Tuesday, March 4, 2008

BMPGlyph


Do you know what is this? No, it's not some random picture. It's a stereogram, more specificaly it's a random dot, diverging, autostereogram. It's a 2D picture designed to trick the brain into believing that it is a 3D picture if viewed correctly.

There is at least one program for Linux that creates that kind of pictures but I decided to make my own algorithm/program which I called BMPGlyph and works on both Linux and Windows. I uploaded a beta version which you can download by clicking here. Just read the instructions in the zipfile.

I will also try to make an algorithm that extracts the heightmap from an autostereogram. It will be very helpful those people that cannot see autostereograms. :-)

6 comments:

  1. You might consider adding the ability to create inverted magic eye images. I think you just have to shift the pixels differently. Some people can cross their eyes and see the invert image but can't relax their eyes enough to see the normal version (my brother is this way).

    ReplyDelete
  2. Hello patorjk! I Didn't knew you were reading my blog! (I don't think you found it through search)

    You are right about the cross eyed (or converging) stereograms. I prefer the parallel viewing because I find it more natural but as you said there are people that can't look behind the screen/paper so I uploaded a new version which includes this option.

    I also made the algorithms I am describing in the last paragraph of the post. It works quite well but it still shows some pixels with completely wrong depth.

    ReplyDelete
  3. Yep, I got here through a feed reader. I usually just lurk, though I felt like saying something since I'm a fan of autostereogram images.

    I'm not sure the cross-eyed version is working right, though maybe I'm doing something wrong, I'll play with it some more later.

    (note: sorry if this is a duplicate comment, I wasn't sure if my first attempt made it through)

    ReplyDelete
  4. I though it worked OK because I made a correct cross eyed autostereogram. Anyway I'll test it again tomorrow. :-)

    (no it wasn't a double post)

    ReplyDelete
  5. I tried the program again and it seems that it creates correct autostereograms (both cross-eyed and parallel viewing).

    ReplyDelete
  6. Just as a follow up, you are correct, it does work for the cross-eyed versions. I think I just kept slipping into wall-eyed vision when I tried to go cross eyed and see the images.

    ReplyDelete

Popular Posts