Layer opacity vs. fill

[Smee Again]

6 hours ago, anon2 said:

 

Forget about blend ranges.

The trick is to Normal blend the green layer with white, the neutral colour for Colour Burn, and then Colour Burn blend that intermediate result with the base image. Use Photoshop's fill percentage for the green layer's opacity and use 100% for the intermediate result's opacity.

colour burn fill.afphoto

 

That technique of blending with neutral can be used to simulate a "fill percentage" for 6 of the Special 8 modes:

• Colour Burn and Linear Burn, the neutral colour is white.
• Colour Dodge and Add (Linear Dodge), the neutral colour is black.
• Vivid Light and Linear Light, the neutral colour is mid grey.

 

While not equal to layer fill, it does give some interesting results. Perhaps with some tweaking . . .  maybe it will be serviceable.

Biggest problem with the method is that it is so easy to accidentally add color to the group layer or change the opacity of the color fill layer. Might just be me, started chemo today and some synapses are not firing and all I want to do is sleep . . . but the possibility of a work around has helped me pass the time.

Thanks. Just out of curiosity, how did you figure this out? What made you think to add a neutral color?

BTW, toying around with the neutral color also gives some interesting results, as will playing with the blend mode of the fill color.

[lepr]

5 hours ago, loukash said:

Playing with the HSL method now, a 50% fill apparently equals S:33 L:67.

No, only some colours mixed at 50% with white, black or mid grey will result in a colour with S:33 L:67.

(Speaking of HSL, I previously provided a way to simulate fill percentage for the two darkening and two lightening Special 8 by applying an HSL Adjustment to a colour, but Serif deliberately changed HSL Adjustment a couple of years ago so that it affects L and S of its result when only the L control is used.)

[loukash]

4 hours ago, anon2 said:

only some colours mixed at 50% with white, black or mid grey will result in a colour with S:33 L:67.

Interesting, thanks. My school math skills are definitely too rusty to grasp all of this at once, after being largely dormant for the past 40 years, haha.
So the HSL method is usable if I don't need or don't want to go "by PS Fill numbers", i.e. when I simply want to quickly adjust by what "looks good to me".

4 hours ago, anon2 said:

Serif deliberately changed HSL Adjustment

Haven't toyed much with that one yet, but it appears to me that it works about the same as PS's Hue/Saturation adjustment. The main difference being that Affinity has Recolor as a separate adjustment.

[lepr]

7 hours ago, Smee Again said:

While not equal to layer fill, it does give some interesting results. Perhaps with some tweaking . . .  maybe it will be serviceable.

It works out equal on my machine, so I wonder if you are accidentally doing something different.

Having said that, Affinity currently has a bug in its Vivid Light blending, so that could be giving you unexpected results.

 

7 hours ago, Smee Again said:

What made you think to add a neutral color?

Intuition, I suppose.

[Smee Again]

BTW, instead of a "fill layer" of a solid color, you might try using a "gradient map".  I tried it this morning and worked great on the image I was toying with. You may just want to adjust the opacity of the gradient map a bit. I added a bit of visibility to the gradient map layer, your mileage may vary.

[Smee Again]

Let's try to bump this thread.

Apply boot to post . . . BUMP!!!

[Smee Again]

BUMP!

[Smee Again]

On 4/14/2021 at 6:07 PM, Smee Again said:

Let's try to bump this thread.

Apply boot to post . . . BUMP!!!

 

On 5/11/2021 at 9:05 AM, Smee Again said:

BUMP!

Hmm . . . is anyone at Serif listening? Would be nice to know it is in the works, or even better on its way.

Have friends who, after trying the demo, see the lack of a layer fill slider as enough reason to remain with PS.

[walt.farrell]

8 minutes ago, Smee Again said:

Hmm . . . is anyone at Serif listening? Would be nice to know it is in the works, or even better on its way.

We have been assured that the planners read this section of the forums and use the ideas in their plans.

However, Serif does not generally disclose future implementations until they are basically finished, and appearing (or ready to appear) in a beta release.

[Smee Again]

On 7/1/2021 at 8:44 AM, walt.farrell said:

We have been assured that the planners read this section of the forums and use the ideas in their plans.

However, Serif does not generally disclose future implementations until they are basically finished, and appearing (or ready to appear) in a beta release.

Ah, I should always mark such questions at "RHETORICAL THOUGHT FOLLOWS:" I guess.

RHETORICAL THOUGHT FOLLOWS:

Also, the squeaky wheel that sounds like fingernails on a chalkboard is usually the one that gets greased first.

In real life, all the wheels get lubed at the same time because it's just easier and keeps things quieter longer. (just an old man's musings, no need to read anything into it)

[robto]

BUMP!

[kirk23]

I am pretty sure APhoto devs have no idea what those "fill" things do  to be able to recreate them .   Unless Adobe have published it somewhere.  I doubt it did. 

Still if anyone could recreate it in Aphoto layer stack  as example   my guess it could be helpful. 

  I am 100% sure it could be done since whatever "magic"  Photoshop do  it's just math or rather simple arithmetic on pixel values .    And as such  it still  could be  done  by other means .  Perhaps a hefty stack of layers and "blend if" in some cases .    

Would be cool if we could convert/save  a group/stack  as our own   "blending mode" ,  "live filter"  , "style"  whatever  and keep in asset library .

 

 

 

 

 

 

[Smee Again]

32 minutes ago, kirk23 said:

I am pretty sure APhoto devs have no idea what those "fill" things do  to be able to recreate them .   Unless Adobe have published it somewhere.  I doubt it did. 

Still if anyone could recreate it in Aphoto layer stack  as example   my guess it could be helpful. 

  I am 100% sure it could be done since whatever "magic"  Photoshop do  it's just math or rather simple arithmetic on pixel values .    And as such  it still  could be  done  by other means .  Perhaps a hefty stack of layers and "blend if" in some cases .    

Would be cool if we could convert/save  a group/stack  as our own   "blending mode" ,  "live filter"  , "style"  whatever  and keep in asset library .

 

 

 

 

 

 

Yep, there's a workaround "kind of". Gives better results than just fill opacity, but layer fill it isn't.

[telemax]

 

[NotMyFault]

I saw this video and was afraid to post it here - after one month of ceasefire 😂

[Max P]

On 3/9/2021 at 3:13 PM, loukash said:

I haven't found any explanation – as in: a mathematical formula – online yet. Everybody knows that the "Special 8" are there, "can" do this'n'that and they love it, but nobody tells you how exactly it works.

But since it's all really just math:
How about adjusting it in APh with a Procedural Texture live filter? Has anybody tried that?
I'm no mathematician whatosever, so this is way beyond my expertize. But I noticed that @NotMyFault has recently experimented with Procedural Textures and seems to understand how it all works.

 

Yees

We can use procedural texture for this

using vec4() for vector with 4 parameters

here vec4(Rx,Gx,Bx, Opacity)

Rx,Gx,Bx, Opacity, are define in [0.0,   1.0 ] For me no sens using 8 bit (Fill Layer is 8 bit!)

 

First  Step we define

vec4 c1 = vec4(0.0, 0.0, 0.0, 1.0); //Black

vec4 c2 = vec4(1.0, 0.0, 0.0, 1.0); //Red

vec4 c3 = vec4(1.0, 1.0, 0.0, 1.0); //Yellow

vec4 c4 = vec4(0.0, 0.0, 1.0, 1.0); //Blue

vec4 c6 = vec4(1.0, 1.0, 1.0, 1.0); //white

 

we want an interpolation slider between White and Black

var c1=vec4(0.0, 0.0, 0.0, 1.0);var c6 = vec4(1.0, 1.0, 1.0, 1.0); c6 * s + (1.0 - s) * c1

var for variable

                      c4 * s + (1.0 - s) * c1  // classic use for mixing or use lerp()

 

And Yess, Pixel layer can gives same result, and with 16 bit!,

32 bit is possible but my pc is limited 16 giga Ram , I5 and  and I am not convinced that it is better

 

Rename each layer very important here, Otherwise photo can get lost in cyber space

 

Turner-Shipreck.afphoto

=>

between Black and Blue

var c1=vec4(0.0, 0.0, 0.0, 1.0); var c4 = vec4(0.0, 0.0, 1.0, 1.0); c4 * s + (1.0 - s) * c1

between Black and Red

var c1=vec4(0.0, 0.0, 0.0, 1.0);var c2 = vec4(1.0, 0.0, 0.0, 1.0); c2 * s + (1.0 - s) * c1

 

 

 

Info and explain

https://thebookofshaders.com/06/

https://www.codeproject.com/Articles/492355/Domain-Coloring-Method-on-GPU

https://fr.mathworks.com/matlabcentral/fileexchange/25773-domain-coloring

[NotMyFault]

13 minutes ago, Max P said:

or me no sens using 8 bit (Fill Layer is 8 bit!)

Nice post.

Never the less, i do not understand the comment about fill layer. Fill Layer always uses the Color Depth of the document, so 8,16,32 is possible.

There are some constraints coming from gradients which use unavoidable "forced dithering", and there is an open bug regarding alpha blending of fill layers (with partial alpha)

[mrakonni]

Hi all, I'm a bit late to the party, but there is a simple workaround for most blending modes that might help you.
The "additive" ones foremost (add, color dodge, the ones that do the glares and the rays and the glowsies :))

This is what it looks like:

TLDR:

- Make a new layer and fill it with 100% black color. On top of that layer put any layer you want to blend.
- Put both of them in a group and assign the blend mode you want to the group, not the layers in it. 
- If the mode is color dodge, for example, you will notice the black layer disappear. That's good!
- Now adjust the opacity of the layer you put above the black layer (not the group) and... voila!
- Don't touch the background (black) layer or group opacity, they need to stay 100% all the time. (Unless you want to experiment later)

For other blending modes u need white (eg - color burn), 50% gray (vivid light) or none at all. 
Play with them, put a background (the bottom most layer) in a group and fill it with either black, white, 50% gray,
or turn the background off completely. Pick a mode for the group and if the background turns transparent - that's the one u need. 
If you realize you do not need a background - you don't need the group, use the layer alone. 

Notes:

The "background layer" does not need to cover the whole document. It needs to cover as much as the 
layer above it, so you can crop it later. If you merge (rasterize) the group it will blend properly, but 
if you change the opacity again it will break.  I leave them in a group, personally, because you can paint in
details or add layer effects to the layer you want to blend and they blend better.  If you add more layers in this group,
note that they will first blend with each other and then the background, and all of them together under the mode of 
the group - so feel free to experiment. 

The long read:

Coming from someone who can't math to save his life: the problem is the way they are doing color math. 
Not saying they are doing it wrong, it's just that their approach is different because (I assume) they 
are using the GPU to do the work, and also the way their graphics engine works. I do game art mostly 
and I came across this in older game engines. We had glare and sun ray (gradients) assets that where 
transparent PNGs and they blended poorly. To blend properly you need a "full" pixel, not a "see through" one. 
If the pixel is transparent the math goes byebye. Photoshop's fill slider (and layer effects, too) somehow blend 
as if the pixel is not transparent, and yet it changes transparency (I know, but someone that does OpenGL dev, 
for example, can explain this better than I ). 😄

There is a "take transparency data and +black or +white or plus 50% gray" thingie to add to the blend mode code
somewhere and it should behave properly. Yes, say hi to "designer math", haha. 😄

Adobe broke blending modes in Photoshop a few years ago and the same workaround worked there, until they added 
the "Transparency shapes layer" checkbox in layer options, and that helped a lot in certain scenarios. What that actually does,
I have no idea, but it blends with a punch.

Sorry for the long post, hope it helps!
Cheers!

[mso1977]

17 minutes ago, mrakonni said:

Hi all, I'm a bit late to the party, but there is a simple workaround for most blending modes that might help you.
The "additive" ones foremost (add, color dodge, the ones that do the glares and the rays and the glowsies :))

This is what it looks like:...

Wow, thanks mrakonni, this really works. I have tried a few things myself, but obviously not that.

[Smee Again]

On 2/10/2021 at 11:10 AM, hanshab said:

Serif has acknowledged to me via the support site that the layer fill does not exist.  I feel that is absolutely needed 

Recent digging into the software has revealed to me that the basic idea / structure is not only already in place, but in use in an obscure place:  It is in the adjustment layers and labeled as "Lens Filter". The filter has a slider named "Optical Density" and it behaves like layer fill, but cannot be accessed unless the "restrictive" adjustment layer is used. This only will allow for single color fill, no gradients or gradient maps will work.

Take a look at that adjustment layer and see what you think.

[Max P]

Smee Again i see your post , thank , if i understand you correctly

For the Fill layer ( or Curve or vectoriel shape )

 it is  possible to use

the Color>

• Grayscale slider  more precise than color wheel
• Or better for me Lab L 50 A 0 B0,  and move L plus or minus
• HSL also

To color layer  use these sliders with the appropriate mode, eventually add a mask to circumscribe the effect on the right area 

No problem for a gradient

I don't know if it does the job,  I have no PS

 

 

[Smee Again]

None of those relate to layer fill. They give you different ways to achieve a certain color but . . . not one of them relates to the amount of color (not to be confused with opacity unless a labrador retreiver is the exact same thing as an oak tree) you see (fill).

Try to think of it as if you were a painter: a "wash" puts less pigment on the canvass, bit does not reduce the actual opacity of the pigment. It is still as it was before the wash, there are only fewer molecules of pigment, the molecules of pigment retain their individual opacity. Basically, you have reduced the population of molecules of pigment, but each molecule maintains its original opacity.

[Max P]

Okay
I understand Lavis
 one method for this dilute a color before using it use only one color
 that will be diluted to obtain different intensities of color
 
 We have a differential of appreciation in the term FILL, and an unsaid

For me
Fill Layers when  Need to add a layer that is completely filled with a solid color, gradient or pattern

 
 Go on youtube

play with opacity colour overlay 
    I retained also that we could use semi-transparent colors on AFFINITY SWATCHES pantone Xxx
    easy
    
  Last video Colour overlay AFFINITY solid color  
 

 

[Smee Again]

L A Y E R   O P A C I T Y  W I L L   N E V E R   B E   T H E   S A M E   A S   L A Y E R   F I L L ! ! !

Until you can wrap your head around that, the only thing you can do (as Serif did by using incorrect nomenclature) is muddy the water as the above post has done. Incomplete sentences that leave the reader hanging (more than one). Who or what is/are "Lavis"?

Layer fill affects the amount of color --- NOT THE ALPHA VALUE OF THE COLOR --- on that layer, and only that layer. When used in conjunction with any of the 8 special blend modes, the way it alters the color is changed by raising or lowering the amount of fill.

This cannot EVER be duplicated by adjusting the alpha (opacity) of the layer as it doesn't change the layer's FILL (quantity, not alpha).

Opacity only changes the alpha of the layer. Blend-if only adjusts the alpha of portions of the image.

ONLY LAYER FILL does what I am speaking about.

 

[Max P]

Painting Wash Lavis is the equivalent  key word water color wash

I seem to have misunderstood your previous comment

for this one

Understand , but not convinced me

 

Well I see that the tone rises also I will leave this thread

thanks to all

because I have made a lot of progress thanks to this thread