What's with the blue channel?

[craigjkharris]

Just finished production on a feature and was very confused with the blue channel off both Epics.
When measuring the footage on a waveform parade, the Blue channel was blurry, spread wide and did not sit on the same plain as the Red or Green channels. Made it impossible to balance black levels using a parade. This was the case in RedCine, Scratch and Resolve.


Craig

 

[Mike P.]

Did you set the white balance?

Traditionally the Blue Channel is the weakest with CMOS sensors (something about wavelength). Using warmer light sources, like tungsten, exacerbates it further. Making the blue channel balance out is easy though (it's raw), however I usually find hat it does some weird pixel fringing... Almost as if the blue channel has been boosted way too much to get it in line.

As with all things, if you're concerned it's an issue with the camera (not sure how, seeing as you had two and they were exhibiting the same thing), you could always post a .r3d and people will tell you right away if what you're seeing is normal or not.

 

[Joseph Winchester]

Several shooters recommended to me to shoot with a 1/4 CTB filter in front of the lens to help alleviate this issue. I just got my filter so I've yet to test it.

This is the one I got in 1/4 CTB (it defaults to 1/8):

http://www.filmtools.com/schneider-optics-1-8-ctb-4x5-650.html

 

[craigjkharris]

Thanks Mike.
I think it is probably the 1/4 Black Frost filter we used on the show.
Screen grab of Parade and image using Black Frost and one of the only shots we didn't use the filter on. You'll see the difference in the blue channel's black and mid-tones.

"With Black Frost Filter"

"No Filter"

Craig

Did you set the white balance?

Traditionally the Blue Channel is the weakest with CMOS sensors (something about wavelength). Using warmer light sources, like tungsten, exacerbates it further. Making the blue channel balance out is easy though (it's raw), however I usually find hat it does some weird pixel fringing... Almost as if the blue channel has been boosted way too much to get it in line.

As with all things, if you're concerned it's an issue with the camera (not sure how, seeing as you had two and they were exhibiting the same thing), you could always post a .r3d and people will tell you right away if what you're seeing is normal or not.

 

[Mike P.]

Was that last shot (the non-Black Frost shot) in daylight? I've seen similar things in my histogram but never really paid much attention because I generally love warmer tones anyway... I find it very interesting (in a bad way) that the blue channel isn't just lower, but actually looks to be softer/resolving less...

Joseph, I hadn't actually considered using a 1/4 CTB on the lens, but it makes sense... My concern is that, even when the blue channel is lined up properly, it lacks resolution/sharpness (which Craig's histograms seem to show by the inherent blurriness) compared to the red and green channels. It'd be nice if someone had hardware scopes to double check that with. Similarly, Joe, please post the results of your 1/4 CTB findings.; I'd love to see how it affects the image. Similarly, I'd love to see some Alexa/F65/Epic RGB parades of the same images to compare...

 

[Timur Civan]

I use half CTB.


RED cameras dont like tungsten. If i need a scene to look warm, i often shoot 5600k with half CTO on HMI's or Half blue on Tungsten. You have to shoot for your medium.

 

[Sidney L. Plaut]

all cmos cameras to my knowledge have daylight based sensors.

 

[Daniel Soderberg]

Timur,

Instead of gelling the lights, wouldn't it be easier to throw an 80A (or B) filter in front of the lens?

I use half CTB.


RED cameras dont like tungsten. If i need a scene to look warm, i often shoot 5600k with half CTO on HMI's or Half blue on Tungsten. You have to shoot for your medium.

 

[craigjkharris]

So as long as tungstun is used with Epics, you will never be able to read black levels without additional filters?

 

[Sidney L. Plaut]

I wouldnt stress about this... just know that you are underexposing your blue channel by 2 stops - when you WB the camera to 3200k. which can generate some noise in the blue channel.

you could offset this by exposing for asa/iso 320 if your are doing green screen/ blue screen.

a 80A takes 2 stops of light, which is a lot in my opinion. a 80D takes much less light and will get you down about 4000k (If I remember correctly).

- this goes for Epic and Alexa.

 

[Timur Civan]

Timur,

Instead of gelling the lights, wouldn't it be easier to throw an 80A (or B) filter in front of the lens?

I prefer it on lights simply because i can vary the strengths on each unit to taste. Really, it can still be done with a filter as well, 6 of one half dozen of the other. I just like keeping less glass infront of the lens.

 

[Marc Wielage]

Blue was (and I guess still is) the noisest color record of 35mm negative as well, I believe because the Yellow layer (opposite of Blue) was the top emulsion layer. If the film was overexposed, the blues seemed to suffer the most. And if you underexposed the film, or the film had to be pushed in the lab, the blue channel had tons of noise, especially in the blacks.

Note that quite a few commercial grain reduction processes will allow selectively controlling what they do by color channel. It's not a big deal to apply more grain reduction to Blue-only (or luminance-only, or whatever) and try to suppress it as much as possible.

 

[craigjkharris]

Can someone from RED comment on this?
It's a pretty serious issue for any colourist who need to balance their shots.