The New Method / Substep / Concept Idea Thread

[wir3sandfir3s]

This is a silly idea I was toying with, it's a stupid little method, probbly nothing special, but here it is:

Solve top face

Orient the bottom face (I suggest you get an epll case)

Solve Last edges.

I dunno if this is any good, probably isn't, but I just wanted to share it.

Isn't this just sandwich?

 

[shadowslice e]

Isn't this just sandwich?

It's more like upside down waterman but you orient more edges than the regular version.

 

[Shiv3r]

if were talking about CF methods, Ill bring my modern spin on minh thais CF method:
-Roux FB on R face(I suggest actually making it white/yellow, even though most rouxers don't do that anymore)
-solve the 2 L face corners intuitively
-some kind of Cll/CxLL (I use CMLL) on R face, like in waterman
-keyhole in the R face
-use one of the algorithms posted on rubikscube.info(rotated Z) to insert the last L edge and orient the M edges
-permute M edges

how to translate the algorithms on that page: all E moves are inverted(E'->E, etc.) and made into M moves, all the R moves are turned into U moves(no inversions)

anyway, this is pretty much minh thai's CF method with a few notable differences:
-the U face in minh thai's method is instead of L face, so you have almost a sandwich-like method
-you block build the "U" face(really L face on here) instead of key holing in the edges
-

-advantages:
-faster than minh thai's method
-blockbuilding is more efficient
-while building FB, there are only 2 pieces to track, and after that only 1
-the <M, U, R> moveset is pretty nice
-for roux solvers, there is only one new angst to learn, with about 7 cases.
-may be useful to help a new transition from a beginner CF method to Roux

just a stupid Idea, may be a good practice for routers onlooking FB and practicing CMLL, because those are the first 2 steps.
what do you guys think?

 

[crafto22]

if were talking about CF methods, Ill bring my modern spin on minh thais CF method:
-Roux FB on R face(I suggest actually making it white/yellow, even though most rouxers don't do that anymore)
-solve the 2 L face corners intuitively
-some kind of Cll/CxLL (I use CMLL) on R face, like in waterman
-keyhole in the R face
-use one of the algorithms posted on rubikscube.info(rotated Z) to insert the last L edge and orient the M edges
-permute M edges

how to translate the algorithms on that page: all E moves are inverted(E'->E, etc.) and made into M moves, all the R moves are turned into U moves(no inversions)

anyway, this is pretty much minh thai's CF method with a few notable differences:
-the U face in minh thai's method is instead of L face, so you have almost a sandwich-like method
-you block build the "U" face(really L face on here) instead of key holing in the edges
-

-advantages:
-faster than minh thai's method
-blockbuilding is more efficient
-while building FB, there are only 2 pieces to track, and after that only 1
-the <M, U, R> moveset is pretty nice
-for roux solvers, there is only one new angst to learn, with about 7 cases.
-may be useful to help a new transition from a beginner CF method to Roux

just a stupid Idea, may be a good practice for routers onlooking FB and practicing CMLL, because those are the first 2 steps.
what do you guys think?

That doesn't make sense, how would you solve CMLL on right if that's where you're block is? I think you need to reformulate this because right now I can't tell what you mean, sorry.

 

[Teoidus]

I'm assuming the FB is actually on the left, just like in Roux.

If so, I think it'd be better to keyhole in 3 edges in the R face, and then go straight into LSE.

 

[Shiv3r]

I'm assuming the FB is actually on the left, just like in Roux.

If so, I think it'd be better to keyhole in 3 edges in the R face, and then go straight into LSE.

The last part is just a different version of LSE, where you insert one edge, solve the other edge while solving EO, then permuting the M edges.

I meant FB on left, im sorry.

 

[Teoidus]

I'm aware, I just think normal LSE will be more efficient. This could be good if you found a fast way to insert those last three edges on R

 

[Shiv3r]

I'm aware, I just think normal LSE will be more efficient. This could be good if you found a fast way to insert those last three edges on R

You could track all 3 simultaneously.

 

[Shiv3r]

or, you could figure out some nice slice turn algs so you can do it rotationlessly(without ARFing, explained in next post). maybe even throw in some waterman "solve 2 redged in ring" stuff, too.

 

[Shiv3r]

first full solve with it: 36.72

given the fact that I average around 24-28 seconds, thats decent for the first time trying a method.

also, I think that for LSE it may be faster to not have to worry about them being ARF'ed( adjust r face) correctly so your ulsolved L/R edged for LSE could be at UL(it always will be there, actually) and RF, or UL and RD, or anything like that. The reason I suggested the LSE variant I did is because you can solve any 3 edges on the R face and recog isn't too bad.

 

[JimCube]

i just had a thought, make crosses all around,except the last two edges on the top cross, then do corners algorithms, which should be similar to 2x2 algs because 2x2 are just 3x3 corners right?, then do the last two edges. however, you would have to devise algs that preserve the crosses. lookahead might be good because crosses are relatively easy to look ahead, altho the move count might be high. i call it CCL2E, crosses, corners, last 2 edges. i might work but im certain it will never beat cfop tho

 

[wir3sandfir3s]

i just had a thought, make crosses all around,except the last two edges on the top cross, then do corners algorithms, which should be similar to 2x2 algs because 2x2 are just 3x3 corners right?, then do the last two edges. however, you would have to devise algs that preserve the crosses. lookahead might be good because crosses are relatively easy to look ahead, altho the move count might be high. i call it CCL2E, crosses, corners, last 2 edges. i might work but im certain it will never beat cfop tho

You have just described beginners F2L, accept you solve 2 LL edges during F2L... Or EF minus 2 edges, at the cost of having to do an alg to solve the orientation of the last 2 edges. Think about that - your method is WORSE than EF...

 

[gyroninja]

You have just described beginners F2L, accept you solve 2 LL edges during F2L... Or EF minus 2 edges, at the cost of having to do an alg to solve the orientation of the last 2 edges. Think about that - your method is WORSE than EF...

No, he's just suggested a worse form of phasing.

 

[wir3sandfir3s]

No, he's just suggested a worse form of phasing.

I don't see it, he never described whether the unsolved edges were opposite or adjacent. It is more of a worse EF.

 

[Shiv3r]

you guys heard of TFOP, the beginners method I came up with that JTWong has gotten sub-10 with?

 

[JTWong71]

you guys heard of TFOP, the beginners method I came up with that JTWong has gotten sub-10 with?

That was only a couple solves, and were pretty lucky.
I average around 15-18 Seconds with TFOP.

 

[Shiv3r]

That was only a couple solves, and were pretty lucky.
I average around 15-18 Seconds with TFOP.

compared to over 20 seconds with normal LBL.

 

[crafto22]

No, he's just suggested a worse form of phasing.

What he never mentioned anything about the edges being phased or not, he just said to solve them. How is he suggesting a worse form of phasing?

 

[Teoidus]

Is TFOP that thing where you solve the E slice, and then solve the F2L corners using 3 F2L algorithms? If so I can't imagine that being too bad, and certainly easier to sub-x than beginner's LBL.

 

[Shiv3r]

Is TFOP that thing where you solve the E slice, and then solve the F2L corners using 3 F2L algorithms? If so I can't imagine that being too bad, and certainly easier to sub-x than beginner's LBL.

thats what it was, yes. Seems like a better Idea to teach than LBL.

I also had an Idea where you solve the D layer, solve U corners, then use setup moves and the H-perm(or T-perm if you want to reuse algs) to solve all the rest of the edges. pretty much a mix of LBL and old pochmann(which should totally be a beginners method BTW)


so for the half-roux, half-corners first method, I had an Idea to rapidly keyhole the R edges:
1.pair up RB and RF(orientation and order doesn't matter) and place them at DF and DB.
1.5: if the DF edge goes to RB, then execute an R2
2:using one of 4 algorithms(1.both oriented, 2/3 one flipped, 4 both flipped) to solve the RF and BR edges simultaneously.
3.solve the DR edge