Lab unusual color, have you noticed? Huge difference!

[Aftemplate]

 

The original image. You can try it yourself. (The above comparison comes from my test)

 

[Aftemplate]

@Andy Somerfield    @Patrick Connor

Is it possible to achieve consistent results for this behavior? Thank you.   

[kirkt]

@Aftemplate - To render the AB composite in AP to look like the AB composite in Photoshop, etc, replace the L channel of the composite with 50% gray.  Then enable all of the channels (L, A and B channels) and the rendered result will be identical to PS.

I did this by creating a spare channel filled with 50% gray (called "50% gray spare channel" in the screenshot).  I made a duplicate of the original image (called "Modified (L > 50% gray)" in the screenshot), and made that layer active.  I then right-clicked on the spare channel and selected "Load to Modified (L > 50% gray) Lightness" to load the 50% gray into the duplicate layer's L channel.  Note that in the second screen shot, the L, A and B channels are all enabled for the composite rendering.  Also note that Photoshop does not display color in the individual A and B channels unless you enable "Show channels in color" from the Preferences > Interface window.

You could make a macro that would do all of this automatically.

Kirk

[Aftemplate]

10 minutes ago, kirkt said:

To render the AB composite in AP to look like the AB composite in Photoshop, etc, replace the L channel of the composite with 50% gray.  Then enable the A and B channels and the rendered result will be identical to PS.

Kirk

This is not a real solution. When you have 100 layers, or more complex layers and group structures.
For those who focus on using Lab space, I rightly know your workaround. That's not what I want.
Thank you for your warm reply! (Although it didn't help me)

[kirkt]

The same thing can be achieved for the individual A and B channels - ie., substitute 50% gray into the L as above, and then view the composite rendering of the L and A to see A in color and L and B to see B in color as you posted above from other apps.

Kirk

[kirkt]

@Aftemplate - You can also apply a curve to the top of the image layer stack.  In LAB mode for the Curve, make the Lightness curve flat across at 50% and then view the composite.  It will look identical to all of the other apps.

Kirk

[kirkt]

see attached

 

[Aftemplate]

11 minutes ago, kirkt said:

The same thing can be achieved for the individual A and B channels - ie., substitute 50% gray into the L as above, and then view the composite rendering of the L and A to see A in color and L and B to see B in color as you posted above from other apps.

Kirk

Why not provide a preference? This different behavior has been changed: the original Photo or, Photoshop generic approach.
Adding this preference does not require a lot of source code changes, and can be implemented with just a little effort.
Clearly, this is the best and most direct way to solve the problem.

[kirkt]

In the meantime, add a curve adjustment with 50% flat L curve and you can see colorized A and B channels, and the AB composite, as in PS.  You would have to ask the devs why this is not a preference, or why the AB composite is different than in PS or similar apps.  I'm just trying to solve your problem.

Kirk

[Aftemplate]

10 minutes ago, kirkt said:

@Aftemplate - You can also apply a curve to the top of the image layer stack.  In LAB mode for the Curve, make the Lightness curve flat across at 50% and then view the composite.  It will look identical to all of the other apps.

Kirk

It was futile.

[Curve adjustment layer]
[Layer A (A only) blend mode-X]
[Layer B (L only) blend mode-U]
[Layer C (L+A) blend mode-H]
[Layer D (A+B) blend mode-T]

Your method will not apply.

[Aftemplate]

9 minutes ago, kirkt said:

In the meantime, add a curve adjustment with 50% flat L curve and you can see colorized A and B channels, and the AB composite, as in PS.  You would have to ask the devs why this is not a preference, or why the AB composite is different than in PS or similar apps.  I'm just trying to solve your problem.

Kirk

As I said, I'm good at using everything about Lab, including all your tips for me.
If you want to (help) solve my problem, you should ask the developer to add a new preference or make it consistent with the general rule.
Thank you very much for your kindness! You are welcome to discuss Lab with me.

[Aftemplate]

1 hour ago, kirkt said:

@Aftemplate - To render the AB composite in AP to look like the AB composite in Photoshop, etc, replace the L channel of the composite with 50% gray.  Then enable all of the channels (L, A and B channels) and the rendered result will be identical to PS.

I did this by creating a spare channel filled with 50% gray (called "50% gray spare channel" in the screenshot).  I made a duplicate of the original image (called "Modified (L > 50% gray)" in the screenshot), and made that layer active.  I then right-clicked on the spare channel and selected "Load to Modified (L > 50% gray) Lightness" to load the 50% gray into the duplicate layer's L channel.  Note that in the second screen shot, the L, A and B channels are all enabled for the composite rendering.  Also note that Photoshop does not display color in the individual A and B channels unless you enable "Show channels in color" from the Preferences > Interface window.

You could make a macro that would do all of this automatically.

Kirk

This operation is damaging and irreversible. It's a far cry from nondestructive editing.

[Aftemplate]

@Andy Somerfield

Hey! If you can respond to this, it would be very helpful. Others can't help. You the lead developer of Affinity photo, taking up a little of your time (I know you busy)
Thank you so much! I hope this doesn't bother you.

[kirkt]

Just to clarify, without knowing why you work in Lab and what you are trying to accomplish, it is not straightforward to find a work around the suits your workflow.  I'm not sure why adding a Curves layer with a flat L curve is not working for you, but it should, and it is non-destructive.

Some examples of what you can do non-destructively - let's say you want to isolate the b channel and blend it into the original in Overlay mode to pump up the blue-yellow color contrast.  Add a Curves adjustment layer and make the L and the a curves flat, as above.  This will produce the b channel, in color, as you have shown in your previous post.  Set the Curve adjustment layer to Overlay and you're done.  I am not sure what you are trying to do, but there are probably ways of getting it done that involve simple adjustment layer operations that are non-destructive - the methods are just different than PS.

Kirk

[kirkt]

17 hours ago, Aftemplate said:

It was futile.

[Curve adjustment layer]
[Layer A (A only) blend mode-X]
[Layer B (L only) blend mode-U]
[Layer C (L+A) blend mode-H]
[Layer D (A+B) blend mode-T]

Your method will not apply.

This is meaningless.  Give the devs an example of what you are trying to do and what will help you do it.  

Kirk

[Aftemplate]

8 hours ago, kirkt said:

Just to clarify, without knowing why you work in Lab and what you are trying to accomplish, it is not straightforward to find a work around the suits your workflow.  I'm not sure why adding a Curves layer with a flat L curve is not working for you, but it should, and it is non-destructive.

Some examples of what you can do non-destructively - let's say you want to isolate the b channel and blend it into the original in Overlay mode to pump up the blue-yellow color contrast.  Add a Curves adjustment layer and make the L and the a curves flat, as above.  This will produce the b channel, in color, as you have shown in your previous post.  Set the Curve adjustment layer to Overlay and you're done.  I am not sure what you are trying to do, but there are probably ways of getting it done that involve simple adjustment layer operations that are non-destructive - the methods are just different than PS.

Kirk

When a problem arises, if you want to solve it, you should first find the root cause of the problem.
The root cause of this "L" variance is:
Other programs: When "L" is hidden (removed), the entire "L" channel is set to 128.
Affinity: When hiding (removing) "L", set to 0 for the entire "L" channel (this is the standard way for RGB, but not applicable to Lab)
It has to be said that your answer is excellent.
But like I said, I want to address the root cause, not to get around it. This is the starting point for the topic.
I'm a loyal user of Lab, and if I want to bypass this difference, I don't need to ask for this topic.

Now, just let's wait quietly  @Andy Somerfield  fix this behavior, just change the source code from 0 to 128.
I only use Lab, and I don't use similar programs that don't support Lab.
When this issue is resolved at the source code level, I'll make two recommendations.

[debraspicher]

6 hours ago, Aftemplate said:

When a problem arises, if you want to solve it, you should first find the root cause of the problem.
The root cause of this "L" variance is:
Other programs: When "L" is hidden (removed), the entire "L" channel is set to 128.
Affinity: When hiding (removing) "L", set to 0 for the entire "L" channel (this is the standard way for RGB, but not applicable to Lab)
It has to be said that your answer is excellent.
But like I said, I want to address the root cause, not to get around it. This is the starting point for the topic.
I'm a loyal user of Lab, and if I want to bypass this difference, I don't need to ask for this topic.

Now, just let's wait quietly  @Andy Somerfield  fix this behavior, just change the source code from 0 to 128.
I only use Lab, and I don't use similar programs that don't support Lab.
When this issue is resolved at the source code level, I'll make two recommendations.

Turns out I learned something today. I've heard of LAB but not sure what it means to "work in" LAB. If the colored bits are true, then I agree with you fully, it should be something that is eventually changed in order to be accurate representation.

[Aftemplate]

38 minutes ago, debraspicher said:

Turns out I learned something today. I've heard of LAB but not sure what it means to "work in" LAB. If the colored bits are true, then I agree with you fully, it should be something that is eventually changed in order to be accurate representation.

RGB is machine-friendly, and in essence, it is not tailored for humans. Lab is the opposite.
I use Lab for digital painting, lab can do anything RGB can do (digital painting only)
About Lab's benefits and many of The Inherent Defects of RGB:
https://rawpedia.rawtherapee.com/CIECAM02
Read this and you'll get a good understanding of Lab. (Lab is not complicated)

Colors can be removed and turned to grayscale. Greyscale cannot be removed, the best way is to compromise, not the all-black all-white (most conspicuous)
That's why, in addition to affinity, the above 4 programs, they all use 128 medium gray.
This issue has not been fixed since 2018.

Thank you for your recognition and support! I hope that affinity will become better.
Stay curious.

[Chris B]

Hey Aftemplate,

I have logged this to be fixed by the developers.

@kirkt - thank you for your input. I have picked up a few tips from you—it's very much appreciated.

[Aftemplate]

9 minutes ago, Chris B said:

Hey Aftemplate,

I have logged this to be fixed by the developers.

@kirkt - thank you for your input. I have picked up a few tips from you—it's very much appreciated.

Thank you for documenting this issue.
Would it be convenient for you to tell which developer to resolve it or when it will be resolved? (Optional)
As far as I know,  @Andy Somerfield  is the lead developer of Affinity Photo.
This issue has not been fixed since it was recorded in 2018 (as of 1.8.4beta)

[Aftemplate]

I will continue to track this issue until it is resolved.

In each beta version.

The (only) record is meaningless until it is actually resolved. Since 2018.

   @Chris B   

[Chris B]

I have already spoken with Andy. He asked me to report back to you and say we'd like to fix this. I couldn't find a log of it (from 2018) so as far as I know, I am the only one who logged it. 

[Aftemplate]

16 minutes ago, Chris B said:

I have already spoken with Andy. He asked me to report back to you and say we'd like to fix this. I couldn't find a log of it (from 2018) so as far as I know, I am the only one who logged it. 

Please be sure to convey my thanks to Andy for me.
Your answer is the most meaningful of the entire response.
That's what I want.
Affinity is the best I've ever seen, far more than Photoshop.
Photo x Designer will have unlimited possibilities.

[Aftemplate]

Fixed in Beta (1.8.4.184) and Beta - 1.8.4.681(Windows)