There are many, many resources on the web that describe typical colour temperatures for the likes of sunrise/sunset, daylight, and overcast. Sunrise/sunset is typically placed well below and overcast well above the “standard daylight” number.
So my question is this. What is a likely “normal” colour temperature for a sunset through (thin) overcast? All I can observe with my own eyes is the light is somewhat “pink” rather than “orange”.
I always shoot at 5600°K, unless it is at night with artificial light, where I might want to show the scene as true colour instead of tinted by the illuminant.
That way I get a visual reminder of what the scene looked like at the time of shooting, relative to daylight, and can adjust it to suit how I “felt” it looked - I don’t believe there is one right temperature.
When I shoot 5" x 4" sheet transparency film, I will use filters to bring the colour temperature to 5600°K, unless I want a shot like this…
… where the colour temperature was only 2300°K and I ended up using four different blue colour correction filters and an exposure time of 17 minutes. You can see the flare from one of the filters on the ceiling lights.
I think this sentence is important for sunsets and in general. as soon as emotions become an important part of an image. Except you want to deliver the absolutely identical colours of a sunset, but with this subject the emotion “day’s over, night begins” (which can be positive or negative) is more important to me than delicately appropriate colour rendering at the last 0.1 K. Since I shoot RAW anyway and maybe need the sunset in context with other pictures you will hardly ever get the “exact” sunset colours from me.
I didn’t use a grey card, I used a colour meter. Since sheet film costs around €10 per sheet, I don’t shoot test shots and have to ensure I get things absolutely perfect in camera, first and only time.
Most of my LF work at that time was documentary, not artistic and wasn’t primarily intended for digitising so the transparency has to look perfect on a light box.
Shooting digital RAW is an entirely different game since the WB recorded is only a “suggestion” and you can change it in post if you want. Compared to film photography, it could be seen as “cheating”
Don’t get me wrong, I love shooting digital when it comes to colour balance. It saves me having to carry around either a colour meter or grey card. Although I do sometimes shoot through a translucent white filter ring cover to create an in-camera “preset” that matches the ambient light…
I also use this kind of diffusor ocassionally but it’s only telling me the average of light colour. It’s good to use it for product photography but with anything else I try to get away with “artistical freedom”.
I was just looking for a “starting point” as I also go “by feel” to suit my taste. When looking up “typical” colour temperatures it occurred to me there was this apparent dichotomy that it would be both warmer and cooler than full daylight. I suspect what such charts are missing is that instead of “overcast” they actually mean “overcast daylight”.
Philosophically spoken, most sunsets happen at daylight . Not at night… also, “overcast” could be some fluffy clouds or some close to unload big dark waterballons… I only care about “white-balance setting to not-auto” when I know I have to stack some images, be it panorama, HDR or focus-stack, I’ll choose something close to approximation.
Not in the sense of colour temperature, they don’t, which is exactly where my problem lay. “Sunrise/Sunset” is always listed as warmer than “daylight” which by inference means (and sometimes is given explicitly as) “midday”.
I know, I was just making fun of the sometimes too oversimplified naming of color temperatures. It’s like giving the other temperatures season’s names. As I didn’t know much about color temperature but found it odd to speak of only Kelvin without bringing “tint” into the equation, I wanted to know more about the definition of “color temperature” which has to do with “white balance”. But not only.
The usual classification of colours into cold or warm hues goes back to a subjective perception and cannot be described by a temperature. Artificial light sources give colour perceptions that deviate from daylight. Mixtures of different types of light sources can even disturb the feeling of well-being.
Quoted from Wikipedia My point is, be it “Sunlight”, “Shadows”, “Clouds” or even a “Color Temperature” given in K - we can only approach a region of color balance and getting a starting point for RAW conversion in to the color we find subjectively appropriate for a picture – or we want to reproduce another image per object and need to take care to light it up with light sources with close to no gaps in their color spectrum. At the moment I think (and might be proven wrong) the RAW file includes all colors visible for the sensor. What I can do with them later on is not depending on my WB setting at the moment of taking the shot - except I was taking JPGs.
I think you’re right. The WB is not part of the RAW file but a value used during the conversion to a RGB image.
There’s quite a difference between the English and German Wikipedia pages.
I also think there’s a difference between the ‘color temperature’ and the ‘white balance temperature’. The color temperature is a value that tells us something about the amount of red and blue in the source light, the white balance temperature tells us how those values are to be corrected during the conversion to RGB. That’s the reason why higher white balance temperatures are reddish and not blueish.
Well, it’s “only words”, but then, these are our base of communication… I accept color temperature as there is a in depth description of it and how it is set-up and how only theoretically it is measured (with anon-existent “black-body”). I’m sure physicists will have endless fun to debate all aspects of it, but my simple brain easily goes in overload mode…
I will not use “white balance temperature” as this thing consists of two values (color-temperature and tint). It’s like a combination of °C (or °F) and hPA to describe a “weather temperature” but using temperature and atmospheric pressure for that.
However, I just did that “fun test shots” and used the camera’s K setting without changing and leaving the Amber-Blue-Magenta-Green settings untouched at 0. The S1R was set to 10.000k, DxO PL5 says 11.573K to that value. At 2.500K they were closer: 2.490 K. My takeaway (which again can be wrong): Numbers I can use to measure and compare. I would not rely on them if two different devices give me two different values. EXIFTool states 10.000K and PL just adds 1.573K
And what would CaptureOne show with the same RAWs:
10.862 for the high one and 2.131 for the low one. Again, “not hitting the target”. And btw. the 5500k of the camera: DxO 5.628K, C1 5.781. Both too high…
And I can equalize all 15 “different” RAWs to exactly the same grey-tone. So, WB is important for JPG-shooters, but not for the ones fiddling with RAW.
Color temperature is telling us something of the quality of the source light. White balance is used to correct that source light. There’s also something as the tint but that is part of the same process.
The converter doesn’t know what light was used. It has to determine that from the image or the camera settings. I don’t know how that first is done. Have been looking for that a while. Every converter has a different result.