Shooting tungsten - 80 vs 3200k

[Lorenzo Levrini]

Hi guys, I have an indoor shoot coming up full of pracs so there's no chance of blueing up the lights. A guy at a rental house told me it is preferable to use an 80 filter instead of balancing to 3200K and fixing in post. Presumably doing all the conversion in post is not only annoying but adds a little noise? I don't like the idea of being forced to keep the 80 in front of the lens at all times, what if I want to remove the matte box for a particular shot? Let me know your thoughts.

[John Sprung]

The rental house guy has it right. You want to make the adjustment at the photon stage rather than in bits, because without the filter you don't have the right balance between the three primary color channels. You'd underexpose the blue, effectively reducing your dynamic range.

 

 

 

 

-- J.S.

[David Mullen ASC]

If you are shooting on the M-X Red One, the noise at 3200K balance is fine, even if it's even better at 5600K. There's no need to use a blue filter unless you are doing a greenscreen shot under tungsten light and need the lowest noise possible.

 

Here are some test frames I shot on the M-X Red One last year at 3200K, no filter. Keep in mind that these have been reduced and highly compressed for the web:

 

400ASA

 

800ASA

 

1600ASA

[Peter Moretti]

I realize that the MX sensor has significantly less noise. But something looks a little bit odd to me in the shadow detail. It's as if the shadows have been "cleaned" and some detail removed. Maybe I'm losing it, as I haven't read anyone else voice this opinion.

 

But to me, it looks like a (somewhat heavy handed) noise reduction is being performed in the chain that is making the footage cleaner.

 

I know the MX is a physicially different sensor. But I--as nutty as it sounds--wouldn't be surprised if the electronics of the sensor are just as noisy as the old sensor. But the new sensor allows for in camera denoising that the old sensor could not work with.

[David Mullen ASC]

I realize that the MX sensor has significantly less noise. But something looks a little bit odd to me in the shadow detail. It's as if the shadows have been "cleaned" and some detail removed. Maybe I'm losing it, as I haven't read anyone else voice this opinion.

 

But to me, it looks like a (somewhat heavy handed) noise reduction is being performed in the chain that is making the footage cleaner.

 

I know the MX is a physicially different sensor. But I--as nutty as it sounds--wouldn't be surprised if the electronics of the sensor are just as noisy as the old sensor. But the new sensor allows for in camera denoising that the old sensor could not work with.

 

I think you're being paranoid, there's no noise reduction happening to a RAW recording in-camera. Sensors don't get built with noise reducers. Don't go by my tests, like I said, they are highly compressed jpegs! But if you are talking about your own tests, I suppose you could be seeing some Redcode compression artifacts, but I doubt it.

 

Besides, you can SEE the noise, it's just very low. The image is not noiseless.

 

I think one of the best things about the M-X sensor images is that the noise looks very "normal" to me even at 3200K / 1600 ASA. Whereas with the old sensor, the blue noise was very large and got mottled by the compression, looked like soft blue oatmeal.

[Sean Lambrecht]

Actually, the latest patent from RED has a good portion of it devoted to discussing several noise reduction techniques, both in-camera and at various stages down the image processing pipeline. It is possible a blue filter could cause denoising to be less aggressive. See the second half of page 42 here.

[David Mullen ASC]

Well, I didn't see anything odd with the noise in the shadows -- if anything, the noise now looks more normal with the M-X sensor camera than it did before.

[Lorenzo Levrini]

Thanks for your replies. I'm afraid we're on the old sensor for this show due to budget reasons, I won't be rating above 400 though. Would you still recommend using the filter?

[David Mullen ASC]

Thanks for your replies. I'm afraid we're on the old sensor for this show due to budget reasons, I won't be rating above 400 though. Would you still recommend using the filter?

 

I just posted this in December in another thread, but this is a shot from "Stay Cool", which I did with the old Red sensor, at 320 ASA, 3200K-ish (may have been 3700K):

 

 

As you can see, the blue noise is not bad for not using any blue filter.

 

If you stick to 320 ASA and below, you should be fine without a blue filter, but if your scene has enough light level, it wouldn't hurt to use one of the lighter blue filters, not necessarily the heavy 80A correction, any amount of blue helps reduce the noise in the blue channel. But it's not a requirement that you use a blue filter. It may get more necessary at higher ASA settings but then you are just chasing your tail because you'd only be going higher because the light level was lower, in which case you don't have the stop for the blue filter.

 

Also keep in mind that most of the extra noise is in the blue channel, if you can leave them image at a higher color temp setting for an orange look, then you won't be boosting the blue channel as much and the noise will be lower.

 

It's easy to just shoot some tests yourself and take some converted Red frames (R3D into TIFF) into Photoshop and look at them.