This content has been marked as final. Show 11 replies
Hi Selby, yes your differences that you mention are expected. In general warm color saturation is much higher under tungsten-filament lighting and this is being reflected in the 2850 k "half" of the dual-illuminant profile. This is absent from the single-illuminant profile because the single-light case uses D50 reference data.
I am somewhat surprised to hear you are getting much higher saturation in daylight images, as the 2850 k table contribution shouldn't be very much.
Of course there is the option to tweak the profile a bit (i.e., reduce saturation). Another option is just to reduce the saturation using the per-image controls within CR/LR.
>However, on some shots the saturation (for example, of autumn foliage) is verging on being somewhat over the top
What's the white balance your camera finds for those? My experience is that many cameras will wrongly creep down the color temperature for shots of autumn foliage because it gets fooled into thinking the temperature of the light source is far lower than it really is because of all the warm colors. This might increase the amount of influence the Tungsten calibration has, consequently increasing the saturation. You might want to try preset white balances such as daylight or cloudy for these shots.
Eric - thanks for the response. Just FYI, the increase in warmth and saturation is very noticeable.
One additional piece of information I forgot to mention - the profile I created gives much different results (increased saturation and warmth) when compared to the Adobe Standard Beta 2.
Also, does the temperature of the illumniants being "off" from the 6500K and 2850K standards matter (not quite sure what I can do about it, but still good to know....)?
Jao - I do my white balance from a WhiBal, so it should be pretty much bang on. In fact, I do a custom white balance in the camera and use that, and then in ACR I double check the value from the shot of the WhiBal against the value in the shots using that custom white balance. So far they always agree.
Thanks for the response,
Well what is it? If you have a temperature between the two extremes, the ACR engine interpolates between the two profiles, so if your temperature even when measured with a card is much lower than 6500K, you get a significant contribution from the Tungsten profile if I understand correctly how this works.
P.S. be careful with white/grey cards in conditions like those. If you get much filtered diffuse light that is coming through a canopy of colored leaves hitting the card, the card is not supposed to be grey in the image but should look yellow (if the leaves are yellow) and you will get a false color temperature reading. This is because the light is no longer a broad black-body like spectrum but has whole bands filtered out of it. Similar situations occur when you have colored artificial lighting or sunset/sunrise conditions. If you use a blue filter on a halogen lamp for example, the grey card will read a very high color temperature and correcting using the grey card reading will make all the blue light look white! The actual temperature you should be using to record what your eye sees in that situation is a tungsten temperature corresponding to the lamp. Grey cards only work correctly when the light source hitting it is white to your eye. Just saying!
Jao, your point is well taken as the temperature in some cases had fallen below 5000K.
In the case of skin tones, where I really liked what I got, I was using flash with a Lightsphere in a room with a white ceiling and neutral walls. The WhiBal gave a reading in the mid-5000s.
In all cases, I was out in the open so the reading was "accurate", but I do appreciate the issue of "what color temperature actually gives me what I see" vs the "actual color temperature". While the eye adapts somewhat to the color of the light, we all see everything look warm in late day sun. Keep moving the white balance slider until it "looks right" I guess... But the WhiBal does give me a good place to start. Have you found any rules of thumb that help?
Selby, as you've observed the human visual system does have limits regarding adaptation. This is why faces seen in candle-lit restaurants have that red glow, for instance. A gray reference also basically assumes a single illuminant which is not really correct in general outdoor situations. But as you say in practice it's a pretty good starting point.
Adobe Standard beta 2 is a mix of scene-referred measurements as well as output-referred "effects" (to address color criticisms given to Adobe in the past), so it is expected to be different from the Chart Wizard profiles, which are purely scene-referred.
>But the WhiBal does give me a good place to start. Have you found any rules of thumb that help
No, I've only used it for portrait shots once in a while and there it did wonderfully and gave me far better results than just trusting my camera. I might sometimes make the temperature a little warmer just to make the skintones more flattering as they did tend to be pale, which reflected reality but not what is usually desired. For my normal photography, which is landscape oriented, I don't think I have ever used it (I should try it sometimes though). I usually just adjust until it looks right. I'm a color junkie anyway so I tend to boost colors over the top, especially if they are warm sunset/sunrise glows.
Eric - thanks for the additional info.
Since you earlier identified the tungsten illuminated image as the source of the additional saturation and warmth, I decided to reshoot the color checker using more wattage (the 1st shot used 100w and a 60w bulbs - this time I used two 100w bulbs). This gave me a temperature of 2800 and a tint of 0. Using this new shot plus the original high temperature (6250 with tint +3 - my initial post had the temperature stated incorrectly), I build a new dual illuminent profile.
The change of +100 degrees made a big difference. The increase in saturation and warmth from the single illuminant profile is now much more subtle.
This then raises some questions:
1. How important is it to get the temperatures exactly at 2850 and 6500?
2. Is one more important than the other?
3. For single illuminant profiles, should the image of the color checker be shot as close to 5000K as possible?
I am thinking the difference has more to do with how you photographed the target or possibly spectral differences between the bulbs, rather than the difference in color temp. Keep in mind that if you have glare off the target then this will wash out your image and the DNG PE will attempt to compensate by adding saturation.
In general it's not just the color temp that matters but the spectrum of the actual illuminant. This info is usually not easy to come by, but if you have a standard tungsten-filament bulb then it will come very close to Standard Illuminant A (~2850 K CCT).
So, the issue with stating "as close to 5000K" as possible is that there are infinite number of lighting conditions that equate to 5000 K but actually they produce different colors. You can have Solux bulbs, CFLs, fluorescent tubes, and daylight that produce 5000K but they will all render colors differently.
PE uses the Standard D50 Illuminant for building single illuminant profiles. For an artificial light source the closest you'll probably come to matching this would be a Solux 5000 K bulb.
Per your question 3, my understanding of single illuminant profiles is the spectral quality of the illuminant on the color checker when photographed is not important - what matters is that the same illuminant is present for images using that profile.
In other words the single illuminant profile does no interpolation based upon the shooting wb, it's a rifle shot and only exact when the same illuminant conditions exist for the final image as for the color checker exposure.
As such, it's great for camera/lens/flash combinations. Practically speaking, it's also useful for daylight illumination that doesn't stray too far from that present for the color checker shot.