Document the RGBA structure

[rurban]

No description provided.

[larshg]

Shouldn't it be RGBA all places, since the alpha part is always present? I couldn't find a RGB only struct at least?

[rurban]

Shouldn't it be RGBA all places, since the alpha part is always present? I couldn't find a RGB only struct at least?

No they are separate, even if they share the same uint32_t storage. pcl::PointXYZRGB != pcl::PointXYZRGBA

[larshg]

Ahh, but there is only a single pcl::RGB struct - which expose the A channel with default 255 value. There are no pcl::RGBA struct.

But indeed there is PointXYZRGB and PointXYZRGBA.
And what I can see PointXYZRGB only has constructors that accept XYZ and RGB - ie. hides the alpha channel.

I don't see the necessary of having PointXYZRGB, since it still has the same size as PointXYZRGBA - but I don't see us removing it now anyways...?
And pcl::RGB should be pcl::RGBA, since it does contain both the space and expose the alpha channel.

I would vote that the documentation explains this for the pcl::RGB struct though?

What do you think @mvieth ?

[mvieth]

Ahh, but there is only a single pcl::RGB struct - which expose the A channel with default 255 value. There are no pcl::RGBA struct.

But indeed there is PointXYZRGB and PointXYZRGBA. And what I can see PointXYZRGB only has constructors that accept XYZ and RGB - ie. hides the alpha channel.

I don't see the necessary of having PointXYZRGB, since it still has the same size as PointXYZRGBA - but I don't see us removing it now anyways...? And pcl::RGB should be pcl::RGBA, since it does contain both the space and expose the alpha channel.

I would vote that the documentation explains this for the pcl::RGB struct though?

What do you think @mvieth ?

Agreed -- removing or renaming point types is not possible any more, but the documentation should say that the alpha value can be stored.