Clipped highlights (255, 255, 255) reduced with the tone curve tool to make the clipping warnings go away, still have RGB values like e.g. 252, 252, 252.
No matter whether you WB from 3x255 or 3x252, the new WB will overrule the camera’s WB settings (e.g. 192, 127, 127, 242 for a RGGB matrix) and this means that R and B are reduced, which will make the green channels overrule the red and blue channels. Result: The Hulk, Green Goblin, Green, Green Grass of Home…kind of tint.
When I try to WB in Lightroom Classic, it tells me that I should choose a darker patch if the one I chose was too bright. Many cameras can’t WB from blown images and some can, which creates a UniWB setting that makes the histogram show something related to raw data rather than to a rendered jpeg preview. This helps to drive exposure to the right, away from noisy shadows.
Isn’t the wb working on the raw data. That by changing the wb, also with the color picker, it forces a re-demosaicing process and all the other edits afterwards?
I downloaded a color chart but that is in jpg. Can somebody make an image of a color chart, in raw.
well, in PL7 overexpose or not gives same results. can’t use white to set white and even though usually green or grey in your image can be use, it still broken. if i remember right, it’s been broken since PL7.3 (PL7.4 or .5 had major issue and many complaints) 7.6 had them fixed but this white balance remain way off since.
sorry for late news, i keep white balance from “.as shot” in PL. Keith method work with RGB channel in curve, but it’s time consuming if you have multiple images to work with.
As a side note: I only WB if a photo has some colour cast or for most of my underwater photos where there usually is a noticeable colour cast. Sometimes the pipette gives me a starting point to tweak the WB with curves, but most of the time just curves are much quicker.
In most cases my “As Shot” WB setting works just fine.
I’m late to the party, but I can confirm this happens to me as well when selecting whites that are too bright (close to clipped colors such as 250 or more). I’ve never had that problem before in PL7, now in PL8 it’s a big problem.
If we’d scale the “As Shot” levels to e.g. 4096 (for a WB at 4096 ), we’d have to multiply the green levels by 4 while R and B are closer to needing factors of 2 or 3 respectively. This will boost the green pixels in relation to what they need to be as shot - if the capture had been taken in the respective light. The result is a green cast…if the sampled area is overexposed or close to it.
welcome to the club. =]
you can select many white stuff in a picture and they all going to give you a different readout, which makes the pipette unreliable.
When the user clicks with the pipette on a given pixel, he is telling PL that the RGB values at this point should normally be equal. Then, PL computes the correct white balance from the differences between the RGB values. But if the area that has been clicked on is overexposed (or underexposed), these differences may not be representative of the actual light source. In this case the PL computation could be full wrong, especially if the pixels around have very different values.
In Lightroom, the eyedropper displays the 7x7 pixels around the clicked pixel. This helps the user make a more consistent selection and LR make more accurate computations or send a warning indicating that it is not able to make a correct computation.
In Photolab, we don’t know how this is done (what pixels are actually taken into account when computing the white balance) and there’s no warning anyway when the clicked and surrounding pixels don’t allow to make a safe guess.
Using the WB pipette doesn’t change the white levels.
Now, back to the RAW and lower the luminance level at the top of the curve to 252. this will get rid of the over-exposure warning. Then export the RAW to DNG…
Read the “pure white”, which now measures 248,248,248. Even if I bring in the top of the luminosity curve to 255,255,255 the WB stays the same when tested with the pipette.
I don’t know what relevance all this has but, at least, it provides a way to use the pipette on 255,255,255 white, which shows as blown, to get a true white balance for the whole image. But at the cost of creating a DNG.
My mind tells me that this problem seems to be linked only to original RAW files.
You have to click on a NEUTRAL coloured item. What is neutral? It is pure grey. In other words it is where all the RGB values are equal.
Now, if you shoot with the wrong WB the grey item will will not be grey but will take on the colour of the ambient light, such as florescent light. To remove that colour cast you use the eyedropper to measure that colour cast on the grey item and remove it thereby restoring the image to a state as if it was taken under pure white light.
Where do you get white light? A number of places: the sun produces white light but also some artificial sources can produce pure white.
Now here is the real reason I don’t use the eyedropper. When you use the eyedropper you are telling the software that you want the image corrected to a state as if you were shooting under pure white light. Now how often is that the case? What if you were taking a photo of a really orange sunset? The ambient light is orange which makes everything in the photo have an orange cast. This IS what you want and you do not want to remove that because then it is not what you saw when you took the photo.
Unless you are a studio photographer, you will seldom take photos under pure white light. Also, your eyes are very good at compensating for colour casts so it is actually very difficult to judge what is correct and the image becomes very subjective. You will normally adjust the image to what you would like to convey to the viewer which will most likely NOT be perfectly White Balanced.
So, I adjust my images by eye and often come back at a later time to check as my eyes get used to certain looks which may not be exactly what I want. Eyes can and often do deceive you when it comes to colour.
Finally, we all see slightly different colours and even one of my eyes sees a slightly different colour to the other!
I hope that gives other forum members a different view on WB. So, edit your image to what you want and ignore perfect white balance and be happy
I generally agree. For many of my images, the ambient light is what I want. And if the camera doesn’t capture that accurately, I will compensate in ways that the picker tool can’t help me with. Nevertheless, there are plenty of situations which call for accurate grays, whites, flesh tones, and so on. Or we simply need a reference while we journey toward the outcome we want. All of that aside, what we’re talking about in this topic appears to be a bug or at least a need for improvement. I’m glad it’s being analyzed and hope it’s also being reported to DxO through support.
It only happens with raw files. No problem with jpg. The difference might be that with raw files the demosaicing process is done again, the tool works on the raw data. If @Joanna couldn’t reproduce this behaviour with dng files, it tells me that the rgb data is used.
The problem is not that much the color temp but the tint. I don’t know that relation.
I just found that when selecting an area nearby clipping on the dark side the image gets a red overcast.
