# The New Method / Substep / Concept Idea Thread

#### Cale S

##### Member
OLL skip or VLS and OLL with EG on skewb would be helpful.
Instead of just EG, you can do EG and orient all top side at once. (EGOLL) This will get rid of L5C and first layer helpful.
If OLL skip is possible for skewb, it will get rid of advanced sarah cases and L5C. It will leave off L4C or L3C. (VLS)

downside is that i cant work on my ksolve program, my stupid command prompt is too advanced and frustrating
There's a document with all cases to solve last layer corners + U center while inserting the last corner, but it's 90 algs and it's almost always faster to do layer then one alg

The EGOLL algs are decent, there's a document with nice algs but it's rarely worth it to use them if you want to be really fast

#### Shiv3r

##### Member
Hey guys, so I was playing around with some 4x4 direct solving methods and I thing I came up with a really nice method(its more of a way to do LSD in stadler, however some variations):
1.2 opp centers
2.roux F2B
you can do CLL anytime between step 3 and step 5
3.Solve D center, B center, and DB edge(I solve D center, then DB edge, then B center; for freedom of all slices, M2U2M2 is your best friend)
4.Finish last 2 centers in M slice
5.Solve last 5 edges similarly to K4 ELL(is a pretty nice step, a little nicer than ELL imo)

what do you think? my PB with this method is 1:20(My PB with my main, meyer, is 1:18).

#### Sue Doenim

##### Member
Wait guys, you don't have to BLD trace

Let's just assume we've got a permutation of corners on R and U faces. We can of course solve DRF and DRB <R,U> leaving a permutation of LL corners on U that we could use to detect a CP fix (aka what people do with COLL or porky style recog), but we can also simulate this solving in inspection with swaps. If we solve DRF and DRB with pure swaps, there's at most 1 extra "correction" swap that we have to do s.t. the 3 swaps performed together form a valid 2-gen permutation (essentially using the property of 2gen group to simulate the <R,U> moves needed to solve DRF and DRB while shortcutting all the tracking you'd need to do so). Then we can much more quickly figure out where our LL pieces would be w.r.t each other than if we had directly tracked what the LL would look like by imagining DRF and DRB solved w/ <R,U>.

So given any permutation of 6 corners we can imagine what the LL would look like if we 2gen solved DRF and DRB by imagining these swaps being done. And then looking at the resulting permutation of LL corners to figure out the key swap

Now this is easy to generalize to 7 corners--whenever we hit the DLF piece, we just pretend it's whatever piece is current at DLF (simulating a swap to solve DLF). Now without any BLD tracing we can extract a key swap, and we can easily express this key swap in a standard form that lets it be easily tracked.
This seems possible. Could you trace CP the same way as in your other method (message #2516)?

#### mDiPalma

##### Member
this method is bad for the EXACT same reason that LBL is bad.

#### Teoidus

##### Member
This seems possible. Could you trace CP the same way as in your other method (message #2516)?
Yeah, that would be possible assuming that you took the key swap and expressed it in a "standard form" where it must involve the DLF piece if DLF is unsolved.

Actually now that I'm here I might as well post how to track stuff

so you have to keep track of 3 things when tracking: generators, DLF piece position, and key swap. Lucky we've taken care of the last two by making key swaps always include DLF if it's unsolved wrt generators. Generators can be encoded by a single sticker: BDL. This single sticker encodes orientation and permutation of DLB piece (and therefore where our generators are).

So when memorizing the "CP fix" of any cube state, we need to know a single sticker marking BDL, a piece marking DLF piece, and another piece indicating key swap.

I said before that <R,U,L,D> are all pretty easy to track--you just watch the swap move around and shift the generators when necessary. Luckily the single sticker encodes our generators just fine, so to track all of <R,U,L,D> we basically just track all 3 items around the cube as you would normally.

Now the nice part about using BDL is that it's visually apparent whenever we make an F/B move, as the face being turned will always be "parallel" to the BDL sticker (as it must reside on the B face wrt generators). So it's easy to know when we will have to do something other than the normal tracking that we'd use with <R,U,L,D>

To track F moves w.r.t generators, we have to rely on some brute force if we want to do it quickly (for now--working on a b better way to do this). For each of the 36 different swaps we can have memorized, we need to know what swap it turns into before and after an F, F', F2, etc. Since F4 = I, F moves have to cycle through swaps in sets of 4. Therefore we can memorize these in 9 sets of 4-cycles between swaps.

So if you want to know what happens to a no-swap when you apply an F move, you just find the picture corresponding to a no-swap and look at the picture 1 to the right (F). For F', look 1 to the left (though keep in mind the cycles wrap around, so "to the left of the very first element in the group of 4" = "the last element in the group of 4")

#### VenomCubing

##### Member
I have been working on this method for a few months now, and I believe it has some potential. I call it the GS method, or Grand Setup method. It is a variant of CFOP and Roux, with it's own unique step. I would love to hear anybody's feedback, positive or negative. I would also appreciate it if anyone could help me by making more example solves. I am also looking for a better name for this method. If you think you found a better name than GS, please tell me. Without further ado, here is the GS method.

Steps:

Step 1:
Build a 2x2x2 block on the bottom in the back right.

Step 2: Build an adjacent 2x3x1 block on the L face.

Step 3: Pair up last F2L pair, but don't insert it

Step 4: Insert last F2L pair with Winter Variation, which orients the last layer corners.

Step 5: Simultaneously orient the last layer edges while inserting the DF edge using only M and U moves. This step is intuitive. I call it SFE, or Setup of Five Edges.

Step 6: Permute the last layer with PLL.

*Example Solve:

Scramble:
U2 L2 F2 U L2 F2 U2 R2 B2 D' L B' D' U' L D2 R2 B D2 L'

2x2x2 Block: D R' F' R' B'

2x3x1 Block: U F U L

Last F2L Pair: Skip

Winter Variation: U2 L' U R U' R' L

SFE: U M' U M U2 M' U' M

PLL: R' U L' U2 R U' L R' U L' U2 R U' L U'

Only 48 Algorithms are needed, (not including reflections,) so most of the solve is intuitive.

After the blocks are finished, the cube can be solved quickly without any cube rotations.

GS requires proficiency in Blockbuilding and Roux-Like edge solving, which may be hard for beginners to do efficiently.

The 2x2x2 Block in the back may be awkward to do without cube rotations, so it may take some time getting used to.

* Sorry the example solve wasn't very good. I don't have much experience with making them.

#### Myachii

##### Member
Ngl it looks interesting. Targetted towards multi-method solvers though. Try get some feedback from roux-cfopers

#### APdRF

##### Member
Why don't insert the last pair -> OCLL (with easy algs that don't preserve EO) -> SFE -> PLL? It's only 28 algorithms, and OCLLs are better than Winter Variation cases + mirrors. And you can even learn COLL (some algorithms should be easier tan normal COLL) and then solve L5E.

#### VenomCubing

##### Member
Why don't insert the last pair -> OCLL (with easy algs that don't preserve EO) -> SFE -> PLL? It's only 28 algorithms, and OCLLs are better than Winter Variation cases + mirrors. And you can even learn COLL (some algorithms should be easier tan normal COLL) and then solve L5E.
OCLL might be good. COLL has worse recognition, though, so corners solved might not be worth it.

#### Loiloiloi

##### Member
What is this method? I'm learning this as an advanced version of 8355 but I'm not sure it could even be considered 8355 at this point.

F2L-2
Insert an F2L edge
Solve third CE pair using VHLS, leaving the solved edge alone
Use a PLL to solve edges
8355 Style L5C

Last edited:

#### Teoidus

##### Member
non-intuitive 8355, looks like.

#### xyzzy

##### Member
What is this method? I'm learning this as an advanced version of 8355 but I'm not sure it could even be considered 8355 at this point.

F2L-2
Insert an F2L edge
Solve third CE pair using VHLS, leaving the solved edge alone
Use a PLL to solve edges
8355 Style L5C
Is there a reason you'd solve the third and fourth slots as edge-then-pair, instead of the more standard pair-then-edge? It looks like your proposal has neither better lookahead nor lower move count.

If you leave the corner buffer to the last slot, you get the added benefit of being able to solve the last five edges with intuitive commutators very easily.

Last edited:

#### Loiloiloi

##### Member
Is there a reason you'd solve the third and fourth slots as edge-then-pair, instead of the more standard pair-then-edge? It looks like your proposal has neither better lookahead nor lower move count.

If you leave the corner buffer to the last slot, you get the added benefit of being able to solve the last five edges with intuitive commutators very quickly.
I don't know how to do commutators, this method is just ease of learning for me. The algorithms are just PLLs I already know and some easy VHLS algs.

#### xyzzy

##### Member
I don't know how to do commutators, this method is just ease of learning for me. The algorithms are just PLLs I already know and some easy VHLS algs.
If you leave the buffer corner to the last slot, you can use a variant of VHLS that doesn't need to care about corners. Alternatively, you can also do Petrus-style edge orientation with the last slot (most cases are either one trigger or two triggers that cancel into each other). It's all fundamentally the same idea, except that VHLS is more restricted (and hence takes more moves) because it needs to preserve the pair.

Examples: 1 2
EO-only examples: 3 4

(Also, 8355 is basically Intuitive Commutators: The Method, so you already know how to use commutators!)

Last edited:

##### Member
Looks like a variant of M-CELL to me. Steps 1 and 2 are SSF2L, 3 and 4 are normal WV (though this is just a stepping stone to L5C/TCLL in M-CELL) and PLL (also an intermediate step in some of the variants).

It's not too bad to be honest and I have a 14-15 global average with it a not quite fully advanced method that has all the algorithms for a fully optimised variant (maybe less now; I don't really practise it specifically- I'm pretty sure that's the UWR though) though it would be interesting to see someone specialise in it so go for it!

#### Loiloiloi

##### Member
If you leave the buffer corner to the last slot, you can use a variant of VHLS that doesn't need to care about corners. Alternatively, you can also do Petrus-style edge orientation with the last slot (most cases are either one trigger or two triggers that cancel into each other). It's all fundamentally the same idea, except that VHLS is more restricted (and hence takes more moves) because it needs to preserve the pair.
I'm a bit lost, what variant are you referring to? Is there somewhere I can find the algs? Or are you just speaking hypothetically?

#### xyzzy

##### Member
I'm a bit lost, what variant are you referring to? Is there somewhere I can find the algs? Or are you just speaking hypothetically?
Just hypothetically, but I think it's better done intuitively than algorithmically. (Except maybe for some of the 4-flip cases, which are nasty no matter what you do.) The point is that it's not too hard to learn how to do this intuitively.

#### VenomCubing

##### Member
Looks like a variant of M-CELL to me. Steps 1 and 2 are SSF2L, 3 and 4 are normal WV (though this is just a stepping stone to L5C/TCLL in M-CELL) and PLL (also an intermediate step in some of the variants).

It's not too bad to be honest and I have a 14-15 global average with it a not quite fully advanced method that has all the algorithms for a fully optimised variant (maybe less now; I don't really practise it specifically- I'm pretty sure that's the UWR though) though it would be interesting to see someone specialise in it so go for it!
Now that I think of it, GS is like M-CELL, and revolves around the same idea. GS just goes about that idea differently.

On a side note, I can't believe you like my method! I was hoping someone with a lot of knowledge would see my thread. The approval of someone like you really means a lot to me. I'm glad to know I'm on the right track.

#### Metallic Silver

##### Member
There's a document with all cases to solve last layer corners + U center while inserting the last corner, but it's 90 algs and it's almost always faster to do layer then one alg

The EGOLL algs are decent, there's a document with nice algs but it's rarely worth it to use them if you want to be really fast
0o0 i wanna see thos algs, can u send them plz?

#### Neuro

##### Member
I created a new method for 3x3. I call it Perry !
The step:
S1: Make b-block + last edge in E-slice ( seems like 3x2x1 block in LD+ 2x2x2 block in BR) (intuitive)
S2: Orient
a) Orient corner (23 cases)
b) Orient edge (intuitive)
S3: Make 2x2x1 block on UBL (intuitive)
S4: Permute last 4 corner and last 3 edge (about 20 cases + symmetry).
Đỗ Viên, I really like the concept, and I generated all the possible algs needed for the last step (lets call it Perry Last Corner-Edge [PCLE].)
There are 4 main categories in PCLE:
• "Pure"-both a corner and an edge need to be put in
• "HKPLL"-only edge insertion
• "TTLL"-only corner insertion
• "PLL"- Ja, Jb, V, Aa, Ab
I define the pure cases using the bottom edge/corner in relation to the UBL block (R=right edge, F=UFR corner) and blocks on top.
HKPLL is mostly identified by corners/blocks and basically same story with TTLL.

Pure has 36 algs
HKPLL has 9 algs
TTLL has 18 algs
PLL has 5 algs (didn't generate these, didn't see a reason as basically everyone knows PLL)

In total there are 68 algs, hopefully I didn't miss any cases. I'd say it's within reason to learn, but you'd probably need a solid recognition system if used for speedcubing. As of now, the algs have been created without the F/B faces, so move-count may be slightly better than whats given here. Keep looking into it, I think you've got some good potential here!

#### Attachments

• 3.2 KB Views: 11