PDA

View Full Version : Dragon White Balance



Rodrigo Prata
09-04-2014, 05:28 AM
I`m still shooting with MX, and something that is common knowledge by know is that the MX sensor performs better with daylight sources then with tungsten. With tungsten you almost always get blue noise in the shadows at 800ISO, that seems to happen because tungsten light has less blue light in it`s spectrum. However, that only happens when you shoot with tungsten lights and try to achieve a neutral look by developing the RAW file at 3200K. That's why many people advocate using a blue filters (80 or 82 series) to lift the blue channel.

I have shot a lot with tungsten light, but I often develop at much warmer settings, frequently developing between 4000K - 5000K. since most of the projects I`ve been involved, called for a warmer look (or maybe that`s just my style), and I noticed that you didn`t need to rate the sensor lower then 800, or use a blue filter, as long as you developed at something closer to 5000K, because then, you are not trying to pull information from an underexposed blue channel to achieve a balanced look.

What seems to be happening at Dragon with the OLPF V2 is the exact opposite. With R3D samples people have provided, I noticed that when developing at close to 3000K, the red noise present in the image would go away. When I first saw that, i said to myself "i bet that the new OLPF is much bluer then the V1". This was later confirmed. So the red noise seems to come from an under exposed RED channel.

So, is not the case for we to forget the whole Daylight sensor mentality when it comes to Dragon OLPF V2 and start treating it as it was tungsten balanced film stock? Maybe people experiencing the red noise should try using a warmer filter when shooting daylight. I know that we have D.E.B now, but some have reported that it is not a good ideia to just "turn it on" since it may affect the rest of the image in an undesired way sometimes.

Has anyone tried using an 85 filter with the Dragon sensor? what was the result? I don`t have access to a Dragon yet, so I can not test it myself.

David Battistella
09-04-2014, 05:40 AM
Colors are much more accurate in tungsten light. It's true. I used to use a 1/8 blue on MX in pure tungsten, but now there seems to be no need for that with Dragon. I would say that the daylight performance is still excellent as all CMOS sensors have a native point around 5000K, but the combination of new COlor science on Dragon actually does leave you with an accurate tungsten balanced stock if you want to go that way.

I have not tried the 85, but I feel there is no need as the colours are rich when treated as Daylight .

Battistella

Rodrigo Prata
09-04-2014, 08:17 AM
I`m sure the colors on Dragon have excellent response under daylight balanced light, the same way that colors have an very good response under tungsten light on MX (and noise free at 320ISO).

My point is that the red channel might actually being underexposed with Daylight light (since the OLPF v2 is so much bluer), and that`s where the red noise is coming from in the first place. A 85 filter would balance the exposure on all color channels, the same way a 80 filter would do on MX sensor.

I guess that my main point is that maybe we should stop thinking of Dragon (with the OLPF v2) as a Daylight camera.

Phil Holland
09-04-2014, 09:08 AM
The sensor seems to favor a division between Tungsten and Daylight (4100-4500K). Meaning it tends to play nicest in the middle, but more or less with Dragon shooting in Tungsten or Daylight is not a concern.

Paul Russell
09-04-2014, 09:10 AM
When shooting corporate talking heads I like to use 5600K fluoros to light the background with a nice big soft tungsten on the talent in front. Lovely warm skin tones that stand out. And here in Asia, where a lot of people are brown, this works very nicely.

Rodrigo Prata
09-04-2014, 09:20 AM
The sensor seems to favor a division between Tungsten and Daylight (4100-4500K). Meaning it tends to play nicest in the middle, but more or less with Dragon shooting in Tungsten or Daylight is not a concern.


Phil, I remember you posted not so longa ago in a thread where somebody posted a lowlight comparison between Dragon (OLPF V2) and MX. The scenario was a kitchen of the OP's house.

You posted a screenshot of the same scene developed at a lower white balanced (below 3200K if I`m not mistake) that showed Dragon had less noise at this settings and that was the first time I saw that the Lower color temperature you develop the Raw from Dragon (OLPF v2), the lower the Red noise. this seems to indicate that the red channel is somewhat always underexposed in comparison to the other channels... If someone could test this it would be great.

I

Aria Khosravi
09-04-2014, 09:20 AM
Rodrigo, I believe you are right and I noticed the same thing long ago, when R3Ds from the new OLPF first came out. I too am treating the Dragon as a tungsten film stock and using the 85 filter when I can. As David says, the color response is fine regardless of color temperature, but as you say the noise color returns to normal when the light hitting the sensor is in the neighborhood of 3200K.

Unfortunately my matte box only has 2 stages so I am thinking to just ride the ISO lower in daylight since we can safely do so on Dragon without ugly clipping highlights. But if you've got enough stages to put Pola, ND, and 85, AND you're not concerned about reflections/quality loss from 3 pieces of glass in front of the lens... then the 85 is definitely something that should live on your camera under daylight sources. Screw-on filter might be an option too. I similarly have used a blue filter on the MX if forced to shoot tungsten sources (or just gelled the lights to full CTB). Even shooting daylight and developing 3200K for a bluish look on MX was not ideal and resulted in blotchy blue noise sometimes. Always tried to develop at 5000-5600K if at possible.

People who doubt what you say can test for themselves by shooting R3Ds with the same exposure under daylight and again in tungsten. The red colored noise goes away and is more of an even mix of RGB. Even if you flip DEB on, the tungsten source still produces a better result in terms of noise pattern as compared to DEB on with daylight sources. Zeb B posted R3Ds from a comparison between a tungsten and daylight Hexolux unit (http://www.reduser.net/forum/showthread.php?115781-Hexolux-Color-Rendering-Dragon-R3D-s) that serve as a pretty nice comparison to see what you're talking about. Just set the white balances correctly and set gamma at REDlogfilm, then take a look at those shadows.

Jarred himself said that the red channel sensitivity was optically changed in the new OLPF to facilitate better tungsten performance. Can't find the post but it was in one of the huge Dragon Noise threads that eventually had to get closed because it got out of hand with misinformation and RED was already working on DEB.


My 2 cents of course. I would provide my own R3Ds for you guys to look at but I just received my own Dragon today and haven't even fired it up yet, and am busy with other things at the moment.

Michael Lindsay
09-04-2014, 10:01 AM
after some testing we felt to match the MX at 5000 we should develop Dragon at 4300... Basically it does sit a bit lower natively.

Rodrigo Prata
09-04-2014, 11:09 AM
Thanks for you input Aria,

I too believe that the best practice is to set your ISO as low as possible whenever possible, since with Dragon it now has this crazy highlight protection.

But what I also believe, is that much of the frustration with the noise on Dragon, has to do with the fact that people that are used to get clean images on MX, are now getting noise on Dragon and they don`t know why.

And maybe that happens because they still treat it like it is a daylight camera.

I remember a while ago, somebody posted that was going to shoot a green screen project on Dragon (OLPF v2) and was worried about the noise. He asked for advice and I remember a lot of people recommended him to use daylight sources if possible... this now sounds to me, like a not so good advice when it comes to Dragon (OLPFv2). They did that because of their experience with he MX sensor.

Zeb B's R3Ds shows that the tungsten lit scene has less noise then the daylight lit one. However he also states in his post that he did not have a light meter, and compensated exposure by eye, so although they seen to point towards what I`m saying, they are not conclusive proof.

Aria Khosravi
09-04-2014, 12:17 PM
Zeb B's R3Ds shows that the tungsten lit scene has less noise then the daylight lit one. However he also states in his post that he did not have a light meter, and compensated exposure by eye, so although they seen to point towards what I`m saying, they are not conclusive proof.

Good catch. Still, you can set them to identical metadata settings (except WB) and the exposure change is minimal, not enough to skew the results too far. It may not be a perfect A/B but for anyone trying to understand what you're talking about, it's a good start.

I also remember the post you mentioned to Phil where it was MX vs Dragon in a tungsten-lit kitchen, and I too recall that as being the first place where I thought that the Dragon may be tungsten native with the new OLPF. Alas, someone will have to do a scientific comparison and settle this matter once and for all.

Ignacio Aguilar
09-04-2014, 04:32 PM
The sensor seems to favor a division between Tungsten and Daylight (4100-4500K). Meaning it tends to play nicest in the middle, but more or less with Dragon shooting in Tungsten or Daylight is not a concern.

I've been done some testing and found similar results (4100 kelvin as the sweet spot, to be exact).

Brandon J.F.
09-04-2014, 04:41 PM
I've been done some testing and found similar results (4100 kelvin as the sweet spot, to be exact).

Is this true only for the V2 OLPF? Has anyone done any tests like this with the V1 OPLF?

Rodrigo Prata
09-04-2014, 05:44 PM
Is this true only for the V2 OLPF? Has anyone done any tests like this with the V1 OPLF?

As I have stated before, don`t have a dragon yet, so cannot test this but John Marchant
Posted a while ago a picture comparing the different OLPFs and the Dragon V2 was clearly much bluer then the V1. The Dragon V1 was similar to the MX one. So it is logical to think that the V2 filters much more red light then the V1

Jacobo Martinez
09-04-2014, 06:12 PM
But if you use an 85, you would be loosing an aditional 2/3 stop of light, and v2 is already making the camera less sensitive.