Mysterium sensor: Color temperature
- Thread starter macgregor
- Start date
Nearly all digital still SLR cameras with CCD or CMOS sensors provide a RAW file option these days. None of the raw files have fixed "native" color temperature, because the unprocessed output of the sensor does not have a blackbody-like response. However, when you process the image you can set the color temperature to whatever you want.
As far as other digital cinema cameras go, if you look at the raw, ungraded output of a Viper it looks green.
http://www.thomsongrassvalley.com/docs/Miscellaneous/cameras/viper/Viper_FAQ.pdf
* What is the normal colour temperature setting on the Viper?
The CCDs in the Viper respond directly to the light falling on them without any colour balance taking place. This means that pictures will normally appear to have a green tint, which is not balanced out at any colour temperature. This can be compensated in post production.
As far as other digital cinema cameras go, if you look at the raw, ungraded output of a Viper it looks green.
http://www.thomsongrassvalley.com/docs/Miscellaneous/cameras/viper/Viper_FAQ.pdf
* What is the normal colour temperature setting on the Viper?
The CCDs in the Viper respond directly to the light falling on them without any colour balance taking place. This means that pictures will normally appear to have a green tint, which is not balanced out at any colour temperature. This can be compensated in post production.
Anders Holck
Well-known member
Speaking of the viper it's closest to something like 5000 kelvin, but with with a major greenshift which is not applicable with the kelvin scale. If you use a CC60M Magenta filter and a CC81B blue filter you get close to 5600, but loose 2 1/3 stop of sensitivity.
The good thing about doing the filtration optically is that saturation (Clipping) of the pixel sensors happen at the same time and that the gains on each channel is equal, so noise should be minimal.
Remember in Filmstream mode when you tweak the color balance you are actually doing electronic gain and attenuation. In low light this can be better than loosing the sensitivity, but if you have enough light you will gain dynamic range and SNR by filtering optically.
With todays low noise sensors and RAW engines the difference might be unnoticeable though...
The good thing about doing the filtration optically is that saturation (Clipping) of the pixel sensors happen at the same time and that the gains on each channel is equal, so noise should be minimal.
Remember in Filmstream mode when you tweak the color balance you are actually doing electronic gain and attenuation. In low light this can be better than loosing the sensitivity, but if you have enough light you will gain dynamic range and SNR by filtering optically.
With todays low noise sensors and RAW engines the difference might be unnoticeable though...
Last edited:
Graeme_Nattress
Red Team
Um... Sensors don't have a colour temperture. They have a colour space defined by the way the sensor is made and it's colour filters.
Graeme
Graeme
CamDiver
Active member
So white balancing is done in post then? I've sifted through some pages of this forum and not found the point i'm looking for. I also take it that the histogram will also negate any need for light metering? Just asking 'coz I'm trying to get my head around the concept design for an underwater housing. I'm chatting with the president, CEO, of Gates Housings regarding this requirement.
Cheers,
Mark.
Cheers,
Mark.
Graeme_Nattress
Red Team
Well, if you're recording RGB, white balance is done in camera. If you're recording RAW, it's done in post, but you can see it being done in camera for the viewfinder.
Histogram does, for the most part, mean you can get accurate exposure without a light metre.
Graeme
Histogram does, for the most part, mean you can get accurate exposure without a light metre.
Graeme
glenn chan
Well-known member
If you assumed that the light source were from a black body radiator (or daylight-series illuminant, etc.) then you could figure out the correlated color temperature of the sensor. ?
Of course, the key assumption here is the spectral distribution of the light source, and this may be wrong (i.e. when shooting tungsten or daylight or HMI). To account for that, maybe you could measure the color temperature for common light sources and then average them and report the error.
1b- Instead of correlated color temperature, it might be more useful to specify light sources + specific gels/filters (i.e. from GAM, Lee, Rosco catalogues) or specific Red accessories.
2- This information is vaguely useful for people who want to filter the camera to get maximum exposure latitude. Similar to the debate about the Viper and whether or not to use Magenta filters. My examination of that is that filters should be used unless you need the extra sensitivity. Electronic white balance has to intentionally clip information, or guess highlights.
Of course, the key assumption here is the spectral distribution of the light source, and this may be wrong (i.e. when shooting tungsten or daylight or HMI). To account for that, maybe you could measure the color temperature for common light sources and then average them and report the error.
1b- Instead of correlated color temperature, it might be more useful to specify light sources + specific gels/filters (i.e. from GAM, Lee, Rosco catalogues) or specific Red accessories.
2- This information is vaguely useful for people who want to filter the camera to get maximum exposure latitude. Similar to the debate about the Viper and whether or not to use Magenta filters. My examination of that is that filters should be used unless you need the extra sensitivity. Electronic white balance has to intentionally clip information, or guess highlights.
Graeme_Nattress
Red Team
Yes, manual white balance in RGB. Histogram should work because all it's saying is "this is what the sensor recieved in way of data" and then you can react accordingly.