daylight Kinoflo fluorescents. Hereby I offer my interim findings in the hope to further optimize and improving the test with your feedback and finding answers to some questions.

As I wrote in the introduction, I have run the IRND filter test on the ARRI Alexa working with in-camera SxS Pro Cards recording in Apple ProRes 444(4) Log C with no colour matrix, since this is the most used workflow here in The Netherlands. For this test I done all my light measurements with the in-camera false colour function (in REC709) and a Kodak Gray Card.

Beside the Kodak Gray Card I placed a self made IR test card (with 'natural' and synthetic 'black' fabrics), a GretagMacbeth ColorChecker Color Rendition Chart and my own right hand (for skintones) in front of a 24mm Ultra Prime Lens. I used the iris on the lens for stop compensation. This time I made a small selection of filters and only tested Schneider regular series ND's (0.3 - 1.2 and stacked them until 2.1), Platinum series IRND's ( 0.3 - 1.8 ) and Tiffen Full Spectrum series IRND's (0.3 - 2.1).

First I shot a clear reference on a 5600K preset and made a second exposure with the in-camera auto white balance on the gray card. I noticed that your exposure value will change when you capture more IR contamination, and setting the (right) exposure after you've done the 'correct' white balancing in the camera. Somehow I don't find it strange that IR contamination will influence your exposure.

It was the second observation with both IRND filter series that set a little food for thought. When reaching IRND1.2 in front of the lens I noticed that you get a different iris setting when you set your exposure after auto white balancing. With one manufacturer you have to open your iris 1/3 a 1/2 stop and with the other manufacturer you can close 1/3 stop on your iris setting. In the last case you also end up with a higher Kelvin value. Somehow I have the feeling that the 'green' tint in the filters influence your correct exposure. Strangely enough I only notice this phenomenon when I wanted to work with a proper white balance on the camera. I am looking forward to your knowledge a findings about exposure with IRND's on the Alexa.

Second, I am very curious how you take care of the colour tint shift coming with the filters. Do you make a camera and filter specific white balance compensation factor profile, and take this is into account when you set your white balance? Or do you colour correct the colour tint shift during grading and then start the actual grading?

Thanking you in advance,

Vincent Visser
(Digital Imaging Technician, Camera Rentals b.v., The Netherlands)

[Thomas Wright]

You have not written which filter did what?

Would help.

Ta

[Bob]

You have not written which filter did what?

With the Tiffen Full Spectrum series IRND filters I had to open the iris on the lens with 1/3 to 1/2 stop from IRND 1.2 and up.

The Schneider Platinum series IRND let me close the iris with 1/3 of a stop from 1.2 and up, and gave a little different (higher) Kelvin value at 1.5 and 1.8 compared to the rest of the filter range.

It also good to tell that I nhad more green tint compensation with the Tiffen's when doing an auto white balance on the camera.

Cheers,

Vincent

[Michael Borenstein]

Vincent,

Thanks passing along this info.  It's very much appreciated!

Regards
Mike

[Thomas Wright]

So, to clarify: When you had an ND1.2 on and you added a 0.3, instead of opening up 1 stop to get 18% grey green in false colour, you had to open up 1 and 1/3 to 1 and 1/2 stops and for the Schneider you had to open by only 2/3 of a stop?

[Bob]

So, to clarify: When you had an ND1.2 on and you added a 0.3, instead of opening up 1 stop to get 18% grey green in false colour, you had to open up 1 and 1/3 to 1 and 1/2 stops and for the Schneider you had to open by only 2/3 of a stop?

I will do my best to explain what I exactly saw, to eliminate the lack of clarity. In the case of using Tiffen Full Spectrum series IRND filter I noticed (with a daylight kino which was a bit green CC-3 and was 5800K):

IRND 1.2 * with a preset white balance of 5600K and CC 0 * I got T5.6 (with false colour)
IRND 1.2 * with auto white balance giving 5500K and CC -6 * I got T5.6- (closing 1/3 of a stop)

IRND 1.5 * with preset WB of 5600K and CC 0 * I got T4+
IRND 1.5 * with auto white balance giving 5400K and CC -7 * I got a T (opening 1/3 of a stop)

IRND 1.8 * WB 5600K and CC 0 * T2.8/4
IRND 1.8 * Auto WB 6200K and CC -5 * T2.8 (opening 1/2 stop)

IRND 2.1 * WB 5600K and CC 0 * T2/2.8
IRND 2.1 * Auto WB 6300K and CC -6 * T2.8- (closing 1/3 stop)

Schneider Platinum series IRND filters gave (with a daylight kino which was a bit green CC-3 and was 5800K):

IRND 1.2 * WB 5600K and CC 0 * T5.6+
IRND 1.2 * AWB 6000K and CC -4 * T5.6+ (I didn't closed the lens, but I could close 1/3 stop)

IRND 1.5 * WB 5600K and CC 0 * T4-
IRND 1.5 * AWB 6300K and CC-4 * T4 (closing 1/3 stop)

IRND 1.8 * WB 5600K and CC 0 * T2.8-
IRND 1.8 * AWB 6900K and CC -6) * T2.8 (closing 1/3 stop)

I hope that the information above helps a bit for the completeness of my questions. It's that your stop deviates slightly when you first set the white balance, and it's that phenomen I try to understand.

Regards,

Vincent

 

[Greg Foad]

I went through the same series of tests and decided for simplicity to grade in post.  I found no manufacturer significantly better than the other, they were just slightly different.

I've decided to let the variations fall where they may and correct later.  I was driving myself crazy with variations on individual filters and color temps.

It feels like IRNDs are as much art, as Science, to manufacture from what I've seen. 

[peterbeeh]

Could you post frame grabs? I'm sure many of us would be thrilled to see the results/effects.

[adrianjebef]

I've used all sorts of ND and IRND's with the ALEXA under many lighting situations. My conclusion is that every setup is different as is every ND or IRND. Even the same filter from the same manufacturer will be slightly different if compared. I recommend using IRND's at N9 and above (N1.2, N1.5, etc.) And I usually find any ND will add a bit of green to the picture. I like to dial this out with the ALEXA's color-temp +/- Color Correction. I find it best to adjust by eye using a proper monitor. It's not unusual for me to have wildly different CC's on two cameras if they've got different NDs.

[Bob]

I agree with Greg that both manufacturers probably have to struggle between art and science. There are indeed slight differences, but also I can't really say that one manafacturer is better than the other. When filming in a ProRes family codec In a multi-camera setup I would advise to use 'equal' filter sets from one brand, and I would still white balance on a good reference grey chart in combination with a proper monitor (like Adrian wrote). Depending on the situation, I would be careful with heavy filtering (especially when good skin tone reproduction is important). I would also set the actual F-Stop on the iris of the lens after setting the white balance, since this order revealed during testing. I still try to understand the latter better on a technical level, and hoping to find an explanation for it. Thanks for sharing your knowledge. 

Best Regards,

Vincent