Pixels and Images and Alphas, Oh My!

[Ldina]

As a long time Photoshop user, I've come a long way to understanding how to work in AP over the last few months, but some things remain a bit confusing and unclear to me. I'm sure I'm not alone. The distinction between images, pixels and alpha is one of them. Exactly what are they and what makes them different? For me, it raises questions about masks, adjustment layers, filters, etc. Is a mask made up of pixels, or is it just pure alpha. Same with adjustment layers, live filter layers, etc. Is an "Image layer" pixels? 

Trying to understand the Channels Panel has brought up a lot of questions. I used channels extensively in Photoshop and was able to easily tweak these channels for blending, masking and other operations. In Affinity, it appears a mask is different from a grayscale pixel layer and needs to be treated differently. How to edit a mask as a pixel layer, so I can apply ANY filter or adjustment, is a bit confusing and seems somewhat complex. I have made progress for sure, but I'd like a more in depth understanding of these differences.

I've watched most of the tutorials I can find, and I've done some searching, but haven't found clearly presented information that has examples to clarify these differences. I'm hoping Affinity can create a video tutorial that addresses some of these questions (or point me to something I haven't found in my searches). Some of the issues I'd like to understand include...

• The distinction between image layers, pixel layers and alpha. What is each made up of and how are they different?
• Exactly what are masks, adjustment layers, live filter layers, etc...(pixels, alpha, etc)
• Is there a way to apply ANY adjustment, filter or effect to a mask? If so, how?
• Why does one sometimes need to 'rasterize' what is already a pixel layer?
• What happens when one "rasterizes to a mask"?
• R,G,B channels are pixels (I assume), but what about the alpha channel?

These are the sort of questions I'd love to see addressed in an in-depth video with plenty of clear examples. It would make the transition easier and help many people get more out of AP more effectively and quickly. At least, it would for me. If such information is already available, I'd love a link. I think this information would be a valuable addition to standard list of AP tutorials. 

Thanks.

[NotMyFault]

5 hours ago, Ldina said:

The distinction between image layers, pixel layers and alpha. What is each made up of and how are they different?

Every pixel, vector, adjustment, filter layer has an alpha channel.

In principle, every layer type which can have colors will always have an alpha channel.

if you use RGB color format, every layer has 4 channels, RGBA, 3 colors and one alpha channel.

Image layers are kind of vector shapes with a fill of type bitmap and a very special case.

The alpha channel is always treated specially and different from color channels.

You can edit, save, restore, convert alpha using special UI elements.

 

[NotMyFault]

5 hours ago, Ldina said:

Exactly what are masks, adjustment layers, live filter layers, etc...(pixels, alpha, etc)

Adjustment layers can modify color information (including alpha ) based the “old” color value of every pixel, and the specific adjustment.

So you can brighten, invert, set to a fixed value, swap color channels etc.

You can’t use color values from pixel at different positions.

Live Filter layers are more powerful. You can modify color information based on the old color of the pixel, and include neighboring pixels within a certain radius (blur, sharpen, distort, …)

 

Mask Layers are alpha values only, no color values.

Adjustment and filter layers have an inherent mask, allowing to remove the layer effect from parts of the document, defined by the mask.

 

[NotMyFault]

5 hours ago, Ldina said:

Is there a way to apply ANY adjustment, filter or effect to a mask? If so, how?

There are some easy options, and more complex.

1. 3 adjustment layers allow to specifically select which color channel (including alpha) are affected: 
   1. channels mixer
   2. levels
   3. curves
2. The procedural texture filters gives deepest control to all channels and alpha
3. other adjustment and filter layers won't let you choose.
   1. Most of them affect alpha, too (blur, sharpen, noise)
   2. Some ignore alpha. 
   3. Some affect only areas which are not transparent. 
4. If you chose “edit mask” you can use almost all pixel tools to edit the mask layer, including (destructive) filters

If you use adjustments or filter layers nested as child layer or in masking position, it can get really complex to realize if the layer will impact the alpha channel or not. This gets even more complex when group layers or their siblings Layer Layer or Symbols come into play, using the default blend mode passthrough.

 

 

[NotMyFault]

5 hours ago, Ldina said:

Is there a way to apply ANY adjustment, filter or effect to a mask? If so, how?

In principle, just add the adjustment / filter layer and either

• nest it in masking position to the mask, or
• If the mask is already nested to a parent layer: put it on top of the mask layer.

please consider that nested mask layers and effects applied to them interact with the parent layer, often in very surprising ways, and some combinations are not really suitable. The main complexity is more in understanding the layer stack structure and how the nesting position affects clipping, masking, and how blend modes play together.

[NotMyFault]

5 hours ago, Ldina said:

Why does one sometimes need to 'rasterize' what is already a pixel layer?

Rasterize will resample the pixel layer to the current DPI of the document, and remove any rotation or sheer / skew, and most important, fractional positions. If you have a pixel layer which has been squeezed or stretched by move tool or transform panel, rasterizing is required to avoid blurriness or other unwanted side effects.

 

[NotMyFault]

6 hours ago, Ldina said:

What happens when one "rasterizes to a mask"?

The pixel layer will be converted into a greyscale layer, and then immediately convert to a mask layer.

As always with mask, black hides, white reveals, shades of grey reveal proportionally to lightness.

[NotMyFault]

6 hours ago, Ldina said:

R,G,B channels are pixels (I assume), but what about the alpha channel?

A pixel has only one single color value, consisting on all 3 RGB color channels combined. Color channels are not equal to pixels. They can be called sub-pixels, but this distinction is only used in specific contexts like sub-pixel anti-aliasing, or Display technology which are not relevant here, and could cause confusion.

The alpha channel is an additional property of pixels: the degree of transparency, similar to glas or transparent film.

Note that most displays do not render transparency. You will never see a pixel on your display which is fully or partially transparent.

Instead, there is always a color like black or white or the grey / white checkered board used to backfill any transparency.

So alpha is the transparency value of pixels in layers, used to calculate the final color as blend or mix of all layers (and the unavoidable backfill color), which you then can see on you display.

[NotMyFault]

6 hours ago, Ldina said:

These are the sort of questions I'd love to see addressed in an in-depth video with plenty of clear examples.

There are countless tutorials already available, some topics covered by official ones from Serif, many more by other creators. The scope of you questions is huge. Some topics can only be answered based on YOUR real life examples, as the combination of all these complexities lead to an exponential explosion which cannot be covered even in thousands of tutorial videos.

If you want to really understand how all this fits together, and works in Affinity, you have to spend about a year of full-time study.

But this is not required at all. A more relaxed option is to ask the forum if you run into an issue, and the community will help you.

[Ldina]

@NotMyFault

Thank you very much for taking the time to write your thoughtful replies. I’ll read them over and try to understand all that you have shared. 

[telemax]

In simple terms, the "image layer" tries to preserve as much information about itself as possible.

 

[lacerto]

In short "Image" layers are basically what Smart Objects are in Photoshop, to give you a more familiar concept:

https://helpx.adobe.com/photoshop/using/create-smart-objects.html

Channel editing is pretty much different than in PS so that just needs to be learned.

[telemax]

1 hour ago, lacerto said:

Channel editing is pretty much different than in PS so that just needs to be learned.

By the way, channel editing is not available for the "image" layer. At least in AP V1.

[NotMyFault]

We can complicate this topic without limits, e.g. introducing more relevant details into the game like anti-aliasing, alpha composting, blend gamma, RGB/32 linear gamma,  and how relevant edge cases are handled. Pixel alignment, dpi mismatch, rounding up/down number of pixels at export are not easy to understand if everything works as intended. Some (hundreds or thousands) implementation bugs make it harder.

I think the only way to understand all these individual topics, and how they relate and combine, needs to be experienced in practice with documents and real-world examples. Video tutorials may give some baseline knowledge. Without practicing immediately and repeatedly, all knowledge will be forgotten.

[Ldina]

I have a lot to digest here and I will be super busy today, but I wanted to send a quick "thanks" for all the responses. 

@telemax I love that non-destructive mask technique using Erase and Blending. I can see that will be very powerful and flexible and it will become a regular part of my tool kit. Thank you.

@lacerto Thanks again for the help. Yes, I now see that "Image Layers" are like PS Smart Objects. That provided some much needed clarity.

@NotMyFault Great material, and I thank you. I agree, learning by doing is the best way to retain info and make it one's own. Actually, what I was really looking for (and I learned some of it in this thread) was exactly what you suggested in your last post here...baseline knowledge about the structure and differences between pixel layers, image layers, masks, alpha channels, adjustment layers and live filter layers, as they have been implemented in AP. Also, how to convert and navigate between these layer types (when and if possible). Many of your explanations deepened my understanding and provided some excellent, useful techniques.

I'll respond again after I have had time to focus on all these responses.

Regards, Lou

[loukash]

18 hours ago, NotMyFault said:

The main complexity is more in understanding the layer stack structure

The Layers panel is the one that I keep open all the time.

[Ldina]

As promised, I wanted to report back now that I've had some time to study and experiment with the previous posts in this topic. Working through the points discussed has grounded me much better in how AP is structured, how these elements work, and more importantly, how I can utilize them in my work. 

Alpha channels are still a bit vague. I understand the concept that alpha is a property that describes transparency, black hides, white reveals, is essential for layer blending, blurs, unsharp masking, soft-edged brushes, etc. But, how to use, modify and manipulate alpha channels is not fully clear to me. Other threads on this forum, (notably a very long one about how to Edit Alpha Channels), suggest I'm not the only one. I used and edited spare alpha channels a lot in Photoshop, and it is VERY different in AP. Perhaps I don't need to understand more, since I've been able to do most of what I need so far, but I like understanding fundamentals, how things are structured and how to use them. I'm not a game developer, don't do 3D renders, etc, so my needs are far less demanding than they might be for others. 

I've learned a lot in this thread and have made significant progress due to all the helpful posts. 

Regards...

[NotMyFault]

The basic question is why do you edit the alpha channel in first place. Probably you want to achieve some kind of effect. If you can explain what effect you are after, we can show a way to achieve the same result - and most often without editing the alpha channel.

Affinity has the key concept of non-destructive editing - meaning to stack lots of filters and adjustment, and put every edit step on a new layer, to allow changing any of these steps later without using undo history. It takes some time to adapt to these workflows.

Directly editing alpha is a kind of no-go. But it requires to learn new workflow techniques more in line with non-destructive editing.

Alpha in Photo is kind of locked down, not a first class citizen like the regular colour channels. So to become friends with Serif Products better don't try to edit alpha directly.

If you really need to edit an alpha channel like a RGB channel:

destructive workflow: simply convert it it via channels panel into a greyscale layer, and back when finishing.

non-destructive workflows are possible, but really require some advanced tricks (workarounds) like linked layers, using blend mode erase etc which could break compatibility to PDF exports.

Again, explain what effect or result you want, and we will find a way.

 

 

 

[Ldina]

@NotMyFault Thanks. Second-class citizen sounds about right. 😝

So far at least, I have found ways to do what I need, but I'll have to do some things differently than I am accustomed to. Working with alpha channels in PS was something I did a lot of. In AP, I've edited grayscale layers for blending, masks, etc, to accomplish some things. I didn't know that Erase blend mode could potentially cause problems with PDF exports, so thanks for that tip. 

Thanks again.