NEW quick tungsten test

[Álex Montoya]

Gamma lifted to see what's going on in the shadows. Here:

I will post a little video shortly.

 

[Julio Quintana]

Wow. That blue noise is pretty bad. I've also been having pretty bad luck with tungsten lighting. In daylight, the footage looks amazing, even in direct sunlight. But under low tungsten bulbs, it almost looks like a different camera shot it.

 

[Álex Montoya]

The thing is, if you process it right, the image is pretty clean.

 

[Fredrik Harreschou]

Exactly. I always grade from a daylight balanced image, no matter what the color temp. The color temp meta data is for offline purposes, IMHO.

 

[Hans von Sonntag]

Just arrived from a studio shoot all lit tungsten.

Using a 80C is the way to go, IMO. Greatly inproves the blue noise issue and also broadens the DR a bit because the red channel is more in line with the blue and green channel and does no clip as early.

The 80C technically steals a stop. A downside, for sure - at first glance. But Reds sensitivity on tungsten is closer to 250 ASA anyway than the 320-500 ASA under daylight. In the end you might lose only half a stop or so wich is in most cases not a dealbreaker.

Conclusion: Like the 85 with film the 80C is an even more important standard colour temperature conversion filter in a filter kit when shooting with the red.

Hans

 

[Evin Grant]

Concrete, you should black balance that camera, you're getting a split screen dude:biggrin:

 

[Michele Gavazzeni]

a lot of noise but it seems organic to me
i like the 5000K process
the 3200 makes the noise worst

Concrete, you should black balance that camera, you're getting a split screen dude

lol

 

[Julio Quintana]

Concrete,
What is the best way to process that footage? I tend to just use the Redcine defaults with Redspace, using the tungsten preset on camera. I know this is kind of amateurish. What's a more hands on way to do this?

 

[Brandon Fraley]

generally speaking, i would think REDspace wouldn't be the best space to convert to. I think rec709 is probably better, but I'm not sure, just my opinion. I think of REDspace more as a Monitoring assist, might be wrong.

 

[Álex Montoya]

I slightly touch the controls to make sure that the whole histogram is there and I export 2K 10 bit DPX's at 5000º in rec709 since it is more of a standard color space.

 

[Gunleik Groven]

Exactly. I always grade from a daylight balanced image, no matter what the color temp. The color temp meta data is for offline purposes, IMHO.

Fredrik, my impression is that you've understood a thing or two...

An other advantage for you (Grading with Scratch) is that the exposure/iso metadata is NOT automatically taken into consideration. That in itself leaves a ton of noise issues out, as compared to grade from a "redspace" starting point...

If it was a default option to grade from the Lin image (similar to hitting F in RC) I think more people would be able to wrap their head around the format...

Just my 2 Icelandic kr...

 

[Gunleik Groven]

weird shit happening in browser - edited away...

 

[Hans von Sonntag]

Fredrik, my impression is that you've understood a thing or two...

An other advantage for you (Grading with Scratch) is that the exposure/iso metadata is NOT automatically taken into consideration. That in itself leaves a ton of noise issues out, as compared to grade from a "redspace" starting point...

If it was a default option to grade from the Lin image (similar to hitting F in RC) I think more people would be able to wrap their head around the format...

Just my 2 Icelandic kr...

Yes, fully agree.

I always keep the camera at 5.000K @ 320 ASA. This way I know that metadata matches more or less what the sensor "sees". If necessary I use filter to WB the picture (eg. 80C or 81EF).

I always create my offline rushes with RedRushes and do a first grade with RedAlert in RedSpace to achieve nice looking and client-friendly footage.

Later when all is conformed I grade off the linear R3D files without any metadata.

Hans

 

[Noah Posnick]

Concrete, can you post a still when it's processed correctly without the gamma lifted for comparison - again with the tungsten/daylight split? That would be most helpful. Thanks.

 

[Álex Montoya]

Here you have it.

Keep in mind that the image was a bit underexposed and that the jpg compression may have introduced some artifacts.

 

[Hans von Sonntag]

Concrete,

I must say that I just realized now that you did a very valuable test concerning the development of R3D files.

Is does show that Red's tools are NOT doing a good job if we talk about WB 3200K. If there is is no possibility to WB in camera using filters, processing at 5000K and CC later is the way to go.

Good job!

Hans

 

[Gunleik Groven]

I've sorta tried to make that point for a while...

 

[Dustin Cross]

One thing I found is if you take tungsten shot footage into RC/RA and set white balance to 5000k then use the color sliders to fix color it is much less noisy than leaving white balance at 3200.

This is the same thing you are doing in After Effects, but on the RAW footage.

Can you test that with this image Contrete?


Mahalo,
Dusty

 

[Álex Montoya]

Ok, I will.

Of course, the credit for this finding goes to people as Gunleik and Gregorios, who have posted prior tests in this direction.

 

[Antoine Fabi]

...and ANY image will show more noise when you lift what was recorded with near absent light.