Shooting with 80A and 80B Filter on Red

[Vigen Vartanov]

Good day . I have basic question about 80A nad B filters with RED.

As i know red is native 5000K camera. Question is : If i am shooting with Tungsten Light is it better to put 80 serial filter and leave balance on 5600 . , or result will be same if i will change balance on my camera .
From my test shooting with Tungsten light with 3200 balance is more noisy .
I have no 80 serial filter and can not make tests , and that's why i am looking for your experience . I am happy with my Daylight lighting , but in some shoots i want to use Tungsten light.
Please share your experiences.
Thanks.

[Daniel Browning]

[I rewrote some of your post for clarity:]


If I am shooting with tungsten light with an 80A filter, is it better to leave the white balance on 5600K?


No. It's the same. The white balance setting is just metadata, it doesn't affect what's recorded to REDCODE. I suggest that you set white balance according to the look you desire to obtain.

[Harry Capota]

Well, It's metadata but the White Balance definetly affects the way you interprete your Raw Data.
You get best results when working wih your footage in RedCine, Red Alert etc above 5000K. So if you're using tungsten I recommend using A filter (80A, 80B) which gets you around 5000. Furthermore if you are looking for a blueish look, always try to obtain it by geling your lights or using a blueish filter instead of going in your software down with your white balance.

cheers

[Pietro Impagliazzo]

The best option is the one that makes you lose less light.

I remember someone posting that a light 80 + light CTBs was a good way to approach this.

[David Wyatt]

So if you're using tungsten I recommend using A filter (80A, 80B) which gets you around 5000.

80A & 80B get you to 5600˚K and 5400˚K respectively. 80C on the other hand gets you to 4900˚K which is closest to Red's native 5000˚K (it's also only one stop of light loss, which isn't too bad).

[michael zaletel]

I think you lose too much light cutting down Red and Green with 80A&B under tungsten. I've read best choice is 80D. Verification anyone?

-michael zaletel
(shooter)

[albert rudnicki]

Every application cries for different tools.
Test it before using it.
80C and D get you close to the Red native.
Filter cuts/changes light that sensor feels, and it's important to understand it. It can save or fuck your ass.
Using 80s is a good idea in tungsten light.
Saying that, I am against filters:)

[Dan Hudgins]

I am against filters:)

Part of the "Video look" is the use of sensor gains to get auto white balance, what you are doing is reducing the dynamic range of two of the three sensors.

There is no Magic in sensor over film that frees you from using optical filters to get true color balance of the light.

If you want the best results from all 12bits the sensor can put out you need to filter the light going onto the sensor to be the right color to match the offsets of the Bayer filter.

It is probably NOT true that any normal K value 5000 or otherwise gives the best color light for the REDONE sensor as most bayer filters overexpose the green and have the red/blue off balance. You might balance the red/blue at 5000K or some value, but green will not clip at the same time. To get all three colors to clip at the same time you need to use a pink or magenta filter plus the 80 series or FLD etc to balance the red/blue (the pink filter adjust the green to match the red/blue levels).

You cannot just balance your REDONE sensor by yourself because the de-moasic software is adjusted for an un-balanced sensor output, to get better results and full dynamic range the de-mosaic software's black and white clip points in each primary color would need to be adjusted for the K value and pink filter used.

Not using optical filters for balance increases image noise and clips the highlights more (when DRX is off).

[Nir Shelter]

Nothing wrong with using filters if you know what look you are shooting for. From my experience though for best result use CTB and CTOs to correct your lights. Even the fittings in the ceiling if shooting in doors rather than filters on the camera as you do lose a stop.

[Chris Kenny]

If you have enough light, filtering to as close to 5000K as you can get will give you a better image. Assuming, of course, that you want a neutral image, rather than something warmer.

If you don't have enough light, then filtering from 3200 to 5000 is just going to mean you're underexposing your red and green channels as well as your blue channel, which is obviously going to look substantially worse.

[Dan Hudgins]

If you don't have enough light, then filtering from 3200 to 5000 is just going to mean you're underexposing your red and green channels as well as your blue channel, which is obviously going to look substantially worse.

Many "practical" tungsten lamps burn at low K values like 2600 to 2800, that reduces the amount on blue they put out and makes the blue noise problem much worse.

Even with 3200K lamps if you boost them to 3800K you get much more blue light out of them.

We use "ColorTran" units for our shooting, you get about a stop more red and about two stops more blue light depening on how much you boost and if you use a variac on the input to adjust the boost past 3400K.

Boosting the lamps to 3800K then filtering to 5000K or so gets you less noise with the same number of fixtures without a large net filter factor loss over non-boosted values.

[Joe G.]

To further complicate matters... I always remember David Mullen's suggestion of using tungsten on the faces, but a daylight or blue gelled background fill, where the noise will show up. This can get you the best of both worlds.

I intend to try this.

[Vigen Vartanov]

Please share relult.


To further complicate matters... I always remember David Mullen's suggestion of using tungsten on the faces, but a daylight or blue gelled background fill, where the noise will show up. This can get you the best of both worlds.

I intend to try this.

[Guidofilippi]

There was a thread not so long ago about the use of the new Schneider CTB filters, which come, I think, in Full, 1/2, 1/4 and 1/8 versions. Film Tools offers them at a very good price. I ordered one of them but my order was rejected because I have a South American credit card...
Guido

[Dan Hudgins]

tungsten on the faces, but a daylight or blue gelled background fill

I shot some tests using florescent fill on the background and tungsten on the face a while back, the result was green shadows under the chin of the subject in folds of the wardrobe. There was too much bounce from the background lighting.

If you use mixed illumination you need to flag off the off K light to keep the shadows from filling with the wrong color light, you cannot fix these off color shadows very well in post since doing so puts crossover into the image.

[Joe G.]

"I shot some tests using florescent fill on the background and tungsten on the face"

Are you talking the crappy florescents with a green spike? Or daylight balanced high end lights?

[Michael Lindsay]

80A & 80B get you to 5600˚K and 5400˚K respectively. 80C on the other hand gets you to 4900˚K which is closest to Red's native 5000˚K (it's also only one stop of light loss, which isn't too bad).

I believe:

a 80c takes 3200k to 4348k

.a 80c has a given mired shift but its kelvin shift depends on the starting temp.

regards

Michael L

PS having said that I do agree the 80c is a good compromise.

[Evin Grant]

I helped develop the CTB series with Schneider specifically for this issue with the Red. The 1/4 CTB is really the best option, only 1/2 stop loss but allows you to shoot at 4200K, only 800 off of native. The 1/2 CTB is a 2/3 stop loss and corrects exactly to 5000K.

[Shawn Nelson]

I assume that since I'd be using nearly always indoors under lower light and without NDs, that the IR isn't necessary whilst using this?

[Shawn Nelson]

I helped develop the CTB series with Schneider specifically for this issue with the Red. The 1/4 CTB is really the best option, only 1/2 stop loss but allows you to shoot at 4200K, only 800 off of native. The 1/2 CTB is a 2/3 stop loss and corrects exactly to 5000K.

I haven't bought one yet, but from asking around it seems 1/2 is the most common, since it does the full correction, why do you feel 1/4? Just because a good tradeoff in lightlost/color shift?

[Mike Prevette]

I use a 80C a lot, it's a single stop and it gets the camera very close to balanced.

[david_winters]

I have 80C and 80D. So far I've only used the 80C on jobs. Tomorrow I'm going to shoot a few simple tests with the 80C & 80D prepping for a job. What is the formula to calculate the K shift, I assume it is not linear?

2900k w 80C = ?
3200k w 80C = ?
2900k w 80D = ?
3200k w 80D = ?

[Esmaile Neissi]

I'm not sure but i think thats the formula:
T2=1/(1/T1+F)
Which is:
T1=source light color temperature (MK)
T2=resulting color temperature
F =Filter shift value in (MK-1)
please correct if i'm rung

[Dan Hudgins]

1) Any mixed light on the subject will make off color shadows, by the degree of difference. If you need to shoot with mixed lights, flag off one from the other so that they do not mix on the subjects, unless you want "bi-color" effects lighting.

2) Boosting the voltage to tungsten lamps reduces the ratio of blue to IR, whereas using a blue filter can increase the IR to green ratio. Its better to boost the lights since the IR goes up a little and Blue goes up very much.

3) Every stop of unbalance reduces the dynamic range by one bit or cuts it in half. If the sensor is 12bits balanced, the de-mosaic must cut that to 10 bits if the blue is 2 stops under-expsoed relative to the green or red. If the blue is 3 stops under-exposed relative to the green or red you are only getting 9 bits of dynamic range, and so on. To get the full 12bits you paid for the light going to the sensor must be balanced (with pink and blue filters as needed, or yellowish for twilight skylight, etc.).

4) You can add a color tint to the 48bpp (16bitx3) frames later by adjusting the red/blue ratio of gammas to get any warm or cool tint without the noise problems you have if you shoot off balance in the camera. The reason for this is that you have more bits in the color correction software, and the digital images are not subject to noise like the analog parts of the sensor. (Do not rotate the color vectors using Hue (sometimes mislabled Tint) in your software, use the red/blue gammas to keep white and black where they should be and to keep the primary colors in their right places.)

[David Wyatt]

I believe:

a 80c takes 3200k to 4348k

.a 80c has a given mired shift but its kelvin shift depends on the starting temp.

regards

Michael L

PS having said that I do agree the 80c is a good compromise.

Not 4329˚K???...:thumbsup:

Michael is [annoyingly, as ever :cuss:...] correct.....

I'm glad I'm not the only one getting it wrong though :rolleyes::

http://www.reduser.net/forum/showthread.php?t=24355&page=2

I think maybe part of the confusion is that a lot of the literature will tell you that an 80C will correct 3800˚K light to 5500˚K (none of these figures are particularly helpful in the real word though!!)...but if you start with 3200˚K light you won't end up with 4900˚K light but with more like 4329˚K (or maybe 4348˚K :thumbsup:), which is where the dark arts of Mired shifts comes in: if I understand correctly 3200˚K tungsten light has a Mired value of 312 and the Red sensor (5000˚K) has a value of 200 Mireds, so to correct tungsten to the Red you need a Mired shift of -112 (which conveniently is what the 80B offers...but with a rather painful 1 & 2/3 stop loss).

Alternatively an 80C has a Mired Shift of -81 which should get you more than half way there with only a one stop light loss (the rest you can safely colour correct in post without too many problems).

[Harry Clark]

80C is a mired shift of -81.
80D is a mired shift of -56.
So,
2900K + 80C= 3790K ish
2900K + 80D= 3460K ish
3200K + 80C= 4320K ish
3200K + 80D= 3830K ish
I think... using a simple Excel calculator (Roscoe has one online; so does AV Net) Hope I'm not adding noise or spreading misinformation.
Cheers,
Harry

[david_winters]

Shot a white card test on Friday. 80D looks to be getting me up to 4700k, much higher than expected. Light source is four 1k open face Micky Moles through a 6x6 1/2 gridcloth.

The bottom of the white card has green spill from the floor. Based on spot meter tests the 80D eats a little over 1/2 stop, 80C a bit more than 1 stop.

http://farm4.static.flickr.com/3559/3500577773_7e445395cf_b.jpg

[Harry Clark]

Interesting.
I only own the 80D, mainly because it was only 2/3 stop loss and "felt" right on some quick tests a year ago. By the numbers, it should not correct quite so much but I felt that it was "close enough". David's test seems to indicate that it is indeed a good choice.
Cheers,
Harry

[Marco Graziaplena]

to david
impressive...
also because you might consider that gridcloth has warmed up a tiny bit your 1k. Did you use a color meter to know what was the effective color temperature during the shoot?

P.S. I think you used the "Pick WB" function on RA to determine that 4700K. where did exactly measure it?

[david_winters]

Hello Marco,

To locate proper WB I adjusted the "Kelvin" slider in Red Alert until the peak "red, green, blue" in the histogram matched. The example shot is of a white card, so the peaks of the histogram represent "white."

I did not have a color meter with me on set. I'm sure the lamps were a but warmer than 3200k, however I'm not sure how much color bias is in the lite gridcloth, anyone know?

Anyone want to buy a basically new 80C Panavision size filter? :)