Álex Montoya
Well-known member
Just a quick point. The RED ONE manual states that the camera is natively 5000 Kelvin, so I don't see any point in processing the footage at more than that.
Welcome to our community
Be a part of something great, join today!
Quick Tungsten Test
- Thread starter Gregor Hagey
- Start date
- Thread starter
- #42
Gregor Hagey
Well-known member
You might be right. I was just using Red Alert's default of 5600K more for the purpose of comparison for the test then a scientifically proven 'best setting' to minimize noise. I haven't done enough analysis to say definitively what Kelvin setting produces the least amount of noise for Tungsten light.
Gunleik Groven
REDuser Sponsor
They've changed the native from 5600 to 5000 as of build 16.
kay madsen
Well-known member
When I read this thread it reminded me of some strange issues I had with RedAlert not too long ago.
While doing some tests I captured a low light shot with an old 18mm cooke speed pancro lens set to wide open at T2, ASA Setting of 320 and with a tungsten practical in frame that had part of it´s lampshade overexposed. When I look at the shot in REDAlert with a color temperature setting of 5000K the way the highlights roll off into the clipped areas looks quite acceptable.
When I pick a white balance in REDAlert or in the camera of around 2800K (the measured color temperature of the tungsten fixture), the clipped areas become tinted in cyan and look posterized. I didn´t touch anything except the white balance. ASA320 in both shots. If I lower the white balance gradually in RedAlert (or in camera) the effect starts to appear at around 3000K. If I go lower it becomes worse...
I would have expected the white pixels in the clipped highlights to stay white if I change the color temperature. I took a tiff still image of the shot into photoshop to check what happens when I use the PS color balance tool on it. In PS the white pixels stay white no matter how aggressively I play around with the red, green or blue settings. If I do the same thing with the r,g,b settings in RedAlert the white pixels change in the same way they did in the shot I uploaded, they get an overall color tint.
Interestingly when I take the original R3D file into Redcine and enter the same 2800 Kelvin settings as in RA the white pixels stay white aswell... I had to go down with the Kelvin settings in RC to about 2000K until the same artifact started to appear. It looks like RedAlert does process the color settings in a different way than Redcine or PS and somehow affect the white pixels in clipped highlights in a very bad way.
Maybe these issues are somehow related to what Gregorios discovered?
While doing some tests I captured a low light shot with an old 18mm cooke speed pancro lens set to wide open at T2, ASA Setting of 320 and with a tungsten practical in frame that had part of it´s lampshade overexposed. When I look at the shot in REDAlert with a color temperature setting of 5000K the way the highlights roll off into the clipped areas looks quite acceptable.
When I pick a white balance in REDAlert or in the camera of around 2800K (the measured color temperature of the tungsten fixture), the clipped areas become tinted in cyan and look posterized. I didn´t touch anything except the white balance. ASA320 in both shots. If I lower the white balance gradually in RedAlert (or in camera) the effect starts to appear at around 3000K. If I go lower it becomes worse...
I would have expected the white pixels in the clipped highlights to stay white if I change the color temperature. I took a tiff still image of the shot into photoshop to check what happens when I use the PS color balance tool on it. In PS the white pixels stay white no matter how aggressively I play around with the red, green or blue settings. If I do the same thing with the r,g,b settings in RedAlert the white pixels change in the same way they did in the shot I uploaded, they get an overall color tint.
Interestingly when I take the original R3D file into Redcine and enter the same 2800 Kelvin settings as in RA the white pixels stay white aswell... I had to go down with the Kelvin settings in RC to about 2000K until the same artifact started to appear. It looks like RedAlert does process the color settings in a different way than Redcine or PS and somehow affect the white pixels in clipped highlights in a very bad way.
Maybe these issues are somehow related to what Gregorios discovered?