Orient First method

[badmephisto]

Lars I think your diagram in table #2, L2 U L' is wrong. The bottom left edge on that diagram should face down and not left.

 

[LarsN]

Lars I think your diagram in table #2, L2 U L' is wrong. The bottom left edge on that diagram should face down and not left.

You're right. I thought I had fixed it, but somehow the old diagram showed up again. I'm not very good at this webstuff

 

[dChan]

Lars, do you think that if a sub-15 average cuber took up your method, he/she would be able to get sub-20 with it in a reasonable amount of time?

 

[pete]

I didn't find any problems with the examples and I've run them through twice. Are you sure that you are doing the right M moves. In my descriptions M follows L.

Try doing it on a solved cube.

i'm sorry, doing this on a solved cube reveals that i'm a retard
i didn't align those 4 edges correctly, i simply assumed all 4 incorrect edges
need to be placed in the top layer.

your examples actually do work correctly, i was referring to the working example under step 3 where you say "If you do a B2...." then it continues "...U move to bring the green edge piece to the B-face, then B2..." followed by "...and finally D2 to align...." and ends with "....and using only L2 R2 and U-face adjustments"

i didn't read it correctly, as it appears the first "B2" move must be ignored,
then it works fine.

as for you actual examples 1 & 2, they work fine although they seem to be very simple examples as in my actual solves I never get such easy cases.

in example 2 :
"z' U M U' - inserted one edge and corrected orientation of another"
this is not explained in your step 2 so I don't understand why I need to
make such a sequence (although it works).
there are surely other scenarios that you have such short solutions for
but it is not explained under step 2 of your tutorial.


having said that, following your 2 examples was very helpful and now I do understand your method in general although there are some minor details
still unexplained.

my biggest confusion initially was in first part of Step 2, because you present 3 Cases, but there are in fact many other cases not mentioned.
for example what if after Step 1 I end up with 3 edges already placed in middle layer and some of those edges are not correctly orientated ?
I understand now that such cases can be eventually converted to any of the other 3 Cases but it is not so obvious to someone who tries this method for the first time, and you probably do know many shortcuts on how to convert
any unusual scenario into any of your 3 Cases but those steps are not exactly
obvious to me yet.

anyway, this is a great alternative method so i'll keep on practising it little more. my first few solves were taking way too long, I average about 2min 30sec, is this way too slow ? (I average 45 secs with my current method after one month of practising - doing cross, keyhole 2nd layer and 3-look LL)

cheers
pete

 

[LarsN]

Lars, do you think that if a sub-15 average cuber took up your method, he/she would be able to get sub-20 with it in a reasonable amount of time?

This is why I hoped that some of the very fast cubers might want to give it a try. It's possible that a more experienced cuber could quickly find the problems with this method. I can detect problems with look-ahead and cuberotations, but I do not now if these problems can be overcome by practise or optimizations.

as for you actual examples 1 & 2, they work fine although they seem to be very simple examples as in my actual solves I never get such easy cases.

in example 2 :
"z' U M U' - inserted one edge and corrected orientation of another"
this is not explained in your step 2 so I don't understand why I need to
make such a sequence (although it works).
there are surely other scenarios that you have such short solutions for
but it is not explained under step 2 of your tutorial.

I made this example early on, then decided that the description of step 2 needed simplification. There are ofcourse other cases than the 3 described. In the description I try to apeal to setting up a case of first having 2 middle layer edges inserted, and then there should only be 3 cases left. I plan to describe some of these scenarios in the advanced section.

my biggest confusion initially was in first part of Step 2, because you present 3 Cases, but there are in fact many other cases not mentioned.
for example what if after Step 1 I end up with 3 edges already placed in middle layer and some of those edges are not correctly orientated ?
I understand now that such cases can be eventually converted to any of the other 3 Cases but it is not so obvious to someone who tries this method for the first time, and you probably do know many shortcuts on how to convert
any unusual scenario into any of your 3 Cases but those steps are not exactly
obvious to me yet.

I recommend that you try work out some of the special cases on your own. As mentioned above I will write about these cases in my tutorial, but trying to work out your own solutions will give you a better understanding of step 2 and you might find some better solutions than I did

anyway, this is a great alternative method so i'll keep on practising it little more. my first few solves were taking way too long, I average about 2min 30sec, is this way too slow ? (I average 45 secs with my current method after one month of practising - doing cross, keyhole 2nd layer and 3-look LL)

cheers
pete

It's hard for me to tell you if you are too slow, since this is not really a tested method. I average 20 secs with full friedrich, my times for Orient First steps are like this:

Step 1: 2 secs
Step 2: 10-20 secs (depending on the case and If I were able to predict the case during inspection)
Step 3: 15-30 secs (much to long, lookahead is very hard in this step.
Step 4: 3 secs (same as my friedrich ofcourse)

 

[pete]

I recommend that you try work out some of the special cases on your own

if I describe your step 2 as two different steps (which I think they are anyway)
such as that orientating middle layer edges is Step 2a and orientating U/D edges is
Step 2b then I discovered that sometimes it is an advantage to execute
4-edge flip (M' U M') from Step 2b before Step 2a.

I execute M' U M' before Step 2a every time I get Case 3 and in some situations
even when I get Case 2 in Step 2a because I don't like those two Cases,
they are longer and cumbersome in comparison to the simplicity of Case 1.

In such situations I execute M' U M' in such a way that I not only end up
with Case 1 for Step 2a but most importantly I leave myself with either
0 or 4 U/D edges to flip for Step 2b because those are by far the easiest
and fastest.

here is an example, scramble with white on left and yellow on right
and blue on top) :
U M2 U z M' U M' z' L

that scramble should leave you with Case 3 in Step 2a (on my cube the Green/Orange edge is the odd one out), I would then do it like this :

z y D' (setup for 4-edge flip where my bad edge Green/Orange is one of the four edges to be flipped)
M' U M' (execute 4-edge flip)
y' z' L R (align last two edges for Case 1 of Step 2a)
U M2 U (execute Case 1)

I got a skip for Step 2b because I planned for it (the move D' to align down face ensured that I get step 2b skip,
otherwise I would have to do long 2-edge flip in step 2b) and I can now continue with Step 3.

note, you cannot always get Step 2b skip but if I can't get 2b skip then at least
I set myself for simple 4-edge flip (M' U M'). I don't know the probabilities but I know I can always avoid 6-edge and 2-edge flip (those are complicated) and
I estimate from my test solves that I get 60-70% of the time 4-edge flip,
less than 5% of the time 8-edge flip (which is easy anyway) and the rest is
0 edges to flip (Step 2b skip).

If I executed Case 3 in Step 2a with your algorithm l2 U' M' U' l2 U M2 U then I would not have control of what case I get for Step 2b (it could be the complicated 6-edge or 2-edge flip).

cheers

 

[LarsN]

You're right Pete, it might as well be two steps for step 2. I think initially I had an idea that you could do both 2a and 2b simultaniously, but I got so stuck on doing them seperately now.

It's a really good observation you did in your example. I've gotten used to doing case 3 fast, but case 1 is still much faster. It might actually be a good idea to always do a 4-flip after step 1. You can look ahead while doing the flip to find the middle edges. I'll give it a try...

 

[dChan]

Maybe Erik would be the best candidate t otest this out? I assume he already knows how to do the corner orientation step as he explains it on his own site and learning step 2 isn't that hard to do.

Anyway, I have temporarily stopped learning Nielsson because I am trying to get some good times for other things(BLD, 4x4x4, etc.) but I am learning it little by little. It is mostly the corner orienation that has got me stumped. I am trying to find logical ways of remembers certain cases but it is a little hard to do. I know about 70 to 80 percent of the cases(sometimes I remember and sometimes I forget) so I just need to log in a little more learning hours and I should be up to speed after a few hundred practice solves. I promise that by San Diego 2008(May 24th) I will know the Nielsson method inside and out.

I really want to see those advanced tricks though, lol.

 

[pete]

It is mostly the corner orienation that has got me stumped. I am trying to find logical ways of remembers certain cases but it is a little hard to do.

I think it's a waste of time to learn those corners (Step 1) as the first thing.
I'm leaving it as last step to learn. I build all corners by intuition at the moment,
sometimes with help of some simple CLL algorithms. It only costs me extra
6 moves (2-3 seconds) on average. There are more important steps to learn first
that improve average times by a huge margin. I think case recognition in step 2
is far more important in saving time (I'm trying to do the whole step in single flow
by doing some M' U M' pre-moves when I see it would simplify second part of step 2).
Practising step 3 is also going to cut your times down. (certainly more than those
3 seconds you would save by learning 16 algs for step 1).

I have changed stickers on my cube - basically I'm now using only 3 colours
(2 oposite white, 2 oposite red & 2 oposite blue). this is great for practising
case recognition all the way to the end of Step 2.

 

[dChan]

Thanks pete. I have actually been thinking of abandoning learning Guimond orientation of the corners because I can usually simplify most cases into a case where I can use a T-OLL algorithm or a six move alg. I'll focus on learning the edges now.

 

[LarsN]

I promise that by San Diego 2008(May 24th) I will know the Nielsson method inside and out.

That's my birthday. Make me proud

I'll hopefully have time to write some of the advanced tricks later today.

Thanks pete. I have actually been thinking of abandoning learning Guimond orientation of the corners because I can usually simplify most cases into a case where I can use a T-OLL algorithm or a six move alg. I'll focus on learning the edges now.

I do recommend learning Guimond orientation at some point. The principal idea was to know the step 1 algs by heart, enabling you to predict the case for step 2 during inspection. I'm close to being able to do that now.

 

[dChan]

Yeah, I know a little more than half of them and the only ones I don't know are the ones at the end of the table(the ones that are longer) and some that don't seem logical(to me) so they are harder to remember. Learning Guimond orientation is actually pretty easy, it is just that I have to get used to the way the algorithms work.

I don't know if I'll be fast enough by SD2008 to use the method in competition but I should be able to show it off to some people. Maybe we can get a bunch of people on the Nielsson method -that would be cool.

Once you write the advanced tricks I'll probably switch back over to full learning mode again. I really want to see those more than anything else, lol.

 

[Harris Chan]

Hmm, about orient-first method; if you want to figure out how fast you can possibly go with this, I suggest you try to do an average with scrambles that are specially generated such that the edges are oriented for F2B2RL and corners are orientated. This might give an idea of how fast you can go (and then you have to consider how fast the orienting edges take).

 

[dChan]

Oh, thats a great idea, Harris. Lars should definitely try this. If the corners can be orientated in 1-2 seconds(I suspect sub-1 is possible) that would definitely give us a great idea of how an accomplished Nielsson cuber would do. Of course, a better suggestion might be that we have you learn the method, Harris! That would be cool.

 

[badmephisto]

Also, Part3 (the F2L) looks like the part that will need most of the practice. Maybe this is plainly obvious, but if you want to practice just that part, you just have to make sure to scramble the cube only using half turns on F,B,R,L... That will set up part3 for you as it keeps everything oriented. So you can only use F2 B2 R2 L2, U U' D D'

 

[niKo]

Corners could definitely be oriented sub-1. (you have the whole inspection period to plan ahead). I can't do it now, but I could time how long each step takes me. I'm looking forward to those advanced tricks.

-niKo

 

[dChan]

He has some up but I suspect there shall be more tricks he has up his sleeve.

Does anyone know a good way to deal with several corners ending up in the D-layer but wrongly permuted during step 3? I always just position a corner over where it should go(over the FR position) and do R2 U R2 U R2 U2 R2 which forces the wrong corner into the U-layer and the right corner into the D-layer but then I have to do this several more times? Is there no other way?

 

[malcolm]

Try R2 U R2 U' R2, much easier alg.

 

[cubeRemi]

I think this method has potential.

what do you all think about this: after orienting all pieces,

-solve bottom corners
-solve bottom edges ( align edges to centers, perform: M' U2 M )
-solve E/M layer edges, this is like in the roux step : http://www.youtube.com/watch?v=afmsuYH5inw

or if parity: use other algs:

switch UB > DF > UB:

F2MR2F2M'R2F2 (7,12)
F2R2MF2M'R2F2 (7,12)
F2MR2F2R2M'F2 (7,12)
F2R2MF2R2M'F2 (7,12)
U2M'R2U2MR2U2 (7,12)
U2R2M'U2MR2U2 (7,12)
U2M'R2U2R2MU2 (7,12)
U2R2M'U2R2MU2 (7,12)
D2M'R2D2MR2D2 (7,12)
D2R2M'D2MR2D2 (7,12)
D2M'R2D2R2MD2 (7,12)
D2R2M'D2R2MD2

or other case: switch UF > UB > UF

U2R2U2R2U2 (5,10)
U2M'R2MU2R2U2 (7,12)
U2MR2M'U2R2U2 (7,12)
U2M2R2M2U2R2U2 (7,14)
M'U2R2U2R2MU2 (7,12)
U2R2U2M'R2MU2 (7,12)

- pll

 

[somerandomkidmike]

I tried to do something similar about half a year ago, but I dropped the idea. Great job developing this method! I think I am going to try practicing it.