The New Method / Substep / Concept Idea Thread

[mark49152]

Hi, today i will be anouncing my new 2x2 method, STYP wich stands for : Skip The Y Perm!

What's so bad about 2x2 Y-perm? It's only 2 moves longer than a T and flows nicely. I'd much rather have that than have to do unoriented CP recognition. Maybe I'm missing something...

 

[TDM]

What's so bad about 2x2 Y-perm? It's only 2 moves longer than a T and flows nicely. I'd much rather have that than have to do unoriented CP recognition. Maybe I'm missing something...

Don't use a T perm for 2x2 then If you compare a Y perm to algs like R' F R F' R U2 R' U R U2 R' and L F' R U2 R' U R U2 R2, Y perms are much slower.

 

[supercavitation]

Comparing to the standard Fridrich(CFOP) OLL/PLL:
avg moves: F2L(6.7*4)+OLL(9.7)+PLL(11.8) = 48.3 HTM
avg moves: F2L(6.7*2)+C3P(~4)+WV(8.07)+ZBLBL(~8.5)+PLL(11.8) = 45.77 HTM

So the number of moves isn't reduced a great amount if at all (I need to make the ZBLBL algs to be sure), though, the algorithm count is reduced a little more. (OLL/PLL = 78 algs; WV/ZBLBL/PLL = 69)

2 things:

Forcing yourself to perform Winter Variation every time actually involves learning 54 algorithms if you can't work out mirrors on the spot, but more importantly, you need summer variation as well unless you want want to wind up performing extra setup moves (R U2 R' or L' U2 L would work, but it increases move count somewhat).

Winter Variation only works once you have the F2L pair setup, so when calculating move count, you can't skip the moves necessary to form the next F2L pair. The insertion of an F2L pair takes 3~4 moves. Let's call it 4 to give the fewest possible moves for setup. That leaves us with another 2.7 moves that need to be added into the move count in order to setup the pair for Winter or Summer Variation. Add that to 45.77, and you get 48.47 moves, or more than standard Fridrich, and based on a quick search through F2L algorithms, the number of moves needed for setup is probably much closer 3.7 (6.7-3) than 2.7.

Someone yell at me if either of those statements was incorrect.

 

[Dane man]

2 things:

Forcing yourself to perform Winter Variation every time actually involves learning 54 algorithms if you can't work out mirrors on the spot, but more importantly, you need summer variation as well unless you want want to wind up performing extra setup moves (R U2 R' or L' U2 L would work, but it increases move count somewhat).

Winter Variation only works once you have the F2L pair setup, so when calculating move count, you can't skip the moves necessary to form the next F2L pair. The insertion of an F2L pair takes 3~4 moves. Let's call it 4 to give the fewest possible moves for setup. That leaves us with another 2.7 moves that need to be added into the move count in order to setup the pair for Winter or Summer Variation. Add that to 45.77, and you get 48.47 moves, or more than standard Fridrich, and based on a quick search through F2L algorithms, the number of moves needed for setup is probably much closer 3.7 (6.7-3) than 2.7.

Someone yell at me if either of those statements was incorrect.

Ah, I see, then my understand of Winter Variation was incorrect. One moment while I modify a few things about my method.

 

[GuRoux]

I've had an interesting thought. After introducing the Diaper Method and ZBLBL, an idea came to me. I need to work on some of the algs, but here is the basic idea.

Skipper F2L (SF2L)

1-While performing F2L, one inserts the first two pairs normally.
2-Insert the corner of the third pair, not worrying about it's edge.
3-Insert the fourth pair using Winter Variation (27 algs).
4-Insert the edge of the third pair using ZBLBL algs (Modified to respect corner orientation, 21 algs. I'm still working on making these to see the avg moves.)
5-You are now left with a 1LLL, being just PLL (21 algs).

Comparing to the standard Fridrich(CFOP) OLL/PLL:
avg moves: F2L(6.7*4)+OLL(9.7)+PLL(11.8) = 48.3 HTM
avg moves: F2L(6.7*2)+C3P(~4)+WV(8.07)+ZBLBL(~8.5)+PLL(11.8) = 45.77 HTM

So the number of moves isn't reduced a great amount if at all (I need to make the ZBLBL algs to be sure), though, the algorithm count is reduced a little more. (OLL/PLL = 78 algs; WV/ZBLBL/PLL = 69)

The executional downsides to this method is the added recognition of inserting the corner of the third pair to prepare for WV (which should actually be very quick, and thus of little consequence), and the standard look-ahead of the last two F2L pairs becomes increasingly complex (not too much though, it just adds looking for 4 piece orientations per pair). The learning con is that it is quite unorthodox, and a good step out of the way of OLL because it replaces it entirely. Other than that, I see this method as being very effective, holding a lot of potential when it comes to speed cubing and even FMC.

What you guys think?

are you including cross in your average count measurements? because 6.7*4 implies you aren't.

 

[TDM]

are you including cross in your average count measurements? because 6.7*4 implies you aren't.

I think he's excluding cross because he is comparing with CFOP, and the cross is the same in both methods. Or he could just have forgotten it, but luckily it doesn't affect what he's trying to say.

 

[Dane man]

My post has been updated with correct modifications to the method.

are you including cross in your average count measurements? because 6.7*4 implies you aren't.

I am excluding the parts of the method that don't affect the move count. For example the cross is done the exact same way. And I really could have excluded PLL, but I wanted to show the count from F2L to finished cube.

 

[mark49152]

Don't use a T perm for 2x2 then If you compare a Y perm to algs like R' F R F' R U2 R' U R U2 R' and L F' R U2 R' U R U2 R2, Y perms are much slower.

Whether you call it a T perm or not, the adjacent swap alg I use is R' U R' F2 R F' R' F2 R2. Is R' U R' F2 R F' U R' F2 R F' R really "much slower" than that? I'll take that over CP recognition any day.

 

[supercavitation]

Ah, I see, then my understand of Winter Variation was incorrect. One moment while I modify a few things about my method.

Not sure if that first one was so clear, don't want to cause misunderstandings. Winter Variation works for R U' R' cases, while the mirror cases work for L' U L cases. Summer Variation works for R U R' cases, and it's mirrors work for L' U' L cases. You can turn a Summer Variation case into a Winter Variation case by doing R U2 R' or L' U2 L, but that increases move count.

 

[Dane man]

Not sure if that first one was so clear, don't want to cause misunderstandings. ... You can turn a Summer Variation case into a Winter Variation case by doing R U2 R' or L' U2 L, but that increases move count.

You were pretty clear, thanks.

Performing the set up moves to do summer or winter variation would add about 3 moves to the total, giving us a move count avg of 48.77 HTM as opposed to CFOP's 48.3 HTM (quite close). Though the alg count increases substantially (54+54+21+21=150), and that is something I can't accept as a well balanced method. Though adding setup moves only for WV would bring up the move count by I think ~5, bringing the move count to 50.77 HTM. But once again the alg count is (54+21+21=96).

 

[TDM]

Whether you call it a T perm or not, the adjacent swap alg I use is R' U R' F2 R F' R' F2 R2. Is R' U R' F2 R F' U R' F2 R F' R really "much slower" than that? I'll take that over CP recognition any day.

I didn't know about that Y perm. It is short... but I really don't like it. Diag swaps are my slowest PBL and adj swaps are one of my fastest (after diag-diag and adj-diag), but I suppose everyone has different algs and prefers different cases. For me avoided diagswaps are good, but it won't be as beneficial for you.

 

[supercavitation]

You were pretty clear, thanks.

Performing the set up moves to do summer or winter variation would add about 3 moves to the total, giving us a move count avg of 48.77 HTM as opposed to CFOP's 48.3 HTM (quite close). Though the alg count increases substantially (54+54+21+21=150), and that is something I can't accept as a well balanced method. Though adding setup moves only for WV would bring up the move count by I think ~5, bringing the move count to 50.77 HTM. But once again the alg count is (54+21+21=96).

If you assume that people can work out mirrors on the fly (I just tried a couple of the cases, it's not that difficult), it brings the total number of algorithms down to 69 again, though at the cost of the slightly higher move count.

 

[Dane man]

If you assume that people can work out mirrors on the fly (I just tried a couple of the cases, it's not that difficult), it brings the total number of algorithms down to 69 again, though at the cost of the slightly higher move count.

True. I simply assume that some mirrors may be harder to do on the fly than others, and in order to get up to a speed that it'd be worthwhile they'd have to either be practiced so much that they become habit, or memorized individually. Some might prefer non-mirror algs for the mirror cases for comfortability in fingertricking. And if I included mirrors in my alg count, I would have to also include them for the OLL cases for a fair comparison. For these reasons, I usually do not exclude mirrors in my alg count. Raw alg count is how I prefer to compare and contrast method power and efficiency in method.

 

[elrog]

I don't see you including algorithms for F2L. I know that a few of the F2L cases (for me atleast) are more algorithmic than some of the WV cases.

Without ever learing winter variation, I do R U R', R U2 R', R U R' U' R U' R', and R' F2 R F R' F2 L F' R F L' (J-perm), L' U R U' L R', and possibly a few more that I can't remember at the moment.

Some of the F2L cases I had to learn to do efficiently with algorithms include: R U' R2' U R2 U R2' U' R, (R' F R F')3 or (R U R' U')3, and R U' M' U2 M2' U' R' U M2 U2 M. I still don't have a good alg for that last one.

 

[mark49152]

R U' M' U2 M2' U' R' U M2 U2 M. I still don't have a good alg for that last one.

Wow, all that to avoid a rotation?! .

I often just solve it intuitively with a rotation but the best alg I know is R2 U2 F R2 F' U2 R' U R'. Start with thumb on top and go R2'. Pull F' with right ring finger

 

[TDM]

Some of the F2L cases I had to learn to do efficiently with algorithms include: R U' R2' U R2 U R2' U' R, (R' F R F')3 or (R U R' U')3, and R U' M' U2 M2' U' R' U M2 U2 M. I still don't have a good alg for that last one.

I use mark's alg for the last case, and [U2] R2 U2' R2' U' R2 U' R2' for the triple sexy/sledge case. It preserves all pieces' orientations, like triple sledge.

 

[TheNextFeliks]

I had a new F2L idea I tried a few solves It might actually work. I don't know if it's been done before but:
CMLBCLL method:
Step 1. Solve cross or x cross if you like.
Step 2. Solve Three of the second layer edges. This goes quick since there is usually only 3 moves for each. This can be combined with step one if you have good look ahead.
Step 3. Use the empty F2L slot and D or D' turns to solve 3 of the bottom layer corners
Step 4. Solve the last F2L pair. Or you can use this to do ZB.
Step 5. Last layer. However you like.

I think it would be something like:
Cross: 6 moves
3x middle layer edges: 10 moves
3x bottom layer corners: 17 moves
F2L : 7 moves
LL: I'll go with 21 moves.
This would be 61, so it's not very efficient... But it's and idea.
.
.
.
.
.
I have a ZZ oh method I like, where you grasp your Eoline with the thumb and middle finger and do L' with ring finger, L with ring finger or pinkie, R and R' with index finger, U with index finger, and U' with pinkie.
It's probably been done before, but it does work good. This is for people who are very strong right handed and can't do good left handed OH. It allows use of the right hand without having trouble doing R turns.

The first method is [wiki]Keyhole[/wiki]

 

[supercavitation]

The first method is [wiki]Keyhole[/wiki]

And the second one is just EOLine on Right (Normally done EOLine on Left), but slightly misoriented so you're still doing <R, U, L>.

 

[elrog]

It should be noted that you can also do keyhole by solving only 3 corners and turning the E layer as you insert the E layer edges.

Also, thanks for the algorithms guys.

 

[Dane man]

I had a new F2L idea I tried a few solves It might actually work. I don't know if it's been done before but:
CMLBCLL method:
Step 1. Solve cross or x cross if you like.
Step 2. Solve Three of the second layer edges. This goes quick since there is usually only 3 moves for each. This can be combined with step one if you have good look ahead.
Step 3. Use the empty F2L slot and D or D' turns to solve 3 of the bottom layer corners
Step 4. Solve the last F2L pair. Or you can use this to do ZB.
Step 5. Last layer. However you like.

That's a very interesting idea. Some are calling it "Keyhole" though I don't think that is correct because keyhole is for inserting edges not corners. This method is for inserting corners.

The cross will be normal.

Each edge would be about 3-5 moves (9-15 total).

The corners though would be more complicated depending on where the corners are. For example, say one of the corners are in the cross layer already, but it is in the wrong place and mis-oriented. Each corner of that type will take anywhere up to 9 moves to put into place using the open slot, much more than simply doing an F2L pair. This means that corners have a range of 3-9 moves, and therefore, combined edge and corner move count is between 6-14 per.

One way you can modify this issue is to place the LL corners into the cross layer when you put the three edges in. This is done by grabbing any LL corner in any orientation and inserting it with the edge piece as if it were an F2L pair. Because orientation doesn't matter for the corner, the edge insertions can stay around 3-7 moves each. Now that all of the cross layer corners are in the last layer, they can easily be inserted using the slot with between 3-7 very fingertrick friendly moves. But this still results in a combined range of 6-14, but the chances of getting closer to 6 is greater.

Very interesting thought though. I think it's kinda cool to try out. Well done.

PS: 500th! (in this thread)