cornerpin / keystoning of data

    May 19 2009 | 7:29 pm
    Its simple but I cant get my head around it: I try to do the classic cornerpin / keystoning of a skewed rectangular (projection). Instead of skewing video I want incoming x,y values to be mapped to the new skewed rectangular. Does that make sense? Can anyone direct me into the right direction? Cheers Torsten

    • May 19 2009 | 8:33 pm
      corner pinning is not exactly simple.
      here is a patch using gl.nurbs to get you started. you could also use gl.mesh. basic idea is to adjust the celss of a 2x2 matrix to move your corners.
      a simple "corner pin" search will turn up a few other techniques.
    • May 19 2009 | 8:37 pm
      Thanks a lot for the quick help, is there any chance to get a patch which works in 4.6? All the best Torsten
    • May 19 2009 | 8:41 pm
      here's the sreenshot. should be all you need.
    • May 19 2009 | 8:55 pm
      thank you, will give it a try.
    • May 20 2009 | 11:34 am
      Thats great, I made the patch and it does exactly what I meant with a video texture. I wonder if there is any way to send a x,y coordinate (position) of a matrix to the and get the altered "nurbed" x,y position. Cheers Torsten
    • May 20 2009 | 12:20 pm
    • May 20 2009 | 1:10 pm
      Cheers, do you have a screenshot or 4.6 patch of that link? Cheers Torsten
    • May 20 2009 | 2:54 pm
      May be somebody can convert it to 4.6 ? Hope it helps!
    • May 20 2009 | 2:55 pm
      another one
    • May 20 2009 | 4:39 pm
      thanks a lot
    • Feb 04 2011 | 11:58 pm
      I have just been experimenting with corner pinning this evening, and would really appreciate it if someone could explain the difference between using nurbs as in this example, compared to a mesh as discussed here:
      I notice that there isn't a need to scale the control matrix bigger than 2x2 as there is with the mesh, so is it simply that nurbs take care of themselves with regards to mapping properly? Is the outcome mapped identically?
      Many thanks
    • Feb 08 2011 | 8:18 pm
      effectively it seems you don't need more than a matrix 2x2 with
    • Feb 13 2011 | 3:06 pm
      Brilliant, thanks for the response :)