White balance doesn't work correctly

[kaffeeundsalz]

20 minutes ago, R C-R said:

As you say, it is all based on perception, so how can that possibly have nothing to do with it?

I didn't say that it has nothing to do with perception. I said that the degree of detail of your scientific approach is completely unnecessary to get to an image that is properly white-balanced. That's because, to stick with my example, if you remember an object to be white, you can make it appear white in the photo, too. Whether it actually has a different color from a physical or perceptional or divine perspective is totally insignificant because you already are where you want to be with your image. Or, in other words: White balance is surprisingly not about finding some universal truth about how we are all expected to perceive color. That would be the same as starting a discussion about the essence of being with someone who just asked for your ID.

[lepr]

.

 

[R C-R]

28 minutes ago, kaffeeundsalz said:

I didn't say that it has nothing to do with perception. I said that the degree of detail of your scientific approach is completely unnecessary to get to an image that is properly white-balanced.

But that depends entirely on what you consider to be the "proper" white balance, right? So for your example let's say you remember an object to be white (whether physically it is or not) & use that with what would be considered to be a "correctly" working white balance tool. Would that really put you 'where you want to be' for whatever you want to do next with the entire image, particularly for the overall magenta-green tint correction?

Or maybe more to the point, consider the image that started this discussion & the various WB adjustments people have tried with it. Do any of them look like they are properly white balanced? Maybe it is just me but none of them look like where anybody would want to be because perceptibly nothing in them I would expect to be white (whether it is physically or not) actually looks white.

So sure, not everything requires a highly detailed scientific approach but at the bare minimum I think it is important to understand that the "color temperature" parameter is really just a "correlated color temperature" & that is an imperfect approximation of most real world lighting, as is the secondary magenta-green correction. Absent that, I do not understand how anybody expects to choose the right tool to achieve whatever it is they want to achieve, much less how to use it to do that.

[MikeW]

Normally I stay out of such discussions as regards what is the "right" look for a photo...especially when I have no idea of what the room actually looks like. Product shoots done by professionals are far easier to "know" what is right.

Here's a quickie in split view. Done in a couple seconds, just not in APhoto. Is it right? No idea. But removing the cast and hitting "white" is easy enough in most software.

In APhoto, I just ran the Auto buttons on the toolbar and achieved this...

It should only take another few seconds to remove the red hue/tint hitting the mattress/cover.

I did try the white balance. I do think it is either broken (not working as Serif intends) or simply not working as per other applications (but working as Serif intends). In either case, I think it needs work.

[firstdefence]

• 18% Gray Card
• ExpoDisc
• White Paper
• Kelvin... Maybe
• Memory

I think it was Monet who did a study on colour perception, he repainted a scene that he'd done when he was younger and noticed the scene he'd repainted was much bluer and less vibrant.

What colours we see today as younger people, those same colours we will likely not be seeing in the same way in 30 years time.

In reality, this is probably just a snapshot of "a bit of decorating I did" and put on FriendFace: IT Crowd SE3 EP5

[kaffeeundsalz]

@R C-R, I think our opinions are closer to each other than you think, but it's amazing how you don't seem to understand what I'm saying over and over again. Maybe it's because I'm not a native speaker? If so, my apologies for not being able to express myself clear enough. But maybe it's because you always quote just some of my lines, so I wonder if you actually read my entire post.

8 hours ago, R C-R said:

But that depends entirely on what you consider to be the "proper" white balance, right? 

Right. Unless you think that the uncorrected original example of this thread has proper white balance. But I already said this.

8 hours ago, R C-R said:

So for your example let's say you remember an object to be white (whether physically it is or not) & use that with what would be considered to be a "correctly" working white balance tool. Would that really put you 'where you want to be' for whatever you want to do next with the entire image, particularly for the overall magenta-green tint correction?

In most cases, yes. It depends on the look you want to achieve. But I already said this.

8 hours ago, R C-R said:

Or maybe more to the point, consider the image that started this discussion & the various WB adjustments people have tried with it. Do any of them look like they are properly white balanced?

In a way, all of them are (except for the unedited original image because that one has a clearly visible orange shift). It depends on the reference point you pick for white balancing. But I already said this.

8 hours ago, R C-R said:

Maybe it is just me but none of them look like where anybody would want to be because perceptibly nothing in them I would expect to be white (whether it is physically or not) actually looks white.

That might be because the images need further editing. Why don't you go and edit that image so that you're satisfied with the result and post it here?

9 hours ago, R C-R said:

So sure, not everything requires a highly detailed scientific approach but at the bare minimum I think it is important to understand that the "color temperature" parameter is really just a "correlated color temperature" & that is an imperfect approximation of most real world lighting, as is the secondary magenta-green correction.

True. But "neutral color" is clearly defined in image editing. In RGB, it's when all three parameters R, G and B have the same value. In HSL/HSV, it's when S is 0. That's what you're white balancing against. Does it correctly reproduce colors of real world lightning? No. Is the perception of real world lightning subjective? Yes. Are there cases when neutral color values are not desirable at all in an image? Yes. But I already said this.

[R C-R]

9 minutes ago, kaffeeundsalz said:

R C-R, I think our opinions are closer to each other than you think, but it's amazing how you don't seem to understand what I'm saying over and over again.

I might say the same about what you & others have said (both about opinions & understanding).

Anyway, to summarize my position on this as briefly as possible, I believe it is always beneficial to understand what the science (for want of a better word) tells us about the quirks of human color perception & its affects on our work, just as understanding the differences between what a camera "sees" & what we do is beneficial for the same reason.

Taking that into account, what seems "clearly defined" often is not, or is considerably more complicated & context sensitive than it otherwise might seem.

39 minutes ago, kaffeeundsalz said:

Why don't you go and edit that image so that you're satisfied with the result and post it here?

I would never be satisfied with that image but I already posted the result of a very simple edit & how I did it using Affinity Photo. Did you miss that?

[lepr]

.

[kaffeeundsalz]

@owenr Certainly I did! 

17 minutes ago, R C-R said:

I would never be satisfied with that image but I already posted the result of a very simple edit & how I did it using Affinity Photo. Did you miss that?

I saw it, but you said "It is far from a perfect fix for this photo", so I thought you could provide one. I was just curious what the result would look like. But if you say you'd never be satisfied, I can understand that. Never mind then.

[R C-R]

10 minutes ago, owenr said:

@R C-R, did you understand my reply to your request for an explanation ...

I did, but did you consider the relevance of my comment about the difference between color temperature & correlated color temperature when trying to determine an appropriate temperature setting? It is an example of what I meant about "clearly defined" things often not being as clearly defined as they seem.

[lepr]

.

[R C-R]

1 hour ago, owenr said:

You are still completely missing the point. Or pretending to.

Neither one is true. I am just considering things that most of you do not seem particularly interested in considering, like the difference between approximation & exactness or more generally why things that seem clearly defined so often are not.

If you don't want to consider such things that is fine by me. As it happens, I do. If that is not fine by you, that is your problem to deal with, not mine.

[kaffeeundsalz]

Just because something can be considered doesn't mean it's necessary. You might trick less experienced people to believe they have to know all of this stuff to master a simple white balance adjustment.

[lepr]

.

[R C-R]

56 minutes ago, kaffeeundsalz said:

Just because something can be considered doesn't mean it's necessary.

If you do not want to consider such things, whatever the reason, then don't. No trickery is involved. It is that simple.

[henryg]

I like the Essentials results, so can I easily or by macro replicate in AP the results I used to get in eg Essentials 14 using the colour pickers in Levels?

I find AP very frustrating in this respect, and really don't care if the tools work as intended or indeed "correctly".

[Sempervivum]

Just updated to Version 1.6.5 and look, the white balance by use of the picker works much better! Well done!