Mouse click detection for overlapping sprites

By Trebor777 Date 2012-04-29 15:16

One acronym : AABB

By erisdiscord Date 2012-04-30 00:23

I think what OP wants is hit testing that takes transparency into account. I assume they're getting hits on the "wrong" sprite (i.e., not the one it looks like they're clicking) when the transparent region of one is overlapping a non-transparent region of another.

OP: Once you've found your candidates by the usual bounding box method, you'll need to figure out some other way of querying the transparent regions. One way would be to build a bitmask based on the alpha channel and test that. You can get the raw image data by calling Image#to_blob.

By erisdiscord Date 2012-04-30 00:38 Edited 2012-04-30 00:52

Here's a magic incantation that returns an array of boolean values for each pixel. true means the pixel has an alpha of over 50%.

mask = image.to_blob.unpack('L>*').map{|x| (x & 0xff) >= 0x80 }.join(' ')

The array is flat, so index with mask[x + y * image.width]. This is also a very inefficient way to store this information (one pointer to a Ruby object per pixel, plus whatever overhead an Array has), but I'll leave improving on it as an exercise for you. :D

Alternatively, you can use TexPlay, which adds a get_pixel(x, y) method (along with several drawing methods and lots of other nifty image manipulation fun that you probably won't need) and test the transparency that way.

require 'texplay'
class Gosu::Image
  def solid? (x, y); (get_pixel(x, y) & 0xff) >= 0x80; end
end

I'm not sure what the implications are for performance if you do it that way, but it's a lot easier.

By psylem Date 2012-04-30 12:35

Thanks erisdiscord, that looks like what I need.

By erisdiscord Date 2012-04-30 17:41

Yeah, I think I might have got the bitmask wrong though for the Texplay example! I think the color Texplay returns is ARGB, not RGBA, so it would really be get_pixel(x, y) >> 24 & 0xff.