Interesting behaviour of masked layer

[DarkClown]

Maybe someone can explain this behaviour to me:
I create an empty document, use a paintbrush to paint anything on it. I add a (pixel)mask to this layer. I choose a black brush and draw a black spot on the mask layer. As expected this spot erases the painting on the layer. Reducing the opacity of the MASK layer now to zero deletes the whole content of the pixel layer - instead of bringing the area of the spot back. Increasing the opacity of the mask layer should from my understanding reduce the effect of the black spot on the layer, but instead it decreases the brightness of the white areas of the mask down to black and empties the pixel layer.

Where's my mistake?

Cheerrs, Timo

[gdenby]

Somewhat surprising. I did some experiments. It appears to me that if the opacity is anything less than 100%, the whole of the pixel layer is made less opaque. When opacity is at 100%, the grey scale of the mask works as expected. If the mask is turned off, the pixel painting is presented as if never masked.

 

My guess is that the intent of masking a pixel layer is to edit the pixels. Then the whole of the layer w. masks can be rasterized to get the desired result.

[DarkClown]

Surprising ... and more than strange ...

[smadell]

I think this is because you are treating the mask as if it were a pixel layer. It is not; it is an alpha (transparency) layer, and only gets represented in black and white (and grey) pixels so that it can be manipulated. I suspect, but do not know, that your opacity setting is affecting the entirety of the layer, and not the mask itself.

[DarkClown]

Hmm, I'm a bit struggeling with how to properly utilize this behaviour since it's not really obvious or understandable for me :-( ... One thing seems to be clear: I can't reduce the mask effect on the layer by increasing opacity ...

[maxen]

Yes, the hole masking thing is definitely a point that affinity should work on. It simply seems impossible to adjust the effect (opacity) of the mask. Instead the mask opacity adjusts the opacity of the layer it's placed upon.

Also, it seems that plain b/w images cant't be used as mask (without "rasterise to mask") in the lastest version of Photo.

I also didn't manage to overlay the mask semitransparent to work in it while seeing the image through it. That in combination is more than annoying.

[carl123]

Add a curves adjustment layer to the mask

 

Switch to Alpha channel

 

Moving left hand side of the curve affects the opacity of the mask

 

 

[maxen]

Thank you for this solution!
Nevertheless the masks opacity slider should do that job...

[R C-R]

I think this is because you are treating the mask as if it were a pixel layer. It is not; it is an alpha (transparency) layer, and only gets represented in black and white (and grey) pixels so that it can be manipulated.

I think this is key to understanding what happens when you change the opacity of a mask. A mask affects the transparency of everything in the layer(s) it is applied to; therefore, if you reduce its opacity, it will reduce the opacity of those layers, regardless of what you 'paint' on it.

 

In other words, changing the mask opacity does not change its 'blackness' because it has no blackness to begin with.

[maxen]

...A mask affects the transparency of everything in the layer(s) it is applied to; therefore, if you reduce its opacity, it will reduce the opacity of those layers, regardless of what you 'paint' on it....

And that's the point. This is also controlled by the masked layers opacity slider.

So actually, these two different sliders do the same. But an opacity slider for a mask should affect the opacity of the mask, not the affected layers one.

[R C-R]

But an opacity slider for a mask should affect the opacity of the mask, not the affected layers one.

It does affect the opacity of the mask. But the opacity of the mask affects the opacity of any & all layers it is applied to. Thus, setting the mask to 0% opacity means any layer(s) it is applied to will also be 0% opaque (100% transparent), & so on for any other values of the mask opacity slider.

 

I don't know why this is so hard to understand. Even though it is represented as one, a mask is not a grayscale layer. By itself, it isn't visible, so there is no reason to expect it to do anything other than affect the opacity of what it is applied to.

[Alfred]

setting the mask to 0% opacity means any layer(s) it is applied to will also be 0% opaque

 

I find that very counter-intuitive! If something is set to 0% opacity, you can't see it, so it's almost as if it isn't actually there. I would therefore expect a mask with 0% opacity to have no effect, rather than having a "0% opacity" effect on the layer(s) to which it is applied.

 

My natural inclination would be to leave the opacity of the mask at 100% and simply use black, white or a shade of grey depending on whether I wanted the affected pixels to be displayed with 0% opacity, 100% opacity or some value in between.

[R C-R]

I find that very counter-intuitive! If something is set to 0% opacity, you can't see it, so it's almost as if it isn't actually there.

That is exactly what happens. But the "something" is not the mask, it is what the mask affects. If you think of it like that, any other behavior would be counter-intuitive.

[Alfred]

That is exactly what happens. But the "something" is not the mask, it is what the mask affects. If you think of it like that, any other behavior would be counter-intuitive.

 

Thanks. I can see that. As I indicated, however, my natural inclination would be to leave the opacity of the mask at 100%, thus avoiding these thorny issues of interpretation!

[DarkClown]

Honestly, this is kind of ridiculous. If I want to reduce the opacity of the pixel layer I'll apply the opacity setting to the pixel layer ... If I want to change the INFLUENCE of the mask (layer) I apply the opacity to the mask layer (same thinking, same state of mind). If the real effect of changing opacity of the mask layer eventually is nothing else but changing the opacity of the pixel layer, why allow masks opacity on the mask layer in the first place at all? In fact it seems extremely illogical - and if by some intellectual interpretation reasonable what so ever - at least inconsequent!

[R C-R]

If the real effect of changing opacity of the mask layer eventually is nothing else but changing the opacity of the pixel layer, why allow masks opacity on the mask layer in the first place at all?

Consider that a mask layer can be applied to more than just a single pixel layer. It can be applied to groups or layers that include images, vector objects, shapes, & so on, in any combination. Its only influence is on the overall opacity/transparency of all of those things in the layer stack.

 

This is not the same thing as applying masks to individual layers, nor is it the same as 'painting' a mask with anything that affects its transparency. It is not a grayscale layer; you are not painting grayscale values onto it, only transparency values. It is a mistake to think of masks like that. That is not what they are & they will not behave as if they were.

[verysame]

This is not the same thing as applying masks to individual layers, nor is it the same as 'painting' a mask with anything that affects its transparency. It is not a grayscale layer; you are not painting grayscale values onto it, only transparency values. It is a mistake to think of masks like that. That is not what they are & they will not behave as if they were.

 

That is not clear to me. Mind to elaborate?

[carl123]

Think of the mask layer as a light that shines on a newspaper in a dark room, reducing the brightness of the light (the opacity) reduces the visibility of the newspaper. 

 

Turning off the light completely (0% opacity) will effectively render the newspaper invisible.

 

It would be nice if there was a simple slider to influence the strength of the mask but that is not how it currently works, maybe this needs a feature request consideration

[maxen]

If I remember correct, in earlier versions it was possible to use b/w-images as mask (as it is in PS). Currently it isn't.
Maybe thats kind of consistency in terms that now only the alpha of the mask (represented by the sadly invisible greyscale values of the mask itself) is taken into account.

At the moment it seems not possible to see what image the mask actually contains without an underlying layer (Only in "edit mask" mode, but then ONLY the mask is visible and you can't paint inside the mask referencing to the masked layer or anything else. Therefore it is very uncomfortable to edit the mask not to mention the missing abilitiy to copy a b/w-image into the mask.

That can't be intended.

Currently you have to "rasterize to mask" an image to use it as mask, and any following editing of the mask is made extremely uncomfortable.

Maybe I just don't see the correct workflow, so I'm open to any tip.

[Alfred]

Honestly, this is kind of ridiculous. If I want to reduce the opacity of the pixel layer I'll apply the opacity setting to the pixel layer ... If I want to change the INFLUENCE of the mask (layer) I apply the opacity to the mask layer (same thinking, same state of mind).

 

That's kind of the way I was thinking when I wrote this:

 

I find that very counter-intuitive! If something is set to 0% opacity, you can't see it, so it's almost as if it isn't actually there. I would therefore expect a mask with 0% opacity to have no effect, rather than having a "0% opacity" effect on the layer(s) to which it is applied.

 

Thinking  about it some more, if it worked like that, then what would an "isn't actually there" (no influence) area of the mask do? A 100% opaque white area on the mask makes the corresponding pixels in the image completely opaque, and a 100% opaque black area on the mask has the opposite effect, so what would a 0% opaque area mean? It would simply look like a hole in the mask, and there would be no way of knowing whether it meant "no white (opaque) influence" or "no black (transparent) influence".

[maxen]

Well, as mentioned: there is no black or white in the mask of the current version. Theres only an alpha channel inside the mask layer. Currently you can't even show it. You have to place a pixel layer or fill layer below it to see what the mask actually does.

By the way: Is there any way to edit a layers alpha channel (preferred by painting on it)?

[R C-R]

I think Carl's analogy could help reduce the confusion about this. Something else that might is to avoid thinking of a mask as a greyscale pixel layer because that is not what it is. A mask maps greyscale (actually lightness) values to opacity values -- when you 'paint' on a mask with a pixel brush you are not actually applying greyscale pixels to it because a mask has no grayness -- if it did you would see that in whatever it masks.

 

The "Edit Mask" function allows you to see the mask's opacity as if it was a greyscale pixel layer, but that is just for visualization purposes.

 

@maxen: You don't have to use the "Edit Mask" function to edit a mask. You can also click on the thumbnail of a mask to select it independently of what it masks & paint on it with a pixel brush. This allows you to see the masked layer(s) so you can reference that as you add or subtract opacity in the same way you would with the "Edit Mask" function.

[maxen]

Yes, I know that. But that you can't see what you are painting, you only see the result. I would need to overlay the mask  as it is done with the quick mask option.

But my main problem is, that I can't use b/w-images as (clipping) mask anymore the way it was possible before.
These converting stuff is more then unnecessary and cumbersome.

I want to copy and paste inside mask layers and I want to paint there an see what I actually do. And I want to see what the mask looks like without going into "edit mask".
But all these problem are only caused by the not longer properly working clipping masks. So I could live with that strange behaviour of alpha masks if the clipping mask would work with b/w-images again.

[R C-R]

Yes, I know that. But that you can't see what you are painting, you only see the result.

??? How is this any different from painting on a pixel layer? If you select the paint brush tool you should see its outline, whether you are painting on a mask or pixel layer. If you move the brush slightly on a pixel layer, you get a preview of its effect, while on a mask layer the preview is always shown. But regardless, you don't see the result until you paint.

 

Also, layer clipping is similar to but not the same thing as layer masking. If you have not already done so, it might be a helpful to read the built-in help topics Layer clipping & Layer masking to become more familiar with the differences. There are also several video tutorials devoted to clipping & masking in section 5 of the In-house Affinity Photo Video Tutorials page that may be helpful.

[MEB]

I'm trying to check the mask opacity layer issue with the dev team.

 

@maxen

Currently it's not possible to edit an existing mask as an overlay, neither copy/mask layers. I advise you to open a new thread in the Feature Requests section for those improvements if they are not there already.

Can you expand a little about the non working clipping masks? As far as i know Affinity never worked with black and white images as masks as Photoshop/InDesign does (1 bit images). You have to convert them first. Is that to what you are referring to? If so i would also add this to the Feature Requests section.