Could you evaluate our cube solver?

[OnlineCubeSolver]

Hi everyone,

not speed solvers yet, and maybe a bit too lazy, we created an online solver on http://www.solveyourrubikscube.com . It uses a rather inefficient algorithm at this point but I am planning on improving it. Building the whole program with 3D and all already took a lot of time. It's primary aim is to be used for people who got a cube but don't know how to solve it and don't want to re-sticker it or break it apart.

Please let me know what you think of it, any suggestions are welcome!

 

[MaO]

Uhm I don't think it works very efficient, I did M2U2 and it was loving for like 2 minutes...

 

[blgentry]

The interface is pretty good. Cube rotation could use some work though, as the rotation of the cube "back and forth" is backwards. I.E. moving the mouse to the right moves the cube to the left. It's also possible to get up/down rotations backwards or locked up if you rotate the cube a certain way.

I can't really evaluate the solve itself. I tried twice and both times started doing the solution with "slow". Slow seems to take around 8 seconds or so for every quarter turn of a face. After sitting through 30 seconds of only two face rotations, I pressed "fast", and it became VERY fast. Like a quarter turn in a half second. Cube rotations just as fast.

So, fast is WAY too fast, and slow is painfully slow. Nearly as importantly, there's no way to go back and see the previous step if it was too fast for you.

It's a nice job for a beta. I may just give it one more try.

Brian.

 

[jason9000]

The solver seems to have an issue solving the scramble D.
http://tinyurl.com/ch6qjv
Mind, it got close to the solution by the end.

 

[OnlineCubeSolver]

@MaO: yes it uses a really basic type of solving. for the next versions we will improve the data model and then the algorithms will be much easier to implement!
@blgentry:thanks! agree this improvements shouldn't be too hard to implement
@jason9000: thanks, i should investigate that problem, probably a bug

 

[Edam]

i put in an anti-sune, and after about 2 minutes of solving on fast it looked like this..

 

[blgentry]

Update: I tried again (with a real scramble) and sat through around 100 moves before trying "fast" again, hitting pause in between each move. It was still to fast, so I switched back to slow. But this time, slow only took around 4 seconds per quarter turn, which isn't too bad compared to the earlier 15 seconds per quarter turn (I timed it this time).

However, at around move #123, either I made a mistake or the solver did something funny because my cube, though very close, no longer looked like the one on screen. So, I entered the state of the cube *again*, and did the solution. This time (from a state of F2L solved, plus top cross solved), it worked completely in around 52 moves.

I've noticed that the solver tends to do lots of rotations. Some times more than a full cube rotation. It also seems to do and then immediately undo single face turns. I'm sure it's something about the way it's coded, but it's strange, as it even happened on the very first move of a solve!

..and yes, as you already know, that method could really use some optimization once you have the other bugs worked out. 170+ moves is likely to try the patience most people who are not cube solvers.

Brian.

 

[Lt-UnReaL]

It looks like it's based off of the method that Dan Brown uses.

 

[cookingfat]

It looks like it's based off of the method that Dan Brown uses.

The solution which the solver generates is inspired by a YouTube video by Dan Brown

that's because it is.

I put in a simple H-perm and it took over 200 moves to solve it. dumb.

 

[thelurch1986]

i put in a sune and it solved the cube correctly. the reason it took so many moves it because it starts over.. no matter what you put it in seems to restart lbl from the beginning.

 

[shelley]

http://www.solveyourrubikscube.com/...444444444666366666222222222111111111353333333

holy crap this is the best U perm ever. I swear I just saw it do an F8.

EDIT: haha, this one is even better

 

[Johannes91]

The solution which the solver generates is inspired by a YouTube video by Dan Brown

I don't think human methods are good for this. Thistlethwaite would use a reasonable amount of moves and is easy to implement.

 

[MistArts]

It makes moves on a solved cube...

 

[qqwref]

It doesn't seem to work at all. Did you just set it to do random algorithms?

This may not be much of a cube solver, but it would make a really neat screensaver.

 

[d4m4s74]

how difficult is it to put in an optimal solver, or atleast a better alg set?

 

[felix]

Or look up the Kociemba method.

 

[daeyoungyoon]

http://www.solveyourrubikscube.com/...444444444666366666222222222111111111353333333

holy crap this is the best U perm ever. I swear I just saw it do an F8.

EDIT: haha, this one is even better

AHAHAHAHAHHAHA lol.

 

[byu]

http://www.solveyourrubikscube.com/...444444444666366666222222222111111111353333333

holy crap this is the best U perm ever. I swear I just saw it do an F8.

EDIT: haha, this one is even better

AHAHAHAHAHHAHA lol.

Sort of inefficient, but nice job with the animations, etc. !

 

[chuckster2.0]

It never solved mine and I know of faster solvers so you have work to do.

 

[OnlineCubeSolver]

Thanks all for the replies. Our conclusions are that we got a lot more work to do. We will skip the debugging of the current solver and try to implement Kociemba's method directly and of cause work on the bugs of the user interface some of you mentioned. Thanks!