New to texturing, question about alpha channels.

[Distul]

It’s my understanding that image files not to the power of 2, or that don’t have an alpha channel will not be compressed.

The power of 2 thing I get, but I’m not entirely sure what makes an alpha channel. I read that if you gave in PNG 32bit format 8bit of it is the alpha channel, the alpha channel being transparency.

But compression rules also say images with transparency won’t be compressed. So this has left me a little unclear.

When the word transparency is used does it mean images that you can partially see through, or images where there are spots that is not filled by color?

Also, my paper2d textures are stating to have alpha channels when I import them. Something like a circle inside a 256x256 box has an alpha channel, but the edges around the circle have no content so is that not transparency?

[Riley1389]

So, PNG are a bit atpyical from other formats as far as alpha/transparency it involved, at least as far as UE4 goes.

First things first, don’t ever use a 32-bit image for anything besides a Targa with its own dedicated alpha channel, its a waste of resources and there’s no benefit.

Now, with PNGs, you can’t save alpha channels, but you can save transparency.

UE4 should automatically convert the transparency into a dedicated alpha channel during import.

For example, here’s is a Petal texture of mine viewed in Windows 10’s photo app, the black area represents transparent areas:

Now, after I imported it into UE4 with no further alterations and just like any other texture:

And here you can see the dedicated alpha channel UE4 generated based on the image’s transparency information:

As for your question - “When the word transparency is used does it mean images that you can partially see through, or images where there are spots that is not filled by color?” - It can mean both, or just one or the other. It all depends how you material is setup to make use of the image’s information, the most important setting to distinguish these to apart is the “Blend Mode” setting. So, if you are using the “Masked” blend mode, the alpha channel will be used to mask/hide information that is not aligned with white pixels in the mask, but won’t change the opacity of the pixels that are still visible.

The “Translucent” blend mode on the other hand, will adjust the opacity in reference to the information provided by the alpha channel. So in that case it controls both, but in saying that, I recommend “Masked” for anything besides objects that necessarily need to be seen through, like glass and water. For no other reason than “Masked” is much easier to work with and is visibly more appealing in various lighting conditions, but I’m also no master with using the additional options in the “Transluscent” blend mode, as I haven’t really needed to tinker with it too much yet.