Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

Quick Tungsten Test

G

Gregor Hagey

Well-known member
Joined
Aug 13, 2007
Messages
155
Reaction score
0
Points
0
Location
Toronto
Website
www.dghagey.com
I just wanted to start a short thread about how to process .r3ds when shooting under Tungsten light. There has been a lot of discussion about the weaknesses of shooting with Tungsten sources, but I think it's more how the RAW files are being processed then Tungsten light itself. This has likely been said before, so excuse me if I'm repeating what others have already posted. I'm using a shot from the feature Shark City to illustrate this.

To see what I'm talking about it's best to view the QT here: http://dghagey.com/Director_of_Photography/Red_Tungsten_Test.html

For the purpose of this example lets say that I want to add a strong blue cast to my shot but I didn't have the time to gel my Tungsten lights with CTB or couldn't get HMI's. Instead I opt to adjust the Kelvin in Red Alert to 2500K. This gives me the heavy blue cast I want, but adds some noise to the shadows. To see the noise look closely in the shadows on the right side of the frame or at the Actor's shirt. If instead I process the .r3d at 5600K the shot is quite warm, but has less noise. If I take both those shots (@1080p ProRes) into Final Cut Pro I can grade the 5600K shot to look like the 2500K shot but without the grain. This is an extreme example. In actual fact I usually set my Kelvin to 4000K for Tungsten lit scenes. If I need to cool the shot down further I'll save it for later during the colour correct. I have also included the shot as it was graded for the film.
 
That was an excellent example.
So this was shot with tungsten light?
What was your K balance on this shot?
What was your process (CC) to blue up the image?
Thanks
 
This is fun.

I am sort of supertempted to put up some "Why RED is not really noisy @ all" examples, I'll see if I get the time. But this is a good (but less brutal than what I had in mind) start.

Cheers! :)
 
Red Guy said:
That was an excellent example.
So this was shot with tungsten light?
What was your K balance on this shot?
What was your process (CC) to blue up the image?
Thanks
Click to expand...

This was shot with 2 Tungsten lights and 1 HMI. The Key light from the left is a 2K Fresnel with 1/4 CTO shining through a 3x3 frame of 216. The Backlight is a 150W HMI Par from the right of the doorway in the hall. The Fill light is a soft silver bounce board. And finally the set light is a 650 Fresnel with 1/4 CTO hung above the doorway in the hallway and is spotted on the back wall. I don't have notes on the exact colour temperatures of the lights, but with the 1/4 CTO on the Tungsten fixtures it had to be below 3200 Kelvins.

To process the .r3d I used the default settings in Red Alert (using RedSpace). The only change was the Kelvin colour temperature (2500K vs. 5600K). I've attached a .jpg of the Red Alert settings.

To CC the .r3d processed at 5600K I used the 3 way color corrector in FCP. I also applied the desaturate lows/highs filter.
 
gunleik said:
This is fun.

I am sort of supertempted to put up some "Why RED is not really noisy @ all" examples, I'll see if I get the time. But this is a good (but less brutal than what I had in mind) start.

Cheers! :)
Click to expand...

I would love to see what you have Gunleik. I have some slow motion material shot in 2K under Tungsten from this show that is a bit noisey for my taste. I will try to get my hands on it to see how good/bad it can turn out.

As a complete aside I had the good fortune to spend a week working in Oslo this past spring. What a beautiful place. The landscape is fantastic in Norway. I hope to get the opportunity to go again. It was the week before the equinox though, so I was feeling a bit out of sorts because the sun never seemed to set (which let us shoot by daylight longer then I'm used to).
 
The conclusion seems to be that all footage should be processed at high colour temperature - am I crazy.

I would love to hear other's opinion on this.

Couple things:

Gunliek - could you post that? I think we would all love to see that.

Gregor - would it be possible to see the .r3d.

Good stuff, folks.
 
I know both the D3 and the RED very well.
On a regular basis I am shooting with both a D3 and RED right beside each other.
The best colorist in the world could not make the red pictures look as good as the D3 pictures.
And the best inferno guy in the world could not make the blacks of the RED images as noiseless as the D3 images.
Not saying that I expect the Red to match the D3. They are different animals.

It is very difficult to get a noisy image on the D3.
On the Red - it is very difficult NOT to get a noisy image in low light.

So to me at least - It is a very easy comparison to make accurately.
 
OK...I don't get it

OK...I don't get it

Could someone explain what is going on in this example please? I see what you are talking about and agree with what's happenning visually, i.e that changing the kelvin does appear to change the noise...but that makes no sense to me at all since the white balance is just metadata. So WHY would changing the color temp in RA change the noise?
 
balance?

balance?

The RED ONE (tm) might be improved if the blue gain (gamma) was adjusted before encoding with REDCODE (tm) since the sensor noise interacts with the compression. Does not seem to make sence unless you know that the compression seems to band more in dark areas, if you make the blue lighter before you compress then the jumps in brightness that the compression compounds the noise by may show up less, maybe the D3 adjusts the K before compression? The RED team decided not to adjust before compression so you compress an unblanced signal, better to compress a balanced signal, so maybe use a 80B and set to 5500K and use more lights.
 
Sir Pancake said:
Could someone explain what is going on in this example please? I see what you are talking about and agree with what's happenning visually, i.e that changing the kelvin does appear to change the noise...but that makes no sense to me at all since the white balance is just metadata. So WHY would changing the color temp in RA change the noise?
Click to expand...

When you develop the files, it's NOT metadata anymore, but baked into the files. So, generally spoken, you should choose when to develop and when to correct. All the K, tint and ASA/exposure switches affect noise and dynamic range in the developed image.
 
more footage

more footage

Here's some 2K footage that gives an even more dramatic view of dealing with noise under Tungsten light. Once again I've treated it as if I wanted to create a blue wash to the shot. I can either do this in Red Alert by lowering the Kelvin to 2500 or leave the Kelvin at 5600 and cool it down later in FCP. I've attached a .jpg of my Red Alert settings. You'll notice that processing your .r3d at a higher Kelvin and cooling down the shot later during colour correction yields much cleaner results. In my opinion this is a lot easier and more cost effective then shooting with an 80 series filter (which means you need to rent higher wattage Tungsten lights to compensate) or lighting interiors to a Daylight balance (which means you have to spend a lot more money on HMIs and/or Kinoflos).

Here is a link to the .mov:
http://dghagey.com/Director_of_Photography/Red_Tungsten_Test_2.html
 
Zakaree said:
this only works if you plan on exporting a baked file tho.. so this doesnt apply if I were to pump out dpx files to grade outside of final cut..
Click to expand...

How you process the .r3d is baked into whatever file you export, be it DPX or ProRes. Only the .r3d contains the RAW data.
 
I get it now, sort of...

I get it now, sort of...

gunleik said:
When you develop the files, it's NOT metadata anymore, but baked into the files. So, generally spoken, you should choose when to develop and when to correct. All the K, tint and ASA/exposure switches affect noise and dynamic range in the developed image.
Click to expand...

OK...I understand about the being bakend in, and I certainly can see how adjusting curves, ASA etc. and then rendering out would/could clearly effect my noise levels. But I guess I don't understand how changing the Kelvin would. Is that having some kind of effect on the gamma, or the curves? Does it boost it or lower it and thus have an effect? Boosting the blues or the reds? Is that it?

I guess I tend to thing of the Kelvin adjustment as a neutral one.
 
Matthew Bennett said:
Looks nice, is this for the vampire movie with S??
Click to expand...

Not yet. This is a clip of Jefferson Brown from a small 16 day feature called Shark City, Directed by Dan Eisen. The previous clip is also from Shark City and shows Jefferson in a scene with Samantha Gutstadt.
 
Sir Pancake said:
OK...I understand about the being bakend in, and I certainly can see how adjusting curves, ASA etc. and then rendering out would/could clearly effect my noise levels. But I guess I don't understand how changing the Kelvin would. Is that having some kind of effect on the gamma, or the curves? Does it boost it or lower it and thus have an effect? Boosting the blues or the reds? Is that it?

I guess I tend to thing of the Kelvin adjustment as a neutral one.
Click to expand...

Maybe someone else here can better explain the technical reason why this works. I just like to test and experiment to discover how to get the best picture out of Red.
 
I'd like to keep this thread going because I think it's valuable information.

My question - If you're shooting tungsten , is there a noise difference between processing it in RedCine at 5600k and dropping it down in FCP, or processing it at 3200k-2500k in RedCine in the first place? It sounds like the former is what is yielding the best results.

Buy why? It would seem to me that the blue channel is still being pulled up, revealing the same noise only later in the process. Or does the high k processing itself change this as it's being baked in?

I'd really like to understand this so I can get the best results.

Thanks!

Jim
 
You must log in or register to reply here.
Back
Top