I believe Sean is referring to the normal ImagesCollide() and ImagesCollide2() functions, not the layer collision detection methods that are new to BlitzMax.
I'm not sure what constitutes transparent or not with regards to image collision detection, but the ImagesCollide2() function has additional parameters to check for collisions between rotated and/or scaled images.
There's no need to remember rotation or scaling with the layer collision methods as the pixels within the layers are set accordingly anyway. Again, not sure how alpha is interpreted using this method.
I'm not sure what constitutes transparent or not with regards to image collision detection, but the ImagesCollide2() function has additional parameters to check for collisions between rotated and/or scaled images.
There's no need to remember rotation or scaling with the layer collision methods as the pixels within the layers are set accordingly anyway. Again, not sure how alpha is interpreted using this method.