By simply making those parts black, you solve that problem. make them gray, and youll get some yellow on top of the transparency
Here is a side by side example of using a mask, and using the straight .png with transparency, as I said before , using the straight .png with alpha causes issues, especially in teh darker colored areas…
Left is with masks, right is using the straight image
I got it! You have problems with the PNG because what you call “black” isn’t black but blue. When you use it as factor, well, it’s just not zero. I punched some alpha holes in the dirty JPEG that I found in this thread, I replaced my biohazard symbol and… It just works.
I made the glass full white too… but that was a bad idea. This render really needs a lot more samples.
As I explained in the other thread about alpha and PNG, you can see the alpha channel as a greyscale image, just like the mask that you insist on doing by hand. When you plug directly your color into a factor socket, Blender actually converts it into greyscale… And for the node tree, it makes no difference if the factor comes from the alpha channel of the image or from another image. It’s just more efficient to use the alpha channel. Everything is in one place, no conversion is needed and it consumes less memory since you probably saved your mask as a full 24-bit image… or worse.
How about a node tree…and teh image you used.
I still maintain the mask method is the best way to do it…
I dont know why its such an issue, it takes just a couple minutes to make one.
Ah…forgot about the alpha output of the image node. It essentially creates the mask for you.
My whole point is that the use of a transparency node was completely unnecessary.
I made a video where I have used a three transparent png-images on top of each other. At the end of the video there is a node setup for the material. Using a transparent png is easy, only UV-maps are a bit challenging at times. You need to play a little bit with them to get it. Basically the main thing is that, when you unwrap one or more faces, all the faces get a UV-map. Only the size is zero or rather one pixel. If that one pixel is colored, then all your “non-unwraped” faces will get that color.
It looks sort of like it acts like a ‘Multiply’. Set it to white, it multiplies by 1, meaning no difference (thus transparent). Set it to red, it multiplies red channel by 1 and green and blue by 0, so it just colours whatever is behind it red.
Yeah, using the png alpha output is simpler, faster and cleaner. It makes me laugh looking at my “alternate” node solution now. Live and learn. Here’s my current corona bottle. Thanks everone.
