5600k vs. 3200k

[Bryce Lansing]

Sorry if this has been covered before.. but I couldn't find a specific thread that had the answer.

I've read that the Red performs best with daylight. If I wanted warm skin tones (but not drastically warm), would it be better to balance the camera to 5600k Daylight, and CTB tungsten lights to bring them up to about 4600k, than to balance the camera to 3200k, and CTO them to bring them down to 2200k?

I want to use purple accents, and I like the way purple gel effects the light on daylight film. I'd like to get the purple I want, while keeping warm skin tones.

 

[Joel Kaye]

I want to use purple accents, and I like the way purple gel effects the light on daylight film. I'd like to get the purple I want, while keeping warm skin tones.

Daylight film and what type of lighting? You should be able to do the same thing with RED and get a similar look.

I find the camera is less noisy under daylight. Shoot daylight and set the white balance at 7k-10K and see if you're getting what you want. You can use the latest alpha/beta builds with the latest Red Alert to get less noise under tungsten if you decide to light that way.

 

[Bryce Lansing]

Actually, after doing tests with some gels and my HVX, I was wrong. The colors are how I want them when the camera is balanced to 3200k. Will the Red be very grainy if I balanced it to 3200, even if I had some daylight in there somewhere?

 

[Jake Wilganowski]

It's not the white balance of the camera that's going to affect "graininess" more accurately called "noise". Its the color temperature of the light fixtures your using. The white balance on the camera means nothing more than getting a creating look on your on set monitor. The red sensor has less noise when exposed properly using daylight balanced lights rather than tungsten, regardless of where you set your white balance in camera.

 

[Michael Schmitt]

Interesting. Think it will be the same for Scarlet? All my lights are 3200.

 

[Jake Wilganowski]

There is no reason you cant get great images using 3200k lights, just don't underexpose. You could also add a blue filter to the lens if you have enough light. BUT, with the new build 20 this is way less of an issue. I'm sure the scarlets and epics will be even better. Don't be scared to shoot tungsten, just shoot daylight if you can....

 

[Michael Schmitt]

I'm not a-scared.

 

[jake blackstone]

Sorry if this has been covered before.. but I couldn't find a specific thread that had the answer.

I've read that the Red performs best with daylight. If I wanted warm skin tones (but not drastically warm), would it be better to balance the camera to 5600k Daylight, and CTB tungsten lights to bring them up to about 4600k, than to balance the camera to 3200k, and CTO them to bring them down to 2200k?

I want to use purple accents, and I like the way purple gel effects the light on daylight film. I'd like to get the purple I want, while keeping warm skin tones.

it's called color grading. Why people keep insisting on doing all this in camera escapes me...

 

[Alex Carr]

Because Colorists can 'never' create the Character of Light that Cinematographer's spend their entire Careers perfecting.

 

[Dan Hudgins]

In camera or not...

In camera or not...

Because Colorists can 'never' create the Character of Light that Cinematographer's spend their entire Careers perfecting.

I would recommend doing tests.

Since RED ONE is not a "true RAW" camera, that is you cannot get the full sensor data out (as far as I have been able to tell), because the de-mosaic software does white balance clipping, this is needed because the anti-OLPF sharpen requires that the data be white clipped before it is sharpened otherwise you get off color rings on the highlights.

If it was true RAW then you could use filters on the camera and re-adjust the clip points to compensate later in CC adjustments to the de-mosaic.

Because you do not have full control over the de-mosaic settings (I guess because they don't want to to goof with them and get odd results) you should not filter too far off center since the built in white balance is not setup for off center "white", it is set up for a line that follows the K values from orange to blueish.

It would be good if you could use a pink filter to compensate for the Bayer filter's bias to green and get extra dynamic range, but without the adjustments using filters in the camera can reduce the dynamic range and so limit the adjustment range in CC later.

Its better maybe to expose the shot so that white subjects are exposed close to the cameras white point so the de-mosaic software can render those subjects as white. You can use gels on the lights to add color accents, but if you want the whole shot tinted you can do that with CC later rather than add noise and clipping in your R3D files.

Color negative had a longer range of stops and softer clipping at both ends, so if you filter off to get a cross-over look from the un-equal contrast in the three layers you have something to pull back later.

Digital does not have a soft foot and sholder curve, no S curve in the camera's "RAW" data, so you cannot introduce "cross-over" with filters in the camera, what you do-do is to add noise in the dark areas and clipping in the highlights with color filters on the camera or general lighting.

My CC has control over RGB clipping, gammas, and curves both before and after matrix, so you can maybe simulate the multi-generational effects of the color masking layers in film stocks (net, master pos, dupe neg) better than using a single LUT, since in film you grade, or should, each generation to keep the subject in the right parts of the S curve in each generation. With my CC you can also introduce film like cross-over if you want that aspect of "the film look" and get un-equal sharpness in R,G, and B like film since film has un-equal grain and sharpness in the color layers, as well as un-equal masking between them, and the masking is effected by the curve of each layer, etc. For Digital "negatives" i.e. "RAW" sensor data, it is best to get the best dynamic range for R,G, and B since the sensor is linear, not S curve like film. When you overexpose one color it will hard clip and that looks very "video" and not at all like film-to-film printing. Using an 80 series filter under low K lights reduces the over-exposure of the red or the under-exposure of the blue, which you don't see because the de-mosaic outputs white balanced images, what you do see is reduced dynamic range, noise in the dark areas and clipped highlights the more of sensor K you get. RED ONE records even gain on the RGB, so does not compensate for low K values before the ADC, so its best to have gray give equal ADC counts.

 

[jake blackstone]

Because Colorists can 'never' create the Character of Light that Cinematographer's spend their entire Careers perfecting.

Alex. You should mention this to the colorist, who's grading your next project I'm sure, he'll love working with you... It seems, in your world all colorists just got off the bus in the big city and they could never spent their entire careers perfecting their craft. How could they possibly even start to understand "the Character of Light that Cinematographer's spend their entire Careers perfecting"? They are all just boobs and they should be kept away from all images as far as possible, before they start to reinterpret "the Character of Light that Cinematographer's spend their entire Careers perfecting". I'm sure, you would rather grade everything yourself, as it seems, you know better and have a great deal of talent and an eye...

 

[Alex Carr]

You took my comment in the exact opposite direction it was intended. Cinematographers create that character of light on the Set. They are then able to Manipulate it with the Assistance of the Colorist in The Color Grading Suite. Colorists are not on Set moving the Lights with the Cinematographer, unless they are in your world

 

[David Mullen ASC]

it's called color grading. Why people keep insisting on doing all this in camera escapes me...

Because what colors you use on set with lighting can have an affect on the noise level of the RED image, so you have to consider the best way to shoot in order to get the best image in post.

For example, what if he said he wanted a blue moonlight look. A colorist might say "just shoot in white light and we'll add the blue in post". But what if the DP then shot the scene in "white" tungsten lighting using a 3200K base on the Red? That's already a bit noisy in the blue channel and pushing the image bluer in post would increase the noise even further. And then the DP would be telling the colorist "but you told me to just shoot in white light and we'd create the blue color in post!"

This comes from the fact that Red is optimized for daylight balance, which a colorist might not know. Because of this, you have to think about what color temp to use for your lights and your camera settings to get the color effect you want with the minimal amount of noise. You can't just leave it to post to figure out.

A color effect that involves major pushing of color channels in post is going to generate more noise in the image than if that final color was created on the set and the amount of signal pushing in post was kept to a minimum. Of course, this assumes that no one changes their mind about the look because then the opposite problem happens, the colorist has to do a major amount of signal pushing to cancel out the look created in camera.

But let's assume that you know the effect you want to achieve. You have to figure out as a DP what parts should be done in camera and what should be left to post to get the optimal effect. And there are a number of good reasons technically do get it right in camera -- the old adage, garbage in, garbage out still holds true for digital color-correction.

If "mrbrycel" wants a half-orange effect, for example, he's better off working with a 5600K base setting for the camera -- in which case, a half-blue gel on a tungsten lamp would give him that half-orange look on the files. Or he could just shoot with HMI's at 5600K, ungelled, and add the warmth in post, just that if he's having trouble affording HMI's, he can use tungsten lights and add half-blue correction, get the warm affect he wants without losing two-stops of light output by trying to correct tungsten lamps to 5600K since it's not necessary if he wants a warm look anyway. All of which is basic DP stuff.

The colorist can contribute mightily to the image, but ultimately, they still are serving the cinematographer's vision, just as the lab timers do, or the gaffer and key grip, operators, etc. If the cinematographer knows what he is doing and wants to create a look in camera, I don't see why the colorist should get bent out of shape over that decision.

 

[Robbie Anderson]

Color grading is important, so is lighting on set.

Color grading is important, so is lighting on set.

it's called color grading. Why people keep insisting on doing all this in camera escapes me...

As a cinematographer I believe color grading can mean the difference between acceptable work and excellent work.

That said, their are lighting techniques using light sources of different colors that work best if done in front of the camera. This is particularly true when there are motivated light sources from different sides of the subject. The movie frame is two-dimensional, but lighting is three-dimensional.

An example might be an actor inside a car at night, illuminated from below by a bluish dashboard light and from outside the car window from a yellowish mercury vapor streetlight. Proper color grading will make this kind of mix of colors more successful and more able to match with other shots in the scene. This is equally true whether the light sources used are the actual dashboard light and actual streetlamp or whether the light sources are movie lights gelled to the desired color.

A considerate cinematographer will, where possible, shoot a color chart under white light accompanying the scene, to communicate to the color grader that the non-standard color of the scene is intentional. That way the color grader does not have to waste time compensating for what looks like a lighting error.

All our roles matter here. Let's respect each others' work.

 

[Bryce Lansing]

Thanks everyone for the valuable info and tips!

Here is my problem (if it is a problem). I figure out I have to shoot at Tungsten to get the colors I want with the gels I'm using (Cyan and Purple). The only way I could shoot 5600k, is if CTB my tungsten lights up to 5600k, then put my gels over that. I'd probably end up losing a good 4 stops of light with that method.

Now if I shoot at 3200k, CTO my keylight down to about 2800k, but use my Cyan and Purple lights in the background and as backlights/cheek kicks, will the higher color temperature in the Cyan work as daylight to the Red's sensor?

 

[David Mullen ASC]

Thanks everyone for the valuable info and tips!

Here is my problem (if it is a problem). I figure out I have to shoot at Tungsten to get the colors I want with the gels I'm using (Cyan and Purple). The only way I could shoot 5600k, is if CTB my tungsten lights up to 5600k, then put my gels over that. I'd probably end up losing a good 4 stops of light with that method.

Now if I shoot at 3200k, CTO my keylight down to about 2800k, but use my Cyan and Purple lights in the background and as backlights/cheek kicks, will the higher color temperature in the Cyan work as daylight to the Red's sensor?

I'm not sure you understand the problem. The problem is what you are setting as your base "white" or neutral color temp -- is it 3200K or 5600K or something in between?

Cyan is blue + green. If you want your cyan to remain looking cyan, you can't use the blue part of it to raise the color temp of a tungsten light to reduce noise... because that means your cyan just becomes green instead of cyan. You've cancelled out, used up, the blue part unless you keep 3200K as the white point, if you put that cyan gel on a tungsten light.

You either need to use HMI's and/or other daylight sources (LED's, Kinos, etc.), or live with the noise level of a 3200K basic setting, or at least add an additional 1/2 CTB to all your tungsten lights and set your white point a little higher than 3200K, like 4800K. Then at least you've reduced some of the noise.

But perhaps you won't have a noise problem with 3200K as your base setting. Have you tested it?

And there's no reason to add 1/4 CTO to reduce the color temp of a tungsten lamp to 2800K unless you need to do that to match it to another uncorrectable 2800K light (like a household practical in the shot.) You can get the same effect by using an ungelled 3200K tungsten lamp and setting the color temp of the camera to something higher like 3800K. Then your tungsten lamp looks slightly warmed up and you haven't quite pushed the blue channel as hard because this native daylight-balanced camera is only trying to reach 3800K instead of 3200K.

 

[Dan Hudgins]

Resolution vs. light's color

Resolution vs. light's color

Something everyone using the RED ONE or other Bayer sensor cameras should keep in mind is that the resolution in the imags varies with the color of the light.

The "sharpness" of the so called 4K image comes from the luma part of the brightness data, and that is best when the ADC counts from the three colors are equal.

When the ADC counts are mostly red or blue you get less resolution contrast in the final image and more noise.

It is not just an issue of trying to re-balance to white or gray, any object colored redish or bluish will have less contrast in the luma when the output is saturated to film like levels.

If you shoot at 2800K your images will be less sharp (lower average resolution) and have more noise and less dynamic range.

Yes, a camera with daylight balance works aganst shooting with tungsten lights which is why color negative films were tungsten balance.

But if you are shooting with a daylight balance sensor, you need to think of it as daylight balance film, which you would never shoot at 2800K and expect to get good results with.

Also you only have 11+ stops of dynamic range with daylight balance lights, when you shoot a lower K values the dynamic range of the non DRX part of the image will be less. If you need 4 stops of blue filter to get balanced, then your dynamic range would be 4 stops less or 7 stops rather than 11 stops.

Whenever possable you should figure on requiring more lights when using a daylight balance sensor if you want to get "4K" resolution out of it since the net resolution goes down with color shift off center.

Also you need to think about the added impact of the REDCODE artifacts. If your end use is SD, then shooting at 4K will let you average most of those out, but if you want to do a 2K filmout that is clean, then you should keep all colors exposed above both sensor noise and REDCODE induced artifacts that tend to show up more in underexposed images do to the noise using up bandwidth.

If you need to rate the camera at EI 50 for use under low K light with enough filter to get even ADC counts than that is what you need to do, otherwise it is another "teacher in space", wishfull thinking does not change the laws of physics.

One thing you can do with tungsten lights is to boost the voltage, we use "colortrans" for almost everything we shoot to get two stops more blue light, if you hook up relays and a foot switch you can boost just when filming to save the lights, that way you get more usable speed for the same number of lamps and it works on Practical lights as well. If you do not have any "colortran" light boosters you can use VariAC into 120 to 220 xformers etc.

 

[David Mullen ASC]

If you have to use tungsten lamps (perhaps they are already part of the location), then at least you can use a blue filter on the camera to shift the overall image, gels and all, up from a 3200K base. Any amount of blue camera filtering is better than none in terms of improving the noise.

But like I said, perhaps you're OK with the noise level at 3200K. Not everyone has a problem with it afterall, this may be a non-issue for this project.

 

[Alex Carr]

All our roles matter here. Let's respect each others' work.

Thats exactly what I meant.

 

[Bryce Lansing]

Sorry for my confusion, but I think I understand now.

I think I'm just going to shoot with Tungsten lamps and a blue filter. Now according to my Color Meter, I'd have to shoot with an 80A filter to convert the tungsten light to white (at 5600k). That takes about 2 stops, could I just the camera white balance to 4000k, and then use an 80C filter instead?