I have a kind of “stupid” question about White Balance.
The problem is: I set-up Pentax camera to Daylight, and shoot Raw (DNG) + JPEG, after that I am trying to process the image in Rawtherapee and when I import DNG file and choose White Balance from the Camera - I’ve got Temperature 5811 and Ting 0.997.
Okay, I decided to set up in-camera WB to Kelvin and set up 5000K, when I imported it to RT I got 6043K and Tint 1.101, okay I’ve tried to set up 5560K (in camera) and in RT I have 6467K and Tint 1.014.
It looks like if I choose in-camera WB in Rawtherapee White Balance settings - the images have very strong yellow ting. However, if I manually set Temperature in RawTherapee to Daylight, or to 5600K - it looks okay (okay - because it still have the difference between Pentax in-camera processed images and RawTherapee - processed images).
So my questions are:
Is it okay if I have such difference between the camera WB and Rawtherapee WB (even if I set up WB in K)?
Can anybody recommend any way to make jpeg produced by camera and Rawtherapee looks similar? (I usually shoot to RAW and doesn’t care about in-camera jpeg, but in this particular situation it would be very useful to have images out of camera, and I would like the images from the camera look as close as possible to the images that I will get from RAW late)
When I open your example in RT it says 6494 and tint 1.049 from the camera.
In any case, DNG uses a weird method of interpolating WB between Incandescent and D65 which I don’t trust at all and I never use DNG. If I had that model, I would personally export PEF, not DNG.
White balance algorithms are not interchange and are not standard, for instance rawtherapee and darktable don’t give the same output when given the same file and identails numerical values for white balance. As long as it looks good, it is good.
I also own a couple of Pentax cameras and would recommend using PEF file and not a DNG. I am unsure if this will solve your issue but it might be worth testing.
I think Dmitry is talking about settings, not image rendition. In other words, if his camera WB is set to some K and tint values, I see no reason why RawTherapee or darktable defaults should say different.
I’ve checked the PEF format now, and don’t see any difference in WB. For the 5560K in camera I have around 6480K in RawTherapee.
I also found in Wikipedia:
@paperdigits is correct. White balance information is not saved as temperature and tint (ok fine, you will see them in the metadata, but that’s only for humans to read). There is no singular temperature/tint model. White balance is stored in a different format to remove ambiguity. When you open an image in different software with different temperature/tint models, you will see different numbers even if the white balance adjustment is identical.
A small difference is expected, but your values are off by about 1000K. There might be a bug.
Set the white balance to Camera and enable the auto-matched tone curve for a decent start. The white balance might be buggy, so you’ll probably have to adjust the temperature.
Is there an auto matched colour profile for the K-1II? In that case you should change the input colour profile from auto matched to default or embedded (they are very close) and turn off the look part of the matrix.
Note that your custom image settings in the camera are only partially transferred to RT. You won’t get the hue adjustments of custom image settings auto translated to RT.
The auto matched profile is very different from the Pentax colours and may result in what looks like a different white balance.
You can also extract the custom profiles from Adobe camera raw but the embedded profile is identical. At least the when using the natural look.
Is there an auto matched colour profile for the K-1II?
As I understand I can take the profile from lightroom or create my own. (all that profiles very fare from what pentax do in-camera. I understand that they do some kind of “magic” and even don’t try to repeat it, because as I understand that’s impossible). But the White balance value doesn’t depend on what profile I choose. In RT WB is around 1000K more than it was set in-camera and it looks around 1000K warmer than in-camera jpegs.
The other problem with pentax to Rawtherapee WB conversion - the WB difference isn’t linear (it isn’t always 1000K), it is curve so if I set in-camera WB to 2700K or 6500K the difference isn’t 1000K already, it vary, and might be 700K or 1500K for example.
If I remember right, the in-camera WB setting is used in the camera for the JPEG (embedded and saved), but it is not used for the raw. And as far as I know, this is true for almost all settings.
Thank you Lawrence37, but I am not sure that I understand properly. The number is different, but the color is also different (in RT it looks 1000K warmer than in pentax jpeg), so for example in pentax I set up 5000K and in-camera jpeg image looks like 5000K, but if I open the raw file in RT the number is 6043K and Tint 1.101, and it looks like it is too Warm, like it was taken in Cloudy mode or something like this, so I need to reduce K (in RT) to, I don’t know, something like 5300K or probably 5400K to have more or less similar image. Is it what should be? and this is what you are talking about? Or you mean that only numbers might be different?
You also mentioned:
A small difference is expected, but your values are off by about 1000K. There might be a bug.
Do you know if it is possible to check if it is bug and if it is what should be done to fix it? (if it is necessary I can provide the set of images from Pentax K1 mark II)
If I remember right, the in-camera WB setting is used in the camera for the JPEG (embedded and saved), but it is not used for the raw. And as far as I know, this is true for almost all settings.
RawTherapee White Balance has Camera option. As I understand it means that it should take the WB that is probably not the same but close to the WB that camera use for the image (probably I am wrong and you will correct me).
Yes, and it works reasonably well for me with the OG K-1. If I set
Input profile to: Use embedded
Auto Matched tone curve
Camera WB
Increase exposure a bit (Pentax matrix has exposure added)
My outdoor images look very similar sooc and from RT. I always shoot with Natural custom image though which might simplify things. There are some shifts in some tones but generally quite good. You have to look carefully to identify differences.
I’ve never seen dramatic WB differences between sooc and RT when selecting Camera WB.
I’m not sure i choose the correct one for the last example as I was comparing several settings.
Another thing to know is that the sooc handling of highlights is quite different than RT. Sooc desaturates highlights more than RT unless you do a colour toning with a L mask like this or similar
The temperature value is expected to be a little different from the number shown by the camera when you select the Camera white balance option, but neutral colors in the jpeg should also appear neutral in RawTherapee. If not, then there is a bug. You can report it in the issue tracker. A good sample image to provide is one where the jpeg contains a neutral color.
Thank you nosle! Use of embedded profile is helpful, but unfortunately I still have the WB problem.
I think, I have to follow @Lawrence37 advise and report the bug in the issue tracker.
I opened your dng in RawTherapee 5.12 and the white balance looks close to the jpeg. The temperature value is way off like you describe, but it doesn’t seem to make the image yellow. Applying Adobe’s DCP for the K-1 Mark II Neutral look and the auto-matched tone curve gives me very similar results compared to the jpeg.
Seems, when I shoot the RAW it was SATOBI custom image settings and white balance was set in Mired (5560K equivalent)
I usually don’t shoot in SATOBI, just forget to switch back to natural, but because it was in RAW, I believe it doesn’t matter. I shoot several images in raw, then converted that images to jpeg (in camera) and after that compare those converted images with RT results.
For me the incamera and RT images side-by-side looks like:
left- incamera, right - RT
And if you look at some neutral parts of image, like asphalt, concreet, etc. you can see that right image is much warmer than left image.
okay, I’ve try to set the 5560K manually, and now it is much colder:
