The New Method / Substep / Concept Idea Thread

Randomno said:

Unpermuted cross > F2L + EO > COLL > last corners.

Click to expand...

But Randomno, the last corners were solved in the COLL step.

TomasCZ said:

I know, that this is not the first method, which skips cross edge permutation.


Hi guys !
Sorry for my english.
I have created a method, which allows me to have a shorter cross (in some cases)
I would like to explain briefly what this method basically does: It orients LL corners ans simultaneously permutes cross edges.

All substeps I had known only solved LL edges or corners, but I thought, that there should be a method, which does something with cross.
So basically it allows me to have 2 edges on cross swapped, and still solve the cube normally. Sometimes cross is VERY easy when I dont care about edge permutation.

So the solve would look like this:

cross (2 edges swapped)
F2L + edge orientation
OLL + cross edge permutation
PLL


Pros:
-only 16 algorithms (basic method)
-good lookahead for F2L
-better cross-F2L transition
-shorter cross

Cons:
-edge control is needed
-algorithms average cca 11 moves


Algs: View attachment 5084


I was also considering permuting the edges, when doing PLL or full OLL, but that would need a large alg set.
Also, all 6 edges could be solved with corners orientation (large alg set again)



What do you guys think of this ? What are your opinions ?

Click to expand...

good job coming up with some algs that are actually nice to preform. but isn't the cross usually pretty easy to make anyway? if you are opposite colour neutral i dont think there would ever be a cross bad enough to justify this. maybe this could be useful for fixed cross colour solvers?? on a related topic, a few weeks ago i actually decided to generate a bunch of algs to swap two D layer edges and permute the last layer. im pretty sure i got them all done, but some of them are far from fingertrick friendly. i'll return with an update on that when i next access that hard-drive.

in all, my experiences tell me that the best methods are both straight forward, and relatively efficient(move-cont efficient, look-ahead efficient, regrip efficient, etc). intentionally solving the cross incorrectly is not straight-forward (it is rather counter intuitive, really) and this over complication is detrimental to its efficiency (having to position the D layer before executing the alg, and it might mess up f2l look ahead).
but experimental method is experimental. who knows what may become of it.

Ross The Boss said:

good job coming up with some algs that are actually nice to preform. but isn't the cross usually pretty easy to make anyway? if you are opposite colour neutral i dont think there would ever be a cross bad enough to justify this. maybe this could be useful for fixed cross colour solvers?? on a related topic, a few weeks ago i actually decided to generate a bunch of algs to swap two D layer edges and permute the last layer. im pretty sure i got them all done, but some of them are far from fingertrick friendly. i'll return with an update on that when i next access that hard-drive.

in all, my experiences tell me that the best methods are both straight forward, and relatively efficient(move-cont efficient, look-ahead efficient, regrip efficient, etc). intentionally solving the cross incorrectly is not straight-forward (it is rather counter intuitive, really) and this over complication is detrimental to its efficiency (having to position the D layer before executing the alg, and it might mess up f2l look ahead).
but experimental method is experimental. who knows what may become of it.

Click to expand...

It's actually pretty useful to do pseudo cross with sufficient practice, especially if you do a lot of CP during OLL. The usual finish is a "parity" EPLL alg.

Just a slight variation to Yau on bigger cubes, maybe it's been done before.
1) solve cross+centers à la Yau
2) freeslice 4 edges and place them on the top layer
3) finish the last 4 edges like L4E with redux.
4) 3x3 stage

I've been playing with this and I can get pretty close to my normal 5x5 timings.

Thoughts?

Sooo I post a lot of things that get rejected here, but have come up with a method that has semi-efficient movecount (more than Roux, Less than CFOP) at least for me
Step 1.EO Cross
Step 2. Solve All E-slice pieces
Step 3. Permute LL edges
Step 4. Solve All Corners using about 40,000 Algs ( I think)
LOTS AND LOTS OF ALGS, but there are ways to reduce it .

Soo I have brought up this method before, but I have recently modified it.
Step 1. 2x2x3 Block
Step 2. EO
Step 3. Solve BL F2L Pair
Step 4. L5C while preserving EO
Step 5. L5E while preserving CO
Also, Because I would like to find algs for this method, could someone tell me how many algs there would be, and if it would be worth it to learn them all

2180161 said:

Soo I have brought up this method before, but I have recently modified it.
Step 1. 2x2x3 Block
Step 2. EO
Step 3. Solve BL F2L Pair
Step 4. L5C while preserving EO
Step 5. L5E while preserving CO
Also, Because I would like to find algs for this method, could someone tell me how many algs there would be, and if it would be worth it to learn them all

Click to expand...

Both would be well in excess of 100. Definitely not worth it to learn them all.

Well, the L5C would basically be COLL+1, which I was told is ~96 Algs, and then for the L5E, It would end up being EPLL+1, which would be around 120 algs I believe, so It would be a total of 216 Algs and some subsets have much more than than like OLLCP and Full EG.

2180161 said:

Well, the L5C would basically be COLL+1, which I was told is ~96 Algs, and then for the L5E, It would end up being EPLL+1, which would be around 120 algs I believe, so It would be a total of 216 Algs and some subsets have much more than than like OLLCP and Full EG.

Click to expand...

COLL+1 is very different from L5C, it's COLL, and solving a single LL edge, which is just learning one extra alg for each COLL case (it's 80 algs), while L5C is 5 corners. L5C would have as many algs as full EG (126, if I haven't counted wrong) for cases when the bottom corner is permuted alone (depending on its orientation, think CLL and TCLL), never mind the number of cases where it's in the top layer. I don't know how many algs L5E is, but this method would require huge numbers of algs.

How to use OP (Owen Petrus) Method

So, my friend owen has invented a new method. (Sort of) There is a back-story behind this which I will explain to you.

So, I have been learning Petrus method, and I told my friend owen about it. Problem was, I only remembered the first two steps, the 2 by 2 block and 2 by 2 by 3 block. I told him this and he tried it. After doing so, he said, "Oh, i get it, Petrus method is a strange way of doing 2 f2l pairs and most of the cross." I told him that was nowhere near it, but he replied " Well in my petrus method that is how it works." And so he called it OP method, for Owen petrus method. I will explain it to you.

You begin with doing the 2 by 2 and 2 by 2 by 3 blocks like in regular petrus method, and you should have some 2 F2L pairs and 3 of 4 cross pieces inserted. You finish the cross and the rest of F2L like a normal layer by layer solve, and finish the rest of the cube like it was a regular layer by layer solve. We have both tried it with mixed results. I actually made my record on the 3 by 3 with this, and I am wondering how the rest of you might do with this.

EO + blockbuild the remaining F2L would be a lot more efficient though...

This has been invented so many times and is not as good as CFOP or Petrus in most situations.

Also OP is the name for another method, so don't call this method OP.

TDM said:

This has been invented so many times and is not as good as CFOP or Petrus in most situations.

Also OP is the name for another method, so don't call this method OP.

Click to expand...

yeah old pochmann

theROUXbiksCube said:

yeah old pochmann

Click to expand...

i was also thinking Original Post but that works too...

quickalt said:

How is Original Post a method?

Click to expand...

lol idk i was confused reading the title...

TDM said:

This has been invented so many times and is not as good as CFOP or Petrus in most situations.

Also OP is the name for another method, so don't call this method OP.

Click to expand...

I know, Owen found it funny, and so it stuck. I preferred witpot, for what is the point of this. The funny name wasn't enough.

Ok, so here goes the idea for a method for 3x3
I feel that this could be used for FMC but maybe not who knows?

1. Cross
2.F2L-1
3. EO (because you did F2L-1 there will only be a max of 4, so less algs to learn to fix this)
4.F2L-1E
5. COLL
6. L5E
An example Solve:
Scramble:U F2 L F2 L2 F' D' F2 B' D2 B R2 D2 R2 B' D2 U
Solution:
U D F L F2 //2x2
U2 R' U' R2 U R U R' U2 R U' R' d2 U M' U2 M //F2L-1
U' M' U M U2 M' U M//EO
U2 R U R' //F2L-1E
U2 R U2 R2 U' R2 U' R2 U2 R// COLL
U R U' R' F2 U' L' U L F2 //L5E

2180161 said:

Ok, so here goes the idea for a method for 3x3
I feel that this could be used for FMC but maybe not who knows?

1. Cross
2.F2L-1
3. EO (because you did F2L-1 there will only be a max of 4, so less algs to learn to fix this)
4.F2L-1E
5. COLL
6. L5E
An example Solve:
Scramble:U F2 L F2 L2 F' D' F2 B' D2 B R2 D2 R2 B' D2 U
Solution:
U D F L F2 //2x2
U2 R' U' R2 U R U R' U2 R U' R' d2 U M' U2 M //F2L-1
U' M' U M U2 M' U M//EO
U2 R U R' //F2L-1E
U2 R U2 R2 U' R2 U' R2 U2 R// COLL
U R U' R' F2 U' L' U L F2 //L5E

Click to expand...

:tu

you can combine steps 3 and 4

place the corner at UFR (3 orientations)

check the EO (11 misoriented possibilities)

so thats only 33 algs (plus the ones with solved EO, but I don't consider those "algs")

so i made the algs for it in cubeexplorer. Now, to optimize them all...

I just came up with this concept. I haven't fully hashed it out yet, and I wanted to throw it out here for some input. I'm thinking about calling it WF2L

Personally, this is how I solve 3x3:
1. EOLine
2. Finish F2L
3. ZBLL

Step three is divided into four main substeps:
A. Look at the rotation of the case, and rotate it into recognisable state by using U, U', or U2.
B. Once rotated, figure out the ZBLL case.
C. Execute alg.
D. Adjust the U layer with a U, U', or U2


I realised that substep A can be combined with the final F2L slot by including a lot of F2L algs.
I think if I can search for the easy cases, this would really help out!

For example:

U' B' F U' R2 U R U' R U F' R B

If I were to do the normal approach with U R U' R', then I'd have to do look at the alg, then do a U, and then look at it again.

However! If I were to do a U2 R U2 R', I can skip that nonsense and go straight into recognition, AND I'd know the corner orientation before I even finish F2L.

I've considered other options to remove that stupid substep A such as:
1. Using a conversion factor to recognise ZBLL from different angles.
2. Using alternate recognition styles.
3. Learning alternate algs to do from other angles. (Ew)

tl;dr: By learning lots of F2L cases, one can omit a ZBLL substep and recognise corner orientation before finishing F2L.

EDIT: Under further consideration, that's A LOT OF ALGS.
I think just learning to recognise from different angles is a better approach.