I know! It's weird! But it's only transparent pixels, so I can work around it by just making my images opaque apparently. (He didn't have a problem in the early version I sent me because I used a placeholder solid image.)
But that's a sad limitation. I wonder if it does it with 100% opaque textured polygons too. Right now all I use is ones with a solid color as the "background color." And that's there the problem pixels occur.
I tried looking at your code and I have no idea where to start and what determines what returns the default color when the image has no color at that position.
Edit: And now I found an even weirder bug on Windows on my own machine! I hadn't tested my own project in my own computer in Windows until now. But the texture coordinates seem to be flipped on the Y axis on my Windows 7 in Parallels on a MacBook Air. (GPU = Intel HD 4000, passed as Parallels Display Adapter)
So many bugs every time I turn around! lol
OS X: (Correct)
Windows 7 under Parallels: (Incorrect, Y-coordinate origin reversed)
Windows on older machine: The video above (Correct alignment, bad pixel output)
I really wish I had some various older machines, both PC and Mac, with assorted specs and 0.8.x on I can use for testing.