How to Edit the Alpha Channel

[telemax]

Hi, @NotMyFault

I watched the video. It seems that the first steps are superfluous?

 

[lepr]

On 2/21/2021 at 8:05 PM, telemax said:

Hi, @NotMyFault

I watched the video. It seems that the first steps are superfluous?

 

 

Yes, and after creating a greyscale layer from the original layer's alpha, the original layer should have had its alpha filled via that menu.

Then, instead of using a Procedural Texture or Channel Mixer to map intensity to alpha, I prefer to simply drag down the left-hand node of the Source Layer Ranges curve in the blend options of the greyscale layer.

(Also, there was an unnecessary complication in the formula of the Procedural texture. There was no need to average its R, G and B channels since all must be equal in a greyscale layer, and so any one the channels could have been mapped to alpha.)

 

Anyway, these workarounds are ridiculous when compared to the straightforward editing of channels that should have been in the app from the beginning.

 

[NotMyFault]

- deleted -

Edited February 21, 2021 by NotMyFault
premature post - deleted

[crgreen]

On 2/15/2021 at 5:07 AM, MEB said:

Hi @crgreen,
Welcome to Affinity Forums
To do what you want select the greyscale image/layer then go to menu Layer > Rasterise to Mask.

This is a good technique; it's a lot like an "Adjustment Layer" in After Effects. But you're wrong in saying it is "what I want", because it effects all layers below it (as Adjustment Layers do). Again, it's good to know, but it is not what I wanted, which was having a mask layer work on one layer only.

Also, the "Mask to below" command isn't exactly intuitive (I'm sure it does something wonderfully useful). If I have the greyscale layer above the rgb layer and use "Mask to below", I would expect it to use this layer as the mask for the layer below, but no, it doesn't work that way. I dont' think "Mask to below" is a good name for this command.

[EDIT]

The missing step to get this to work as I wanted is to then take the now-rasterized-to-mask layer (made as you suggested) and drag it onto the right side of the rgb layer's icon in the layers panel to nest it as the layer's mask.

This is good.

[EDIT 2]

It occurs to me that a mask layer acts not just like an Adjustment Layer, but also (and perhaps more) like a "clipping mask" in Illustrator.

Edited February 22, 2021 by crgreen
add'l info about needed step for technique

[MEB]

Hi @crgreen,
Yes, the command i outlined was just to create the mask, since from your steps you seem already aware how to nest it to a layer (Mask to Below). And like adjustments and live filters if affects all layers below if not nested to a specific one.

Regarding the Mask to Below command specifically, it's not working for you because the greyscale layer you have placed over the image is a regular fully opaque pixel layer (not a mask at this point) which will be used as a clipping mask instead (basically a rectangle the size of the canvas) to clip the layer below which remains fully visible because it matches/fulfils the clipped area (both have the canvas size). If you use the command Rasterise to Mask on the greyscale layer (to convert it to a mask first which will affect all layers below) and then use the command Mask to Below on it, the converted mask will then be nested to the image, used as a mask and affect only that layer (notice the different icons in the thumbnails when its being used as a clipping mask and when used as a mask).

Notice this command (Mask to Below) can also be used with other types of layers - for example with vector shapes which will be used as clipping masks as well. To use a pixel layer as a clipping mask make sure parts of the pixel layer are transparent, otherwise seems it does nothing because the whole layer will act as a clipping mask rectangle the size of the canvas leaving everything visible - this is what (i believe) confused you.

[Sandrooo]

22 hours ago, NotMyFault said:

Hi,

i made a quick tutorial how to edit the alpha channel as pixel layer, using any pixel tool you want to use, including copy & paste.

There are some constraints:

1. Works for pixel layers, may not work for other layers types like vector shapes, text, etc
2. Workflow is not yet non-destructive.
3. May not work if original layer is deeply nested in other layers or groups.

Looking foreward for your feedback.

Currently investiganting to extend this idea into a non-destructive workflow, and to non-pixel layerds like vector shapes.

 

 

2021-02-21_19-36-46_edit_alpha_channel.mkv 52.18 MB · 8 downloads

That has nothing to do with what we are talking about here  just had a very very quick look and will watch it later fully.... besides literally in Photoshop and GIMP all you do is COPY > PASTE  - no one said anything about non destructive, we are talking about channels, if I pack texture in a channel di per se is non destructive as if I want to update a channel I just re-open that file and paste an up to date version of that texture in there (for example a metal map or roughness, which are just B&W images, that' why is so easy with GIMP and Photoshop, exception done with Alpha as if using PNG the best way is by SUPER PNG free plugin for photoshop), moreover I can easily manipulate said channels as any other channel with a brush if I want to modify something in detail, again this is already there in the software just serif does not allow you do do so. Not sure as I did not use Affinity much due of the limitation but with photoshop you can easily do manipulation and selection with channels in seconds without having to spend 3 minutes per step, again this is just a comment not based on experience as I found no use to even try it further, not even this second time with free trial.

Alpha should be the same, after all it is just another channel, there is no need to do the mess that Affinity does for a simple channel manipulation, that is just that, a channel... I appreciate the different size of the two companies and how Serif is "new" (the fact that it was funded in the "900'" means nothing as Affinity is what made them relevant to the public) however as someone stated they keep chasing the white rabbit, a software is never finished and they keep not adding features "because we need to fix this or fix that". We are in 2021 we know what markets generate billions, there is not even need to do market research really, two simple modification would allow to reach a billions dollars market instead of doing 50% discount and selling an otherwise good software for £50 like if it is a "Chinese rip off", most functions are just there but totally overlooked.

It is one off payment, once you sold X copies you have saturated people with interest in it, no company offers software 50% off if it sells like hotcake... fact is that for anyone doing anything but photo editing it is pretty much a bad purchase, still, as you still have to pay Adobe (or use Gimp if you are not a studio but indie), as many have stated above these are "minor" changes as there are lots of open source addons for Alpha / DDS however no one can force them to do something, not even 15k views of this own topic, we had many examples in history of tech of company disappearing or being relegated to the corner for arrogance in business.

I mean if you are a wedding photographer or the very few left professional photographers doing only that then fine, but as stated above nowadays you get people do more than one thing even more so companies doing multimedia, basic for surviving in a competitive market that includes developing countries doing work for half the price, that dictate one has to be competitive.

For the rest this is good for hobbyist photographer, I am one of them for example, yet I do also some game development as "side job" for a well renown platform and despite being a side job I get my addon published by a known company in this platform, so despite I like the software I still use Adobe as it covers both my uses and I assume for other people is the same, this without taking into account the actual pro who wrote and visualized this thread who need these feature as must and for a precise scope.

 

 

 

[kirkt]

On 2/20/2021 at 12:42 PM, Sandrooo said:

Came back a year later, downloaded the trial and see as bad as before, still no use for gaming, 3D (VFX too) or GUI so except photographer, if pure photographers still exist as nowadays you have to offer more services if you want to be more than a hobbyist and are keen in putting food on the table, the software still pretty much a £25 quid worth, if one wants to pay for the extra few features it offers compared to GIMP that is.

Besides this alpha channel topic, packing in general is absolutely tedious and remind me of the process it took to do some simple folders navigation and file manipulation in DOS compared with drag and drop of the last few decades with windows. Even this was never addressed with a simple COPY AND PASTE in the already existing channels, one has to do some weird steps and over 30 different clicks to simply allow to have a B&W maps in channels which is already there and already editable, without even talking about the alpha itself as one can question it has to be added, the RGB is already there. 

Let's say your images that you want to pack exist as separate grayscale images (in this example, let's assume you have an AO pass, a ROUGH pass and a METAL pass that exist as grayscale images - pretty typical for a CG render workflow).  You can use the Apply Image operation to pack the ROUGH and METAL grayscale images into the G and B channels of an existing basis document that you make from your grayscale AO grayscale image, as you described in a previous quoted post.

1) Open the grayscale AO pass in AP and convert it to an RGB8 image, so it supports R, G and B channels.  This will be the new WORKING DOCUMENT for your channel packing exercise.

2) Use Filters > Apply Image and select "Load Source From File ..." and select the ROUGH grayscale image.

3) in the Apply Image dialog, check the "Equations" box and enter the following equations:

DR = DR

DG = SG

DB = DB

DA = DA

where "D" denotes the Destination of the operation, and "S" represents the Source of the operation.  In this case, this will place the GREEN channel from the ROUGH image (the "Source" or S in the Equations) in the Destination Green (the DG) and leave the rest of the channels in the Destination document as they were originally.

4) Do the same as Step 3, only use the METAL image as the Source and enter:

DR = DR

DG = DG

DB = SB

DA = DA

This places the METAL image's blue channel into the blue channel of the destination working document.

5) Save As (for an .aphoto doc) or Export (for a JPEG, TIFF, etc) the resulting file and your are done.

It does not require 30 steps and is pretty straightforward.  You do not even need to open the other files (ROUGH and METAL) explicitly into AP' interface to select them and incorporate them into the G and B channels of the AO doc to pack them.  QED.

You could avoid having to open the ROUGH and METAL documents explicitly in PS to to pack them into the AO-based RGB doc's G and B channels by using .... you guessed it - Apply Image!  Same operation as described above for AP, same amount of steps.  Sometimes you have to change your routine slightly to adapt your longstanding workflow to new tools.

Have fun!

 

Kirk

 

 

[Sandrooo]

7 minutes ago, kirkt said:

Let's say your images that you want to pack exist as separate grayscale images (in this example, let's assume you have an AO pass, a ROUGH pass and a METAL pass that exist as grayscale images - pretty typical for a CG render workflow).  You can use the Apply Image operation to pack the ROUGH and METAL grayscale images into the G and B channels of an existing basis document that you make from your grayscale AO grayscale image, as you described in a previous quoted post.

1) Open the grayscale AO pass in AP and convert it to an RGB8 image, so it supports R, G and B channels.  This will be the new WORKING DOCUMENT for your channel packing exercise.

2) Use Filters > Apply Image and select "Load Source From File ..." and select the ROUGH grayscale image.

3) in the Apply Image dialog, check the "Equations" box and enter the following equations:

DR = DR

DG = SG

DB = DB

DA = DA

where "D" denotes the Destination of the operation, and "S" represents the Source of the operation.  In this case, this will place the GREEN channel from the ROUGH image (the "Source" or S in the Equations) in the Destination Green (the DG) and leave the rest of the channels in the Destination document as they were originally.

4) Do the same as Step 3, only use the METAL image as the Source and enter:

DR = DR

DG = DG

DB = SB

DA = DA

This places the METAL image's blue channel into the blue channel of the destination working document.

5) Save As (for an .aphoto doc) or Export (for a JPEG, TIFF, etc) the resulting file and your are done.

It does not require 30 steps and is pretty straightforward.  You do not even need to open the other files (ROUGH and METAL) explicitly into AP' interface to select them and copy/paste them to incorporate them into the G and B channels of the AO doc to pack them.  QED.

You could avoid having to open the ROUGH and METAL documents explicitly in PS to to pack them into the AO doc's G and B channels by using .... you guessed it - Apply Image!  Same operation as described above for AP, same amount of steps.  Sometimes you have to change your routine slightly to adapt your longstanding workflow to new tools.

Have fun!

 

Kirk

 

 

No offence Kirk but this is not even viable for me that I do it as a "second job" so I have more time to create my asset and import them for then place them accordingly, let alone people who do this as daily job with busy pipelines.... again photoshop is a simple COPY AND PASTE, we are talking about worse case 10-15 seconds compared to still a cumbersome way in Affinity as many, way more experienced, developers have been saying in this topic and Reddit (if you want to do a quick google search, as Discord channels won't appear in it).

By the time you get to your point three I have done and saved a texture with three maps in it...

No reason in trying to continue this further as many other have said Affinity is too specific for photographers only and is of no use for developers, not even for who like me does bits and pieces (although published) as I prefer to use an app that can do both, I see no point in bloating my hard drive with something that can barely do something SO simple, it is a shame as I am very ethical when choosing software (or food really) and I would rather give my money to a smaller company, but hey...as said...their product, if they are happy by selling at £25 quid like in a street market that's totally up to them, I do not code specifically so not sure if they lack of knowledge to make such as simple implementation (considering the tools are already there) as perhaps isn't that simple, so cannot judge.

As the long list of people here and online have said this software is just a disappointment, the company is willingly holding it back and sure enough Adobe is happy, they know money is with studios and gaming business, generating more money than Hollywood as we know, that's why they could not care less of offering anything more convenient, than the current plans, they have no reason too, what's the point in wasting resources to go after a niche (because let's face it, small photographer are a niche).

[kirkt]

@Sandrooo - It is not 30 steps and it takes 15 seconds tops.

No offense taken, if your workflow is that brittle, then it sounds like AP is not a viable solution for you.

best,

kirk

[roger rosa]

On 1/23/2021 at 8:15 AM, opermaks said:

Not “just one”, but “just one to expand the community”. Understand? Just one to allow a bunch of people even to star work with the product. Just one that will bring new money for development not at the expense of discounts. 
 

That shows why you're still writing. As if you want to overcome someone without delving into the essence of his Message. How nice it is to accuse someone of negativity and try to heroically put it in its place, taking the words out of context. If this is not the case, then do not write. I described my problem to be heard, not to be silenced. And I haven't seen a positive response from the developers yet. If I missed it somewhere, please share the link with me. I'm ready to wait my turn in peace. But in the meantime, I have to ask to let me and other artists into this queue officially. And please don’t imagine how I demand freedom for artists with a machine gun and a can of gasoline. If you communicated with me looking me in the eye, then you would not have any doubts about my adequacy.

Happy photoshopping, bro.

 

17 minutes ago, kirkt said:

Let's say your images that you want to pack exist as separate grayscale images (in this example, let's assume you have an AO pass, a ROUGH pass and a METAL pass that exist as grayscale images - pretty typical for a CG render workflow).  You can use the Apply Image operation to pack the ROUGH and METAL grayscale images into the G and B channels of an existing basis document that you make from your grayscale AO grayscale image, as you described in a previous quoted post.

1) Open the grayscale AO pass in AP and convert it to an RGB8 image, so it supports R, G and B channels.  This will be the new WORKING DOCUMENT for your channel packing exercise.

2) Use Filters > Apply Image and select "Load Source From File ..." and select the ROUGH grayscale image.

3) in the Apply Image dialog, check the "Equations" box and enter the following equations:

DR = DR

DG = SG

DB = DB

DA = DA

where "D" denotes the Destination of the operation, and "S" represents the Source of the operation.  In this case, this will place the GREEN channel from the ROUGH image (the "Source" or S in the Equations) in the Destination Green (the DG) and leave the rest of the channels in the Destination document as they were originally.

4) Do the same as Step 3, only use the METAL image as the Source and enter:

DR = DR

DG = DG

DB = SB

DA = DA

This places the METAL image's blue channel into the blue channel of the destination working document.

5) Save As (for an .aphoto doc) or Export (for a JPEG, TIFF, etc) the resulting file and your are done.

It does not require 30 steps and is pretty straightforward.  You do not even need to open the other files (ROUGH and METAL) explicitly into AP' interface to select them and incorporate them into the G and B channels of the AO doc to pack them.  QED.

You could avoid having to open the ROUGH and METAL documents explicitly in PS to to pack them into the AO-based RGB doc's G and B channels by using .... you guessed it - Apply Image!  Same operation as described above for AP, same amount of steps.  Sometimes you have to change your routine slightly to adapt your longstanding workflow to new tools.

Have fun!

 

Kirk

 

 

I would ask also, Can you add a 4 th grayscale channel to pack in the alpha channel of that image and save it out ?
So far this does not work as packing since Affinity just pre multiplies (is that the word) the RGB with the alpha instead of keeping each channel separate.
 

[opermaks]

It is obvious that the people who are trying to challenge the essence of the problem, in fact, are not even familiar with the pipeline of the gaming and media industry. It is not clear why they even try to answer then. First, it would be worth understanding that there are artists in this world whose life is inextricably linked with certain technical requirements. The result of our work is a complex multi-layered material, where the information in each channel is stored. Stored even in the alpha channel, and not destroyed. And when editing channels, the user does not have any questions about editing them. If you select a channel, you can edit it. If you select another one, you edit another one. And that's it! Don't need some space lessons on image programming using C++. At the end, the entire RGBA is packed in any format that the alpha channel supports. When opening this file, the user should see the information in all channels as it was before closing. Including the alpha channel. And the alpha channel should not cut out information from other channels leaving white pixels. The materials needed for every pixel. That's right, every pixel is important without exceptions. All the information is needed to be read by game/3D engine or other CG software. That's why we need all the information in the file. This is not a wedding photo. What else could be unclear?

[NotMyFault]

Hi @opermaks,

thank you for your friendly comment. I just clicked "unfollow" to this thread and will leave you kids alone.

 

[opermaks]

Better unregister on this forum at all and uninstall OS. Do it really in an adult way

[kirkt]

On 2/22/2021 at 3:15 PM, roger rosa said:

 

I would ask also, Can you add a 4 th grayscale channel to pack in the alpha channel of that image and save it out ?
So far this does not work as packing since Affinity just pre multiplies (is that the word) the RGB with the alpha instead of keeping each channel separate.
 

I believe this is where AP falls apart, because it does not preserve the alpha channel as a separate channel without burning its effect into the RGB channels.  Ideally, one would be able to pack the alpha as well and leave it intact as its own channel, instead of burning its effect into the color channels.

The exception is exporting your packed document as EXR, which may not be a viable option for a gamma-encoded texture workflow, for example.  Just because it can be done, does not mean it is an efficient and useful workaround for a particular workflow.  

Kirk

[telemax]

3 hours ago, kirkt said:

Ideally, one would be able to pack the alpha as well and leave it intact as its own channel, instead of burning its effect into the color channels.

I found that AP keeps the RGB channels intact if a Mask Layer is used instead of the Alpha Channel. But the mask elements must extend beyond the canvas, or have a black background, all over the canvas.
Test file Mask_to_Alpha.afphoto

[roger rosa]

Definitely, I don't think it is an issue of AP itself, it has to do with AP file format Reader/writer for targa or Tiff.  Since other application support preserving the alpha and RGB channels separately.
I am no programmer but I would not think that this is such a major work to rectify, a save as Packed / or preserver Alpha tick on export ?  

[MattyWS]

On 2/23/2021 at 7:23 AM, opermaks said:

It is obvious that the people who are trying to challenge the essence of the problem, in fact, are not even familiar with the pipeline of the gaming and media industry. It is not clear why they even try to answer then. First, it would be worth understanding that there are artists in this world whose life is inextricably linked with certain technical requirements. The result of our work is a complex multi-layered material, where the information in each channel is stored. Stored even in the alpha channel, and not destroyed. And when editing channels, the user does not have any questions about editing them. If you select a channel, you can edit it. If you select another one, you edit another one. And that's it! Don't need some space lessons on image programming using C++. At the end, the entire RGBA is packed in any format that the alpha channel supports. When opening this file, the user should see the information in all channels as it was before closing. Including the alpha channel. And the alpha channel should not cut out information from other channels leaving white pixels. The materials needed for every pixel. That's right, every pixel is important without exceptions. All the information is needed to be read by game/3D engine or other CG software. That's why we need all the information in the file. This is not a wedding photo. What else could be unclear?

Man, I have been getting that a lot on this forum... We ask for the methods of our workflow to be better or just possible, explaining exactly why it currently isn't viable in AP and get remarks from other forum goers like "just do this 50 step thing using these 5 other applications, whats the problem?". The problem is always quite simple, the solution is always clear... yet people on this forum seem addicted to resisting feedback on behalf of the devs and it washes out the discussion.

If people dont agree (understand) with feedback, requests or constructive criticism of the topic, don't bother coming into the topic just to disregard it.

[opermaks]

Yes, the discussions on this forum are conducted in a truly strange way

[Jowday]

On 2/26/2021 at 12:17 PM, MattyWS said:

Man, I have been getting that a lot on this forum... We ask for the methods of our workflow to be better or just possible, explaining exactly why it currently isn't viable in AP and get remarks from other forum goers like "just do this 50 step thing using these 5 other applications, whats the problem?". The problem is always quite simple, the solution is always clear... yet people on this forum seem addicted to resisting feedback on behalf of the devs and it washes out the discussion.

If people dont agree (understand) with feedback, requests or constructive criticism of the topic, don't bother coming into the topic just to disregard it.

Bingo. Sad but true.

On the positive side I found the price tag on my professional software quite acceptable now. And love that professional software even more.

[Christoph Werner]

On 2/26/2021 at 12:17 PM, MattyWS said:

Man, I have been getting that a lot on this forum... We ask for the methods of our workflow to be better or just possible, explaining exactly why it currently isn't viable in AP and get remarks from other forum goers like "just do this 50 step thing using these 5 other applications, whats the problem?". The problem is always quite simple, the solution is always clear... yet people on this forum seem addicted to resisting feedback on behalf of the devs and it washes out the discussion.

If people dont agree (understand) with feedback, requests or constructive criticism of the topic, don't bother coming into the topic just to disregard it.

Completely agree. Editing alpha channels/masks in Affinity Photo is still a no go, if you come from Photoshop.

Sorry for this, but why the hell I can't just edit an alpha channel like a regular pixel channel?! And I don't mean to paint something with a brush in the alpha/mask.
I mean functions like changing color levels, curves or adding blur... Things that works like a charm in Adobe Photoshop.

Don't misunderstand me: I like Affinity Photo a lot and try to work in this tools as much as possible. But alpha editing is still one thing that really su...s. Or I didn't understand until today, how Affnitys solution is better. What solution it ever is, to change levels in alpha channels in some seconds.

😪

[dmstraker]

Wow. Long conversation. I won't pretend to understand some of the discussions.

However something I haven't seen here is mention of the Erase blend mode.

A Pixel layer with Erase blend mode seems to work the same as a Mask layer, with Mask white=Pixel layer with  transparent, and Mask black=Pixel layer opaque. Non-pixel layers work in the same way. Coupled for example with Blend Ranges and inserted into a group (to limit the erase layer extent), it seems this could be of use at least in some applications.

Am I barking up the wrong tree (or just barking)?

Edit: Here's video:

 

[crgreen]

On 3/9/2021 at 1:08 PM, dmstraker said:

Wow. Long conversation. I won't pretend to understand some of the discussions.

 

That is some SUPERB barking. Great info. Thanks! ("ta" in your native tongue 🙂 )

[dmstraker]

Just now, crgreen said:

Tha is some SUPERB barking. Great info. Thanks! ("ta" in your native tongue 🙂 )

You're welcome. Ta works. Also Diolch (Welsh).

[Jops]

I observe this topic for years now and also for me (working as a 3d Artist) the way alpha masks are edited is a kill argument. having Photo installed but basically never used besides panorama work. It is just sad to see a so capable program laying around on the harddrive because (and I have to put it that way) the deveolpers are to stubborn to implement a simple solution to this (by just make the alpha and chanels a normal work area and enable copy and paste). It remembers me of blender when developers and a fair amount of users said that the right click paradigm (and the UI) is fantastic and the world should adjust to that. then two years later they changed it and boy look how the userbase of blender developed. shure, compared to that, changing alpha handling will make a smaller impression, but there are a hell of a lot of artists out there that mainly do 3d visuals and games.

best regards

Jops

[jamie_c]

UPDATE: I thought the below was correct, but with further testing have found some continuing issues with the Alpha channel bleeding through and modifying the RGB channels using channel packing. I'm still not sure why my first test image worked and the subsequent 9 tests didn't.

 

I've tried to read through all the posts to see if anyone else has a similar method to what I'm doing. I'm on day 3 (maybe 4) of a trial of Affinity Photo and like many of you use the Alpha Channel to pack more information into a texture, I've also been frustrated by painting on the RGB layers and having the values affected by the Alpha channel.

To get around this, I have a Spare Channel as a "Blank Alpha" and another one as my "Final Alpha". I've also been doing the same with the RGB channels and any variations and references for a material. If I need to make any changes in RGB, I load the Blank Alpha into Pixel Alpha of the layer I'm editing, make the changes and prior to save, load Final Alpha to Pixel Alpha. For export it's pretty much the channel packing video, but from my test I'm selecting from a longer list of spare channels, currently 16.

This allows you to edit other channels independently in grayscale without that weird multiply effect based on the alpha, it also allows you to edit RGB normally (switching Blank Alpha with final and back). It's not perfect, but it may work for you as it is working for me.

Keep in mind I've only tested saving to PNG and importing into UE4, I've seen other people mention TIFF, I have not tested this format. Hope it's useful.

 

Jamie

Edited May 16, 2021 by jamie_c
Ongoing testing with the software to find a solution found my original method only worked in one case and hasn't worked since.