Try a real Maze

[JuanTutors]

Originally Posted by sonjohan

I tried to see if windows Paint was able to find the solution for me( you know, the fill button)...
Guess what: it didn't really appreciate me doing so..... it crashed.

Heh I did the same thing a few days ago and was writing about it when I accidentally hit the browser back button and erased my entire post!!

I actually had to go through this whole mess just to get all the walkways completely white. (The filler wouldn't work if I didn't do that.)

It turns out almost the entire maze is traversible. There are two main spots that are not, and those two spots happen to be clearly untraversible. Other very small sporadic spots are also untraversible.

I was as careful as possible to make sure the walls of the maze stayed intact when I turned all the walkways white. Actually you can see the corners turned fuzzy, but I don't think any walls of the maze actually broke, i.e. no pixels on the walls turned pure white and let any red through. However, the map is HUGE and I didn't try that hard.

It took my 2.4GB P4 computer over an hour and it filled almost my entire 768MB ram. The time bottleneck was probably the ram, since it's pc100/133.

If anyone wants to post the file PM me and I can email it to you. It's over 5MB in jpg format, though.

[sonjohan]

Originally Posted by sonjohan

I tried to see if windows Paint was able to find the solution for me( you know, the fill button)...
Guess what: it didn't really appreciate me doing so..... it crashed.

OK I continued my search and started to close the gates of the start line.
It seems there are quite a few openings that return to the same spot, which is refused by paint, as it produces a loop. Paint might find the answer; but it's still pretty long, and I wasn't patient enough.

I also started to close the access to the end gate. This allowed me to .... find an answer!

PS: When I say gates: it's every possibility to start/end.
Closing them means making sure that if you use the filler on the lane, you don't accientally fill all other lines too.
You'll end in the fifth lane under the end opening. The image allows you to understand what I meant with closing gates. Three dotts means that they were looping at first. Use the same closing procedure at the end and you'll see that several of these dotts are connected.

Here is a solution based on the Breadth First Search (BFS) algorithm, resulting in a shortest path. The bitmap used is 2072 x 1566, the path has a length of 112116.

[sonjohan]

You ,just took the fun of finding it on your own away.

I'm however delighted to see that the path I was taking was the correct one...

[biwema]

There is actually more than one solution (Because there are any loops). This is just the shortest one.

On the start side you can also take the thirs or fourth path to right, but then you also need to walk around the big square nest to the two coils on the left side.. (Actually the fifth one as you suggested does not work).

[sonjohan]

I never said the fifth line from the start, I said the fifth line from the end.(which is where the shortsest one from the answer ended).
As usual, the answer is the easiest to find starting from the end, rather than from the start.

[THILLIAR]

Quote: "I never said the fifth line from the start, I said the fifth line from the end.(which is where the shortsest one from the answer ended).
As usual, the answer is the easiest to find starting from the end, rather than from the start."

Actually it ends in the middle it was originally two distinct mazes, together designed to be difficult from either end or start if you will.

[biwema]

How did you generate this maze?
I think it is too complex to generate manually in reasonable time and it is too special to be generated with a program (it is kind of art).

[THILLIAR]

AutoCAD