I have a question: doesn't the sRGB encoding ignore the alpha channel by nature? I don't heard of the difference of alpha between sRGB space and linear space and non-linear alpha encoding.

VK_FORMAT_R8G8B8A8_SRGB specifies a four-component, 32-bit unsigned normalized format that has an 8-bit R component stored with sRGB nonlinear encoding in byte 0, an 8-bit G component stored with sRGB nonlinear encoding in byte 1, an 8-bit B component stored with sRGB nonlinear encoding in byte 2, and an 8-bit A component in byte 3.

Vulkan VK_FORMAT_R8G8B8A8_SRGB doesn't use non-linear alpha, too. I doubt if non-linear alpha has any use cases.

I have a question: doesn't the sRGB encoding ignore the alpha channel by nature? I don't heard of the difference of alpha between sRGB space and linear space and non-linear alpha encoding.

VK_FORMAT_R8G8B8A8_SRGB specifies a four-component, 32-bit unsigned normalized format that has an 8-bit R component stored with sRGB nonlinear encoding in byte 0, an 8-bit G component stored with sRGB nonlinear encoding in byte 1, an 8-bit B component stored with sRGB nonlinear encoding in byte 2, and an 8-bit A component in byte 3.

Vulkan VK_FORMAT_R8G8B8A8_SRGB doesn't use non-linear alpha, too. I doubt if non-linear alpha has any use cases.

You are correct and this PR aims to explicitly state exactly what you described in your comment for those who are unaware of this.

Since there is no other case about the alpha channel, why is it necessary to explicitly state it?

Since there is no other case about the alpha channel, why is it necessary to explicitly state it?

Both Aaron and I were uncertain about what encoding was common or expected for the alpha channel of color data, especially because this Color data structure can hold either linear or nonlinear encoding for the RGB channels. Since both of us had this uncertainty, we expect that other users may similarly be unsure. Adding this explicit statement will help others have clarity on this point.

Since we're here, it might be also good to add something like this to the class description:

Color data may be stored in nonlinear sRGB encoding or linear RGB encoding. Use srgb_to_linear and linear_to_srgb to convert between these two color spaces.

Since we're here, it might be also good to add something like this to the class description:

Color data may be stored in nonlinear sRGB encoding or linear RGB encoding. Use srgb_to_linear and linear_to_srgb to convert between these two color spaces.

After thinking more about it, this text might make a problem with the ambiguity of Color encoding worse. Maybe the following text would be better?

Red, green, and blue values of Color use nonlinear sRGB encoding within Godot, unless stated otherwise by documentation.

If this is ever untrue a documentation issue should be opened.

Thoughts?

(I can make this sort of change a separate PR if you'd prefer.)

After thinking more about it, this text might make a problem with the ambiguity of Color encoding worse. Maybe the following text would be better?

Red, green, and blue values of Color use nonlinear sRGB encoding within Godot, unless stated otherwise by documentation.

Because this somewhat goes beyond the intended scope of this PR, I've created a new PR #104666 that builds on top of this one, adding more information about colour encoding of RGB properties. These two PRs can be merged separately or I can squash mine with @aaronfranke as a coauthor.

Thanks!