The New Method / Substep / Concept Idea Thread

[Cale S]

fun LSLL idea for ZZ or maybe CFOP

example: R U2 F' U' F U' F' R' U' R U' F R' U2

R U' R'
r U R' U' r' F R F'
U' R' U' R' U' R' U R U R U

basically just corner>COLL>L5E
the L5E algs seem pretty nice
not always worth it but for bad F2L cases (like this) I think it might be

 

[Shiv3r]

cool! i

fun LSLL idea for ZZ or maybe CFOP

example: R U2 F' U' F U' F' R' U' R U' F R' U2

R U' R'
r U R' U' r' F R F'
U' R' U' R' U' R' U R U R U

basically just corner>COLL>L5E
the L5E algs seem pretty nice
not always worth it but for bad F2L cases (like this) I think it might be

it actually looks viable.

 

[Shiv3r]

fun LSLL idea for ZZ or maybe CFOP

example: R U2 F' U' F U' F' R' U' R U' F R' U2

R U' R'
r U R' U' r' F R F'
U' R' U' R' U' R' U R U R U

basically just corner>COLL>L5E
the L5E algs seem pretty nice
not always worth it but for bad F2L cases (like this) I think it might be

Im going to try some of these with a petrus Idea me and wong had, Ill share some example solves.

 

[4Chan]

fun LSLL idea for ZZ or maybe CFOP

example: R U2 F' U' F U' F' R' U' R U' F R' U2

R U' R'
r U R' U' r' F R F'
U' R' U' R' U' R' U R U R U

basically just corner>COLL>L5E
the L5E algs seem pretty nice
not always worth it but for bad F2L cases (like this) I think it might be

OHhhHHhh, that'd be sick if the corner was already placed.
This is pretty nice. o.o

 

[Shiv3r]

OHhhHHhh, that'd be sick if the corner was already placed.
This is pretty nice. o.o

I agree, the 2-gennness is amazing.
what if we solved the F2L corner while affecting the LL corners in some way?
like a CPLS with only the corner not the edge so after that we only have 2-gen OCLL algorithms and then the L5E?

 

[wir3sandfir3s]

I agree, the 2-gennness is amazing.
what if we solved the F2L corner while affecting the LL corners in some way?
like a CPLS with only the corner not the edge so after that we only have 2-gen OCLL algorithms and then the L5E?

So solve a corner and permute the rest? I think it is better with a whole slot because then you can do 2GLL

 

[Shiv3r]

So solve a corner and permute the rest? I think it is better with a whole slot because then you can do 2GLL

or do CLS-I algs, insert it(unoriented) and use EJF2L to orient them, then permute corners and do the L5E.
all the EJF2L algs can be 2-gen.

 

[wir3sandfir3s]

or do CLS-I algs, insert it(unoriented) and use EJF2L to orient them, then permute corners and do the L5E.
all the EJF2L algs can be 2-gen.

EJF2L can be counterproductive. People say it's easier to insert unoriented, but I don't see the point. Just insert it solved with 1 extra move, which gets rid of a longer alg and 9 cases. At that point, your just doing COLL, which isn't even that efficient...

 

[crafto22]

or do CLS-I algs, insert it(unoriented) and use EJF2L to orient them, then permute corners and do the L5E.
all the EJF2L algs can be 2-gen.

That's a horrible idea. Inserting it oriented takes the same number of moves as inserting it unoriented. You're adding an extra look

 

[crafto22]

A super efficient method with a ton of freedom but very difficult recognition:

1. Solve two 2x2x1 blocks on B like Roux blocks without the front pairs, but they don't have to match in colour in any way
2. Solve the front corners and then the U layer corners using CMLL
3. Solve 2x3x1 blocks like Roux blocks but they don't necessarily have to match
4. L6E

In order to achieve the greatest possible efficiency, one must choose their first two blocks wisely in order to achieve a low movecount. After doing a couple dozen solves I've found that I can usually solve the blocks in about 8 moves. For the third step, it's important to choose your blocks wisely. You want to solve the block efficiently, but also influence the upcoming L6E for maximal efficiency.

Here is an example solve:

D U' L' F U L2 U' R B D' F D F2 B2 R2 D' B2 U L2 U F2 U2

x2 R' B' U F R2 B' L // Blocks (7/7)
U' F R U2 B' R' F R2 B R2 F' // D layer corners cancelled into a Sune CLL (11/18)
M2 U2 F M2 F' // 2x3x1 blocks (5/23)
U2 M' U2 (M R) U R' U' M' U R U' r' // L6E done using an ELL finish (12/35)

This example solve was actually one of the longer solutions I've gotten but since people are always so skeptical I didn't want to use one of the best solves as an example. 35 ATM is still very very good if you ask me.

Okay so obviously this has one huge problem: recognition. In this particular example solve recognition was not really a problem since it just so happened that the most efficient blocks also made recognition easy, but it won't always be this simple. If I could invent some recognition system this could be really good.

 

[JTWong71]

Okay so obviously this has one huge problem: recognition. In this particular example solve recognition was not really a problem since it just so happened that the most efficient blocks also made recognition easy, but it won't always be this simple. If I could invent some recognition system this could be really good.

Wouldn't fingertricks also be a problem for the method?
It looks like you are using a decent amount of B Moves during the Example Solve.

 

[wir3sandfir3s]

A super efficient method with a ton of freedom but very difficult recognition:

1. Solve two 2x2x1 blocks on B like Roux blocks without the front pairs, but they don't have to match in colour in any way
2. Solve the front corners and then the U layer corners using CMLL
3. Solve 2x2x3 blocks like Roux blocks but they don't necessarily have to match
4. L6E

In order to achieve the greatest possible efficiency, one must choose their first two blocks wisely in order to achieve a low movecount. After doing a couple dozen solves I've found that I can usually solve the blocks in about 8 moves. For the third step, it's important to choose your blocks wisely. You want to solve the block efficiently, but also influence the upcoming L6E for maximal efficiency.

Here is an example solve:

D U' L' F U L2 U' R B D' F D F2 B2 R2 D' B2 U L2 U F2 U2

x2 R' B' U F R2 B' L // Blocks (7/7)
U' F R U2 B' R' F R2 B R2 F' // D layer corners cancelled into a Sune CLL (11/18)
M2 U2 F M2 F' // 2x2x3 blocks (5/23)
U2 M' U2 (M R) U R' U' M' U R U' r' // L6E done using an ELL finish (12/35)

This example solve was actually one of the longer solutions I've gotten but since people are always so skeptical I didn't want to use one of the best solves as an example. 35 ATM is still very very good if you ask me.

Okay so obviously this has one huge problem: recognition. In this particular example solve recognition was not really a problem since it just so happened that the most efficient blocks also made recognition easy, but it won't always be this simple. If I could invent some recognition system this could be really good.

I totally agree with you, this method is amazingly efficient just terrible recog. To fix recog (or make it better), you have to see where it is bad, in this case CxLL and the second block. To fix the second block is easy: plan first in inspection, and look ahead while building it. For CxLL, you have the freedom of two extra edges so we can use this to our advantage and create a new alg set, but i don't see the point in that as two extra free edges isn't really too much freedom. We can also influence the corners during previous steps, but I like the simplicity in this method and I don't want to ruin that. I think the best thing for this case is trying to improve the CxLL execution.

Also, the example isn't working for me, but that may just be human error on my part.

 

[KevinM]

This is a 3x3 last slot subset, I would like to call KMLS (Kevin Min Last Slot) and this will take a butt load of effort to accomplish. Although I do not expect anybody to learn the whole thing, but it is nice to know some of the east ones ( ill include in the post)
Basically, this can be called WVCP (Winter variation cp) which means on your last slot, when all edges are oriented, you will use an algorithm to permute the top colors, and the corners. Sounds INSANE. Right? But there are algs that are simple.(Coming up on yt. I will update this post when it does) Such as an alg which is just a J or T perm with R U R' cancelled, or some insertion cancelling into a coll.
I will get most algs from experiences, or using cube explorer, and find the most finger trickable.
If anybody is interested in helping me(it is a lot of algorithms to do alone!) Private message me or email me at [email protected]

 

[gyroninja]

This is a 3x3 last slot subset, I would like to call KMLS (Kevin Min Last Slot) and this will take a butt load of effort to accomplish. Although I do not expect anybody to learn the whole thing, but it is nice to know some of the east ones ( ill include in the post)
Basically, this can be called WVCP (Winter variation cp) which means on your last slot, when all edges are oriented, you will use an algorithm to permute the top colors, and the corners. Sounds INSANE. Right? But there are algs that are simple.(Coming up on yt. I will update this post when it does) Such as an alg which is just a J or T perm with R U R' cancelled, or some insertion cancelling into a coll.
I will get most algs from experiences, or using cube explorer, and find the most finger trickable.
If anybody is interested in helping me(it is a lot of algorithms to do alone!) Private message me or email me at [email protected]

Bilde(man) has a complete wvcp alg sheet. It's probably posted earlier in this thread (and also in it's own thread?).

 

[shadowslice e]

This is a 3x3 last slot subset, I would like to call KMLS (Kevin Min Last Slot) and this will take a butt load of effort to accomplish. Although I do not expect anybody to learn the whole thing, but it is nice to know some of the east ones ( ill include in the post)
Basically, this can be called WVCP (Winter variation cp) which means on your last slot, when all edges are oriented, you will use an algorithm to permute the top colors, and the corners. Sounds INSANE. Right? But there are algs that are simple.(Coming up on yt. I will update this post when it does) Such as an alg which is just a J or T perm with R U R' cancelled, or some insertion cancelling into a coll.
I will get most algs from experiences, or using cube explorer, and find the most finger trickable.
If anybody is interested in helping me(it is a lot of algorithms to do alone!) Private message me or email me at [email protected]

Do you mean this?

 

[Shiv3r]

I had an idea for a newb method, kind of a transition between a beginner method and CFOP.
backstory:
When I was learning CFOP, I didn't learn F2L for a while. For a lot of the time, I didn't even do corners->edges. I would solve the F2L edge, then solve the corner using some quick 2-gen algorithms. So I decided to try to expand that to make a method that is reasonably fast and can be used as a stepping stone to intuitive F2L.

so here are the steps for TFOP(T-style F2l Oll Pll)
-Solve a cross
-Insert the F2L edges as in Keyhole without adjusting D face.
-Use one of 3 fast algorithms to solve the F2L corners
-Last layer however you like

 

[crafto22]

I totally agree with you, this method is amazingly efficient just terrible recog. To fix recog (or make it better), you have to see where it is bad, in this case CxLL and the second block. To fix the second block is easy: plan first in inspection, and look ahead while building it. For CxLL, you have the freedom of two extra edges so we can use this to our advantage and create a new alg set, but i don't see the point in that as two extra free edges isn't really too much freedom. We can also influence the corners during previous steps, but I like the simplicity in this method and I don't want to ruin that. I think the best thing for this case is trying to improve the CxLL execution.

Also, the example isn't working for me, but that may just be human error on my part.

Nope example solve works. Also the block isn't the problem, the problem is CMLL recog and L6E recog, although I've been able to improve that quite a bit. For now I've been orienting all edges as if the second block matched and then I simply imagine that that edge is a different. Is sorta hard to explain, maybe I'll do a video example solve to explain it better. Also I got a 25 move solution this morning (5 move 2x2x1s, 7 move corners, 3 move 2x3x1s and 10 move L6E), and the L6E wasn't particularly good. I think corners could be as low as 5 with cancellations and L6E can occasionally be 8. Also the 2x3x1s can quite often be solved with a quick M U M trigger or something short like that. I'm pretty sure a lucky speedsolve could take as few as 21 moves, which would be almost a third of CFOP's average moves. Even with the horrible recog this method could be super efficient which would make up for the pauses.

 

[crafto22]

Wouldn't fingertricks also be a problem for the method?
It looks like you are using a decent amount of B Moves during the Example Solve.

You could rotate and execute them as U moves, I simply didn't show that in the example solve. In terms of ergonomics this method isn't actually too bad.

 

[Shiv3r]

Nope example solve works. Also the block isn't the problem, the problem is CMLL recog and L6E recog, although I've been able to improve that quite a bit. For now I've been orienting all edges as if the second block matched and then I simply imagine that that edge is a different. Is sorta hard to explain, maybe I'll do a video example solve to explain it better. Also I got a 25 move solution this morning (5 move 2x2x1s, 7 move corners, 3 move 2x3x1s and 10 move L6E), and the L6E wasn't particularly good. I think corners could be as low as 5 with cancellations and L6E can occasionally be 8. Also the 2x3x1s can quite often be solved with a quick M U M trigger or something short like that. I'm pretty sure a lucky speedsolve could take as few as 21 moves, which would be almost a third of CFOP's average moves. Even with the horrible recog this method could be super efficient which would make up for the pauses.

I came up with this with me and jtwong, but either solving corners like above, or building the 2 blocks and use some kind of L2S to skip CMLL.
The problem was that it could simply not be fast, its faster to solve roux FB and SB than building 2 1x2x2s and solving the corners.

 

[crafto22]

I came up with this with me and jtwong, but either solving corners like above, or building the 2 blocks and use some kind of L2S to skip CMLL.
The problem was that it could simply not be fast, its faster to solve roux FB and SB than building 2 1x2x2s and solving the corners.

How so? Technically you're solving Roux blocks without two edges then just CMLL. Finally you insert those last edges but you have way more options and you can optimize L6E. So basically this method HAS to be faster than Roux (in theory, and only if recog was good)