The New Method / Substep / Concept Idea Thread

[ttran9235]

This is very similar to Petrus.

It is supposed to be some kind of variant

 

[supercavitation]

It has the same LSE cases as Roux, just the bad edges ones. Then you will ALWAYS have at MAX 8 moves to finish your F2L, and then, bring you to EPLL. Look at the example to see more clearly
I will post a vid in a moment on how to solve it how I mean to solve it

This is exactly what he was referring to, a variant that has been thought up and dismissed by many different cubers. As Guroux already said, a more efficient variant would be to skip the EO step, solve F2L, and then do ELL, but even that's simply not nearly as efficient as Roux with standard LSE.

 

[Ninja Storm]

It has the same LSE cases as Roux, just the bad edges ones. Then you will ALWAYS have at MAX 8 moves to finish your F2L, and then, bring you to EPLL. Look at the example to see more clearly
I will post a vid in a moment on how to solve it how I mean to solve it

Lower movecount and better fingertricks for Sandwich(is this the right term?) instead of EPLL.

 

[AlexTheEmperor]

I have discovered in the last 5 minutes a technique that I originally used for OH but can be advantageously applied to 2H. It is designed for my main method of CFOP but there might well be applications to block building in other methods.

When I was OHing, I found it annoying how I could not insert pairs into certain slots easily without doing y. I overcame this by transferring my Cross on B to a COL and reverting after the insertion. In 2H, it could be used for certain inserts into BL and BR which would require a y, the core of this method being that you do a z to put make the insert or even to pair the CE if you can spot bad EO.

Why not do normal F2L, you may think but my response is that a z rotation is far quicker to do than a y and the z (insert) z' clause that would be used can be executed in the same wrist position and without regripping.

Why does y instead of x mean a faster F2L, it is because ys, is the generally accepted slowest rotation to do since it requires a total regrip of the RH; whereas z can be executed with the thumb static around the F center and the index finger can be ready to flick U both before and after z with the LH only requied to let gravity rotate the cube.

This technique can be used to free your F2L of ys, the appropriate adjustments for each slot's insert can be made by doing wide Ls as needed.

I will put together an example soon, any initial thoughts or feedback?

Some names for this technique could be:

The Z technique / ZF2L

or

Goodhead F2L / GF2L

GZF2L

F2L-Goodhead handling

 

[donal56]

I dont know if it exists but I modified the cage method (its for big cubes) to work in 3x3. Its not the fastest but I like to solve the cube way different.Starts like this.

1.- Solve all the cornes in the way you want, I solve with ortega
2.-Move the centers to the right position
3.- Insert the bottom face edges, it can be inseted from the top or with E slices
4.- Rotate 180 and insert the bottoms edges. First locating the edges, moving to the color's face and in the oposite side do R' D R or L D' L' (according to the case), then insert edge and reverse algorithm.
5.- The midle edges can be oriented doing R2 and E moves intuitively. There are 4 cases, all in pure flips, 2 edges in vertical (almost all the time) , 2 diagonal (sometimes), all 4 (very few times) or solved (rare)

The first two cases can be solved with a ELL and the third doing it twice. As I said before it is just an idea not to speedsolve. [emoji28]

 

[goodatthis]

I dont know if it exists but I modified the cage method (its for big cubes) to work in 3x3. Its not the fastest but I like to solve the cube way different.Starts like this.

1.- Solve all the cornes in the way you want, I solve with ortega
2.-Move the centers to the right position
3.- Insert the bottom face edges, it can be inseted from the top or with E slices
4.- Rotate 180 and insert the bottoms edges. First locating the edges, moving to the color's face and in the oposite side do R' D R or L D' L' (according to the case), then insert edge and reverse algorithm.
5.- The midle edges can be oriented doing R2 and E moves intuitively. There are 4 cases, all in pure flips, 2 edges in vertical (almost all the time) , 2 diagonal (sometimes), all 4 (very few times) or solved (rare)

The first two cases can be solved with a ELL and the third doing it twice. As I said before it is just an idea not to speedsolve. [emoji28]

First, it's not really a modification on the cage method, it's more of just another corners first method. Seems to me it's the most similar to Waterman. Generally the term corners first is just a blanket term for methods that solve the corners first, you could solve the edges anyway you like. Also, one thing I like doing is going on the wiki and browsing on the methods list, it's pretty cool how many methods are out there, and you can learn a lot.

 

[brian724080]

I have discovered in the last 5 minutes a technique that I originally used for OH but can be advantageously applied to 2H. It is designed for my main method of CFOP but there might well be applications to block building in other methods.

When I was OHing, I found it annoying how I could not insert pairs into certain slots easily without doing y. I overcame this by transferring my Cross on B to a COL and reverting after the insertion. In 2H, it could be used for certain inserts into BL and BR which would require a y, the core of this method being that you do a z to put make the insert or even to pair the CE if you can spot bad EO.

Why not do normal F2L, you may think but my response is that a z rotation is far quicker to do than a y and the z (insert) z' clause that would be used can be executed in the same wrist position and without regripping.

Why does y instead of x mean a faster F2L, it is because ys, is the generally accepted slowest rotation to do since it requires a total regrip of the RH; whereas z can be executed with the thumb static around the F center and the index finger can be ready to flick U both before and after z with the LH only requied to let gravity rotate the cube.

This technique can be used to free your F2L of ys, the appropriate adjustments for each slot's insert can be made by doing wide Ls as needed.

I will put together an example soon, any initial thoughts or feedback?

Some names for this technique could be:

The Z technique / ZF2L

or

Goodhead F2L / GF2L

GZF2L

F2L-Goodhead handling

I'm not sure about the whole speed with rotations, but I'm pretty sure that x rotations require a complete regrip too. However, I do know that using double layer moves is a pretty standard technique in advanced F2L.

 

[Randomno]

If this doesn't exist, I'm calling it CRY for sure (CRosses, Y perm).

1. Make a cross.
2. Put in 3 middle layer edges.
3. Solve last layer cross, making use of the empty slot.
4. Identify where the UBL corner should go.
5. Bring it to DFR. The corner does NOT have to be moved so that it will have correct orientation after best step.
6. Y perm.
7. Repeat for all corners.
8. Find two unoriented diagonals and put them at UBL and DFR.
8a. If there are none, move one diagonally opposite the other. After the Y perms, undo this move.
9. Y perm, then x' z2 (rotate so other corner is in DFR).
10. Y perm. Repeat 3 times.
11. If the two corners are now oriented, repeat step 8. Otherwise, repeat step 9.
12. Solved.

Probably hard to follow, I'll do an image guide later today.

 

[qqwref]

Wat? So it's like edges first and then doing the corners with a really slow Old Pochmann variant?

 

[Randomno]

Wat? So it's like edges first and then doing the corners with a really slow Old Pochmann variant?

Yeah, it's a reasonably simple 1 alg method.

EDIT: Just timed myself, 2:45.55. CRY is a good name, it's what you'll do after all those Y perms.

 

[Chenkar]

If this doesn't exist, I'm calling it CRY for sure (CRosses, Y perm).

1. Make a cross.
2. Put in 3 middle layer edges.
3. Solve last layer cross, making use of the empty slot.
4. Identify where the UBL corner should go.
5. Bring it to DFR. The corner does NOT have to be moved so that it will have correct orientation after best step.
6. Y perm.
7. Repeat for all corners.
8. Find two unoriented diagonals and put them at UBL and DFR.
8a. If there are none, move one diagonally opposite the other. After the Y perms, undo this move.
9. Y perm, then x' z2 (rotate so other corner is in DFR).
10. Y perm. Repeat 3 times.
11. If the two corners are now oriented, repeat step 8. Otherwise, repeat step 9.
12. Solved.

Probably hard to follow, I'll do an image guide later today.

Interesting idea. I'm surprised that I was able to follow it. I'll try it out later and think of some ways to make it a fast-ish method

Edit: By LL cross, do you mean orient and permute, or just orient?

 

[Randomno]

Interesting idea. I'm surprised that I was able to follow it. I'll try it out later and think of some ways to make it a fast-ish method

Edit: By LL cross, do you mean orient and permute, or just orient?

Same edge method as 8355 or Sexy.

 

[mDiPalma]

Not sure if this idea is taken before but:
1. 2x2x2 block
2. Expand to 2x2x3
3. Then expand to F2L minus 1CE
4. Orient the remaining edges
5. Solve the F2L and the last layer edges (2-gen)
6. Solve the last the remaining corners using L4C

Minimum amount of moves can be 25 with L4C skip and at least maximum of 55 moves (bad L4C case or block)

2x2x2-8 moves (8)
2x2x3-5 moves (13)
F2L- 1CE-6 moves (19)
Orient edges-3 to 6 moves (22 to 25)
F2L and LL edges- 7 moves (29 to 32)
L4C-12 moves (41 to 44)


Total of 84 algorithms

It's a good method. It's actually almost what I used to do for Heise (but instead of full L4C, I used 2-look commutators to solve).

But you overestimate the movecount. You don't need that many moves for the third step.

The only problem is that L4C algs truly suck.

But to be honest, this is the best legitimate variant proposal that I have ever seen in this thread.


Edit: you can also use https://www.speedsolving.com/wiki/index.php/LPELL to solve the LL edges while inserting the last pair

 

[2180161]

I believe I have a keyhole variant:
Cross
F2l-Corner edge pairs do not have to match, but do however have to be in correct orientation so when you do a D, D' or D2 move, it pairs up the corner and edge, then orient LL edges (2-look OLL step), and then you pair up the corner edge pair, remove, correctly place corner edge pair, repeat
OLL
PLL
Pros-F2l pairs dont have to match, making for more options, LL edges oriented
Cons-LOTS of moves.

 

[supercavitation]

I believe I have a keyhole variant:
Cross
F2l-Corner edge pairs do not have to match, but do however have to be in correct orientation so when you do a D, D' or D2 move, it pairs up the corner and edge, then orient LL edges (2-look OLL step), and then you pair up the corner edge pair, remove, correctly place corner edge pair, repeat
OLL
PLL
Pros-F2l pairs dont have to match, making for more options, LL edges oriented
Cons-LOTS of moves.

That would waste a lot of moves, and moving around the first layer corners without messing up EO is going to waste even more, and will almost certainly require several algorithms. There's absolutely no way this ever turns out worth it.

 

[lerenard]

So, I'm not trying to be like the guy in the 19th century who said everything that could be invented already existed, but are there seriously any new "methods" that haven't been found yet? I mean, so many people have looked into so many different things, I feel like we basically already know of everything that is worthwhile by now.

I'm not trying to hate on newness or originality, I just feel like there really aren't very many ways to solve a 3x3 outside of all the different ways everyone has already come up with (or small variations on these).

Also, FreeFOP basically covers anything anyone could think of in the future. ha.

 

[mDiPalma]

I'm not trying to hate on newness or originality, I just feel like there really aren't very many ways to solve a 3x3 outside of all the different ways everyone has already come up with (or small variations on these).

i can easily write down BILLIONS of methods that have never been thought of, never will be thought of, and bear no resemblance to FreeFOP

think about it. you can solve any property of any piece on the cube in any order, and still move the cube closer to the solved state. My first step could be orienting the BR edge while placing it in the D layer, my second step could be orienting the corners with respect to L/R while placing UBR in the left layer. are we solving the cube? YES. has this dumb approach been written down yet? NO.

the point of this thread, and speedsolving in general, is to have fun. thinking of new methods is fun. finding new algs for our new methods is fun. practicing our new method until we get a random sub-15 is fun. and looking at all the stupid methods that other people come up with is also fun.

 

[lerenard]

So, the purpose of this thread is not necessarily to discover new methods that are better than the ones we have now, but simply to share things we think of and have fun? If that's the case, awesome. I just misunderstood its purpose.

My method I currently use is basically FreeFOP, but there's a lot of different ways people can do that, since it's basically just "build the F2L however you can." My previous method being Petrus, I usually focus on blockbuilding, or an X-cross minus one edge (a.k.a. a 2x2x2 + one edge) and then do it Fridrich style with a free M-slice, limiting rotations, but I'll fill in the final edge piece as a 2x2x1 block if I see it can be done easily. And I almost always orient the edges before I get to the LL, often using VHLS (only like 5 cases) and I know EJF2L as well. Sometimes I'll use full Fridrich if there's an easy cross. I don't use Roux though. I've tried, and it just is way slower than anything else I use (not because of the method itself, I'm sure) but when I do use it after the 2 blocks I orient corners, then fit in one of the D edges and then orient the rest of the edges and place the other D edge using the technique shown below. I then finish with a PLL. I don't do ZZ either, because it's not like you just pick up a cube and say "oh hey I have a really easy EOLine case" unless you have practiced with it a lot, and I'm not willing to put in the work for that xD. Basically, though, I'm just trying to use as many different methods as possible so I never have the problem of "agh where is that one corner I need" and can use whatever I see. I average a 32-move F2L with Petrus. Other ways I solve make up for using more moves by being fingertrick friendly and/or limiting rotations. Here are a few example solves:

Scramble: (White as U, Green as F) F' R2 F' U' R2 U D R U' F B' L2 R' U2 R F2 D2 F' D2 U' R' U' B' L2 F2

Inspection:
Okay, so the first thing I notice is the RY edge is already in place, and we have two edge/corner pairs: RBY and RGW thus, I will build around the red face and my LL will be orange. All I have to do is not use any orange pieces.

L' U (2x2x1 RBY block)
D F' D' (2x2x1 RGW block)
L' F B' U' F2 (connecting all the solved pieces to where they go)
now I inspect the cube and see the RBW pair can be easily matched up
z(? rotate the cube so Orange is U and Green is F. I've never understood cube rotation notation) U2 R' U2 R U' R' U R (RBW pair done Fridrich style)
U2 L' U L U' F R U' R' F' (RGY pair with VHLS because I don't know full OLL.)
28 move F2L
btw this was a lucky scramble, I almost never get that much to start with AND it's that easy to put in. Usually they're all jumbled in with each other and it's a pain to solve anything without messing up the stuff the scramble gave you. Here's another example:

Scramble: R' F' L' F' B2 D2 F D' L2 U' R B2 L2 U' D2 B' F' R2 U' R' B2 R L2 U' R2

Inspection: I have a WBR pair and two White cross edges in place (one of which is blue). Finally I see the RB edge and figure out how to place it while solving everything else.

R U' B' (RB edge)
U' F2 U' (X-cross -1 / 2x2 + 1) So my color this time is White, and I have to avoid all Yellow pieces.
x2 (Blue is F, yellow is U)
I see the WBO pair that would normally be a pain, but is fairly easy to handle here because almost the whole L face is free.
L U2 L2 U' L
I realize I should have been looking ahead here, but I'm only averaging 30 seconds (maybe that's why?) so the next thing I see is the WGO corner (not the WGR pair, which is easier... So I would look to the B face for its edge.
U' R' U2 R U M U M' (I think that's how you write M notation (M follows L). The pair is now connected) U L U' L'
Now I would look for my last edge piece and resist the temptation to try and fit in the GW edge as part of a block (that never works (quickly) when one of the F2L pairs is already inserted)
R' U' R U R' U' R U' R' U R - RGW pair
U M U' M' U' M U M' last edge and edge orientation.
42-move F2L (STM), but the last two steps were 19 moves (!) and would be executed very fast in fingertrick fashion.

So yeah, not really a new method per se, just how I approach FreeFOP. It works for me, but I need to look for more tricks and get used to it, because I still have consistently better times with Petrus when I happen to use it (the corner I use after the 2x2 happens to be the right one for a 2x2x3). I even set a PB of 20.87 yesterday. Also, if this isn't the right place to put this since it's basically just FreeFOP, sorry. I'm just trying to show how /I/ do it.

 

[mDiPalma]

or an X-cross minus one edge (a.k.a. a 2x2x2 + one edge) and then do it Fridrich style with a free M-slice, limiting rotations, but I'll fill in the final edge piece as a 2x2x1 block if I see it can be done easily. And I almost always orient the edges before I get to the LL, often using VHLS (only like 5 cases) and I know EJF2L as well.

One thing that you could do is after the "2x2x2 + one cross edge" is build and insert the next F2L pair (into the open F2L slot that you made) while either:

1) orienting the edges.
2) permuting the corners via ZZ-Porky V2, or some alternative

the former will leave you with a ZZ-esque finish (my ZZ soft-spot is showing)

and the latter will leave you in some stupidly awkward cube state with messed up edges, but really nicely permuted corners (so you only need to use 2 faces of the cube, and slices to solve) which is mad weird, but would probably be kind of ergonomic and efficient.


You should give these a try!



or another idea for option 2)

you could expand your first step to be "2x2x2 + 1 cross edge + orienting the E-slice edges"
and then solve the missing F2L pair in your second step while solving the Corner Permutation
third, solve the next two F2L pairs without solving the last cross edge, but inserting the last F2L pair with winter variation so your corners are completely 100% solved relative to each other, but have a ton of screwed edges floating around in between
fourth, solve the remaining 5 edges of the cube with either 1) commutators or 2) EO and EP5 (which is, in my opinion, how the roux method should ACTUALLY finish)


lol That's a horribly complicated method if I ever saw one! but it probably outperforms CFOP movewise, which makes me frown and smile at the same time.

 

[lerenard]

One thing that you could do is after the "2x2x2 + one cross edge" is build and insert the next F2L pair (into the open F2L slot that you made) while either:

1) orienting the edges.
2) permuting the corners via ZZ-Porky V2, or some alternative

the former will leave you with a ZZ-esque finish (my ZZ soft-spot is showing)

and the latter will leave you in some stupidly awkward cube state with messed up edges, but really nicely permuted corners (so you only need to use 2 faces of the cube, and slices to solve) which is mad weird, but would probably be kind of ergonomic and efficient.


You should give these a try!



or another idea for option 2)

you could expand your first step to be "2x2x2 + 1 cross edge + orienting the E-slice edges"
and then solve the missing F2L pair in your second step while solving the Corner Permutation
third, solve the next two F2L pairs without solving the last cross edge, but inserting the last F2L pair with winter variation so your corners are completely 100% solved relative to each other, but have a ton of screwed edges floating around in between
fourth, solve the remaining 5 edges of the cube with either 1) commutators or 2) EO and EP5 (which is, in my opinion, how the roux method should ACTUALLY finish)


lol That's a horribly complicated method if I ever saw one! but it probably outperforms CFOP movewise, which makes me frown and smile at the same time.

That might be a good idea in general, but not for me. I like my LL the way it is. 28 algs, 2 steps. I also know 2-sided PLL recognition (although my recognition time is not the fastest...). Basically, I have enough invested that I'm not interested in permuting anything until everything's oriented. I have considered WV, although I tried to learn a few algs and they didn't come to me very easily. But yeah, I would be more interested in different ways to build the F2L than how to influence/solve the LL.

The problem I see with filling the F2L slot as soon as you build it is 1: if you're going to solve the pair immediately, why not just build a 2x2x1 block? and 2: The whole point I moved to FreeFOP is that you aren't tied down to a specific system/order. You don't have to look around for that one corner you need if you see 2 pieces right in front of you that can already be matched up. filling in the cross edge is beneficial because I can do it without much/any thought (and I'll only do it if I already see how to do it easily) and it allows me to just make the corresponding F2L pair at any time (whenever I happen to find it). I don't have to worry about making it immediately just because I have one edge that goes with it.