Default color for a new correction layer or a mask is not white anymore

[walt.farrell]

43 minutes ago, Alfred said:

For the partially selected edge pixels (the two at each end in my example) whatever you apply will be partially applied, depending on the degree of selectedness.

So the entire pixel is affected, but the effect is only x% applied?

[Alfred]

11 minutes ago, walt.farrell said:

So the entire pixel is affected, but the effect is only x% applied?

That’s my understanding. Or misunderstanding, if I’m missing something!

[smadell]

To understand selections, one must understand channels. Because a selection is a channel.

Look at the Channels panel. Each and every layer is composed of a number of channels. When we look specifically at pixel layers in an RGBA document, each layer has red, green, and blue channels which denote color; there is also an alpha channel which is used for transparency.

Look toward the bottom of the Channels panel. There is a separate channel called Pixel Selection. This channel contains the active selection. (1) It is fluid, in the sense that its contents will change depending on what, if anything, is selected. (2) It does not belong to a layer, but exists on its own. (3) Most importantly, like all channels, it has a single entry for each individual pixel.

That last notion is important. A pixel is the most elementary unit of an image. Each pixel can contain one, and only one, color. Each pixel can contain only one level of transparency. And each pixel in the Pixel Selection channel can also include only a single value.

This is why a single pixel cannot be two different colors. This is why a single pixel cannot be half opaque and half transparent. And it is why a single pixel is either selected or not, but also why a selected pixel can be partially selected.

Let’s first clear up the “where is the 50% line” question. In the image below, I’ve created an elliptical selection and applied a 100 pixel feather to it. On the right, I’ve zoomed into the marching ants at 1400%. The marching ants line indicates the border at which pixels are 50% selected. PLEASE note that the line does not go through individual pixels. It denotes entire pixels, because each pixel can only be selected by a specific fraction (0% to 100%, or any value in-between)

The answer to questions by @R C-R and @walt.farrell about how partial selection affects subsequent operations is really a fairly straightforward one. (1) Once a selection is made, that selection is stored inside the Pixel Selection channel. (2) In many cases, for instance in straightforward use of the marquee tools, the Pixel Selection channel is white where the selection is active, and black where the selection is not active. (3) In some other cases, such as Selection From Layer, Luminosity based selection, and feathered selections, the Pixel Selection channel contains pixels that are all shades of grey. These grey pixels within the Pixel Selection channel indicate document pixels which are partially selected. (4) Subsequent use of the selection acts on whatever layer is activated in the Layers panel. The specific action that is undertaken (whether that is a Copy command, the creation of a Mask or and Adjustment or Filter, or anything else) is modified by the values in the Pixel Selection channel.

What that last clause means is that my actions are multiplied by the value of the Pixel Selection channel for each and every pixel. In other words, if the Pixel Selection channel is black for a particular pixel, then my action is multiplied by 0. If the Pixel Selection channel is white for a particular pixel, then my action is multiplied by 1. And, if the Pixel Selection channel is, for instance, 45% grey then my action is multiplied by 0.45.

Once you start thinking of selections as a separate channel, not attached to any particular layer, they become easier to understand. Also, it becomes a lot easier to understand how a pixel can be partially selected, and how a mask or adjustment layer created based on such a selection may contain areas which are grey.

[R C-R]

11 minutes ago, smadell said:

That last notion is important. A pixel is the most elementary unit of an image. Each pixel can contain one, and only one, color. Each pixel can contain only one level of transparency. And each pixel in the Pixel Selection channel can also include only a single value.

This is why a single pixel cannot be two different colors. This is why a single pixel cannot be half opaque and half transparent. And it is why a single pixel is either selected or not, but also why a selected pixel can be partially selected.

The underlined sentence makes no sense to me because the last part contradicts the first. A pixel is either selected or it is not, period. There is no way for the Affinity apps to partially select a pixel because it is indeed the smallest, most elementary, indivisible unit of a raster image, a pixel channel, or anything else that is based on pixel units.

What the apps can do is provide ways to adjust the strength of whatever is applied to selected pixels.

[Alfred]

1 minute ago, R C-R said:

A pixel is either selected or it is not, period.

How do you explain the fact that when you make a pixel selection of a fully opaque patch of colour, feather the edges of the selection and then copy and paste the selected region, the resultant edge pixels are less opaque than the central ones?

[smadell]

@R C-R I think we're arguing semantics here. I am using the term "partially selected" because that is the term that Serif uses, and it is the one that is generally used amongst most photo editing programs. In that sense, "partial" selection implies two things: (1) the entire pixel is selected, since as you correctly note – a pixel is either selected or it is not; and (2) the fraction by which it is selected can vary – not the fraction of the physical size of the pixel, but rather the fraction of the strength of the selection.

Since I think it is clear that a selection constitutes a channel, let's look at Masks instead, since these are more understandable for most of us. In the graphic below, I started with a photo, added a mask to it and drew a grey line on the mask. I've shown the result on the right side. Look at those pixels in the middle: are they masked or not? Of course they are. Are they fully masked? No, they are partially masked. And the fraction by which they are masked is exactly the fractional amount of greyness that exists in the Mask channel.

Why can we not agree to apply the same logic to Selections, since Affinity Photo clearly sees the selection as a channel?

[R C-R]

14 minutes ago, Alfred said:

How do you explain the fact that when you make a pixel selection of a fully opaque patch of colour, feather the edges of the selection and then copy and paste the selected region, the resultant edge pixels are less opaque than the central ones?

What needs an explanation? All the pixels are fully selected, including those not enclosed by the marching ants. It is just that what is pasted takes into account the feathering applied to those pixels; IOW, the strength of what is applied to them during the copy process. In this instance that is the reduced opacity applied to the feathered pixels.

7 minutes ago, smadell said:

... and (2) the fraction by which it is selected can vary – not the fraction of the physical size of the pixel, but rather the fraction of the strength of the selection.

It is a bit more than just semantics: it is the strength of whatever is applied to the pixels of the selection that can vary, not the strength of the selection itself.

[smadell]

You're putting the cart before the horse. It is indeed the selection that varies in strength. This should be obvious when you inspect the Pixel Selection channel.

Also, assuming that there is an active selection, an activity such as creating a Mask or Adjustment Layer means that the selection is applied to the Mask (for example), not that the Mask is applied to the selection. This might seem like a trivial distinction, but it is crucial to understanding the way channels work.

Channels are multipliers. When displaying a color, Affinity will put up a color that is G(128) when the Green channel is 50% grey for that corresponding pixel. Basically, it uses the multiplication (Green * 0.5). If my Mask is 80% grey, then the pixels being displayed are (Full Transparency * 0.8). And, if I use a Selection whose Pixel Selection channel contains pixels that are, for example, 30% grey, then the action subsequently performed on those pixels takes the general form: (Action * 0.3).

[Old Bruce]

What is happening here is that the selection of a pixel can be only all of the pixel's area but only some of the 256 discrete levels of grey (I am using 8 bit for this). Ten percent of a pixel being selected is 100 % of its area and 10 % of the value of the channels's (1 up to 4 if we ignore the alpha) grey values. A pixel that is more than 50% selected will be enclosed within the marching ants, less than 50% is outside of the marching ants. 

[smadell]

So, @R C-R – I will amend my previous post a little, because I still believe we both understand how to get from Point A to Point B and we're just explaining it (to ourselves and others) in different ways.

Once I create a selection, I have inserted certain values into the Pixel Selection channel. Those values can be anything from 0 to 1 (or 0 to 255, in 8-bit parlance). When I use that selection to create a mask (as an example) the mask reflects the selection in the sense that selected areas within that mask are black, white, or grey depending on the state of the selection.

Since multiplication is commutative, it is always the case that (A * B) is the same as (B * A). So, it really doesn't matter too much if the mask is applied to the selection, or the selection is applied to the mask. The result is always the same, since (Mask * Selection) is the same as (Selection * Mask). You can understand this in any way you like, and any way that makes the most sense to you. But I think we're still seeing the same basic process going on. And, I think we both know how to get the result we anticipate.

[R C-R]

3 minutes ago, smadell said:

You're putting the cart before the horse.

Nope. You cannot just ignore the fact that a pixel can either be fully selected or not selected at all. There is no wiggle room here. There is no such thing as a partially selected pixel.

The channel multiplication you describe is just another example of how the strength of what is being applied to fully selected pixels can vary.

[smadell]

It is becoming clear that you believe the phrase "partially selected pixel" is an oxymoron. This is conflating the area of the pixel with the degree of its selection. While the entire area of a pixel can be selected, that selection can also be less than full strength. Both can be true at the same time.

If you don't believe that a pixel can be "partially selected," then you place yourself at odds with Serif, Adobe, and virtually all other photo software companies who quite freely use this phrase to explain the selection process.

[R C-R]

2 minutes ago, smadell said:

If you don't believe that a pixel can be "partially selected," then you place yourself at odds with Serif, Adobe, and virtually all other photo software companies who quite freely use this phrase to explain the selection process.

At least for the Affinity apps, I do not think this is true. As I said earlier, the Affinity help topic does not say a pixel can be partially selected, only that an area of a layer or layer group can be partially selected. If there is anywhere in any Affinity help topic that explicitly states that a pixel can be partially selected, I cannot find it.

I think "partially selected" only makes sense when referring to some area of the document. In part, this is because "marching ants" selections do not belong to any layer or any other object in the document. In fact, as you mentioned earlier, they can be made even when the document has no layers, no objects, or no pixels in it at all, like in this Just ants.afphoto example.

[smadell]

What happens when an irresistible force is applied to an immovable object? Unfortunately, this is the game we are playing.

I mean no disrespect, but it seems to me that you are inventing facts in support of a foregone conclusion. Nevertheless, I am done. I know that I cannot convince you with my interpretation, and it should be clear that you cannot convince me. I am OK with that. But, this debate is officially over.

[R C-R]

9 minutes ago, smadell said:

I mean no disrespect, but it seems to me that you are inventing facts in support of a foregone conclusion.

I am not inventing the fact that I could not find anything in the Affinity help that explicitly states that a pixel can be partially selected. If you or anyone else can, then I am prepared to reconsider my conclusion.

As it is, I think the only correct interpretation is that pixels can either be selected or not selected; thus, that there is no such thing as a partially selected pixel.

[NotMyFault]

1 hour ago, R C-R said:

I am not inventing the fact that I could not find anything in the Affinity help that explicitly states that a pixel can be partially selected. If you or anyone else can, then I am prepared to reconsider my conclusion.

As it is, I think the only correct interpretation is that pixels can either be selected or not selected; thus, that there is no such thing as a partially selected pixel.

https://affinity.help/photo/English.lproj/pages/Selections/selections_fromlayers.html

Case closed.
 

[walt.farrell]

1 hour ago, NotMyFault said:

Case closed.

That Help was already referenced, I think, in this thread. Where does that say that the pixels are partially selected? It says the areas are partially selected.

[NotMyFault]

This seems to be the greatest example of ultimate cross talking.

Pixels are selected if the opacity is greater than 0. But every pixel is bound to a kind of hidden mask layer, which holds its opacity value. The UI hides a lot of these underlying complexities. Selection is almost equivalent to an alpha channel or mask layer.

In essence, the whole discussion is only about names for selections, masks, alpha channels. Hair-splitting at its best.

If it makes you happy, you can say pixel are not partially selected - but then you must agree that every pixel has a shadow channel/layer containing the alpha value, which could be partially white or below 1. That is what 99,999% call a partially selected pixel. Language is used differently by different people and depends on context.

Time for my 🛏💤

 

[R C-R]

38 minutes ago, walt.farrell said:

That Help was already referenced, I think, in this thread. Where does that say that the pixels are partially selected? It says the areas are partially selected.

It was & it does not say anywhere in it that pixels are partially selected, just that areas are.

The case is not closed.

[R C-R]

1 hour ago, NotMyFault said:

Pixels are selected if the opacity is greater than 0.

What about documents with no pixels? I have already posted an example of that.

Please keep in mind that we are talking about the areas shown by "marching ants" selections. We have already established that they are independent of any layer, any group of layers, any masks, & (maybe??) of any alpha channels. They are in no way, almost or otherwise, equivalent to any of those things.

EDIT: as a second example please consider this Just ants 2.afphoto file. No only does it have no layers or pixels, the marching ants selection (defining an area of the document) is not aligned to integer pixel values -- to see that just toggle on the Quick Mask, switch to the Move Tool, & refer to the Transform panel.

It does) have a composite alpha channel, so it is ready to be applied to a new layer, but as it is, it is just an area.

[NotMyFault]

19 minutes ago, R C-R said:

What about documents with no pixels? I have already posted an example of that.

Please keep in mind that we are talking about the areas shown by "marching ants" selections. We have already established that they are independent of any layer, any group of layers, any masks, & of any alpha channels. They are in no way, almost or otherwise, equivalent to any of those things.

The active selection will be applied to the current selected layer. It is kind of a joker, waiting for the user to make his choice. Having an active selection in a document containing no layers proves nothing. It is a hair-splitting edge case with no practical impact. If you want to apply the selection to something, you need to create layers.

 

The selection is a one-of-its kind special document-level alpha channel in the dimension of the document. As soon as you select a layer and an edit action it is used in that context. And the selection remains unchanged (not consumed) unless you deselect or change it.

Try this:

• Create a 1x1 GREY/8 document
• add a pixel layer
• edit->fill with 50% gray
• use info panel to check IA values (128/255)
• use quick mask to Edit>fill a selection with 50% grey
• close quick mask
• use edit->cut.
• use info panel to check IA values (128/128)
• see that the pixel will become 50% transparent.

So how do you want to call the 50% transparent selection:

• Partial selection?
• Full selection with 50% transparency?
• other ideas?

It does not matter how you call it, it will work as described (and not as called by any name)

 

[R C-R]

23 minutes ago, NotMyFault said:

If you want to apply the selection to something, you need to create layers.

Yes, you do. But that in no way changes the fact that document pixels cannot actually be partially selected. Areas of the document can be partially selected, just like the help says, but it is not the same thing.

EDIT: Sorry, I hit save before I answered your question. My answer is it is a full pixel selection with 50% transparency.

[Ron P.]

1 minute ago, R C-R said:

Yes, you do. But that in no way changes the fact that document pixels cannot actually be partially selected. Areas of the document can be partially selected, just like the help says, but it is not the same thing.

I've been reading all the back-and-forth on this. Yes the whole pixel is being affected, but the strength of that pixel is being affected. Look at it as how much of that or those pixels is visible. In affinity, they say that the pixel(s) need >50% will get the ants (marching or otherwise) 😄

So RC-R, you won (sort of), so why persist? What are you trying to prove? GEESH...........

[NotMyFault]

If we we speak of partially select pixels, we do not mean that the a portion of the area the pixels is covering will be selected (getting sub-pixels). Insofar a pixel is the smallest unit of selection.

We mean with “partially” is that the RGBA values (or what ever color format is used in the document) will be used partially according to the transparency level of the mask for that individual pixel.

If you are referencing to “areas”: this “partial” will be applied individually to every pixel contained in the selection, using its individual transparency level defined by the selection. It does not refer to selecting x% of pixels within the selection area.

So, you have a different usage of the term “partially”, that is absolutely ok. But please let the other 99% stay with our definition.
 

[R C-R]

7 hours ago, NotMyFault said:

If we we speak of partially select pixels, we do not mean that the a portion of the area the pixels is covering will be selected (getting sub-pixels). Insofar a pixel is the smallest unit of selection.

I am not sure of the "we" you are referring to but it is evident from various discussions in this & other forums that a lot of people do not understand that a pixel cannot be subdivided into different color/opacity regions, or that "partially selected pixels" does not literally mean that only a part of each pixel is selected.