Full LMCF 3x3 method now available

Pyjam

In the french community, someone just got this scramble:
L2 F2 R' U2 D2 R' D F B' R' B2 L2 D' R2 U D' B2 R2 L2 U B2

efattah

Member
In the french community, someone just got this scramble:
L2 F2 R' U2 D2 R' D F B' R' B2 L2 D' R2 U D' B2 R2 L2 U B2

LMCF solutions for the above scramble:

Scramble: L2 F2 R' U2 D2 R' D F B' R' B2 L2 D' R2 U D' B2 R2 L2 U B2
z y M E M U M2 // solve green center and two green edges
z U' M' U L2 // solve blue red and set up triplet
U l' U M U' L U' // E2L triplet
L' D' M D L2 M U2 M U' M' U' M U' M' U M2 // L6E BDL set XOO
32 STM
16 moves for E2L phase, 16 moves for L6E

Another solution to the same scramble:
z D' M' D M' D2 M // solve Green/Blue centers and blue orange edge
y M2 U M2 // solve green-orange and green-red edges
z L' U M' U' R2 // green white edge
L2 U' M' U l' L' U' M U // E2L triplet
M2 x U' M U' M' U' M U' // L6E DFL set
31 STM
23 moves E2L, 8 moves L6E

Member
This method's great! I've been trying it out, but since I don't know how to orient the edges while solving the last one, what I do is orient and solve them using Roux once I've solved 3 edges on each side. So for instance, this is how I'd solve the above scramble:

Scramble: L2 F2 R' U2 D2 R' D F B' R' B2 L2 D' R2 U D' B2 R2 L2 U B2
z y M E M U M2 // Solve green center and two green edges
z U' M' U L2 // Solve blue red and set up triplet
U l' U M U' L U' // E2L triplet
L' x2 // Setup for Roux L6E
U M' U M' U M U M' // EO
U2 M' U2 M U M2 U' // L6E
L2 // ALF

alg.cubing.net

efattah

Member
Got video of a 13.29 Mo3 and 14.13 Ao5 today. Not PB's but fastest on cam.

Neuro

Member
Going to do some tests pinning LMCF to my "phase" solving style to see what happens. What I'm thinking is to do corners intuitively and solve the L12P using pairs, triplets, comms and other edge stuff. IDK if I'm gonna break the edge portion down to separate substeps but as of now I don't see much of a reason to.

efattah

Member
Going to do some tests pinning LMCF to my "phase" solving style to see what happens. What I'm thinking is to do corners intuitively and solve the L12P using pairs, triplets, comms and other edge stuff. IDK if I'm gonna break the edge portion down to separate substeps but as of now I don't see much of a reason to.

Good luck, interesting to see what comes of it. A super human person who could solve the 12 edges in no particular pattern (not L/R first), could solve the cube in a ridiculously short number of moves. However recognition and lookahead would be extremely hard.

For what it is worth I am still developing the method and making great progress. The number of tricks I have found is so huge that it will require another huge document. In particular, I am learning different variants of E2L pair/triplet algorithms which if selected on the fly improve the ergonomics. As a result the ergonomics of the E2L phase is radically improving, which was previously one of the weak points of the method.

I am still a comparatively slow solver of the corners. I don't like to practice that segment. But I can now routinely finish the E2L phase in sub-5 seconds and the last six edges in sub-2, and that is still with a relatively low TPS of around 3.5 - 4.

Just Use Roux

U

Underwatercuber

Guest
Just Use Roux
LMCF IS the best! It has high TPS and low movecount

Jokes aside though if you really like roux for low move count then LMCF is actually way better

Attachments

• B7CE3C73-BF26-4416-BE92-595E57D4BCCF.png
183.9 KB · Views: 24

efattah

Member
To be fair Roux is still more developed than LMCF. I would say a Roux solver who knows EOLR and multiple CMLL's to skip bad LSE cases has an advantage over an LMCF solver... but not for long. The progress I am making improving LMCF is so rapid that I'm not sure that Roux will have an advantage for much longer. LMCF has unparalleled flexibility for innovations, speed short cuts and on-going developments which will continue to speed it up over the next while. Even before any of that, I am also re-working the algorithms for better ergonomics & speed and even lower movecount, as well as improving the ergonomics and lookahead in the E2L phase.

1001010101001

Member
To be fair Roux is still more developed than LMCF. I would say a Roux solver who knows EOLR and multiple CMLL's to skip bad LSE cases has an advantage over an LMCF solver... but not for long. The progress I am making improving LMCF is so rapid that I'm not sure that Roux will have an advantage for much longer. LMCF has unparalleled flexibility for innovations, speed short cuts and on-going developments which will continue to speed it up over the next while. Even before any of that, I am also re-working the algorithms for better ergonomics & speed and even lower movecount, as well as improving the ergonomics and lookahead in the E2L phase.
I don't use LMCF because of the waaayyyy high algorithm count. I already have problems learning OLL/PLL so... 400 algs...

Juqe

Member
I don't use LMCF because of the waaayyyy high algorithm count. I already have problems learning OLL/PLL so... 400 algs...
I started learning it 1.5 months ago and the algorithms ( especially the basic set ) are really easy to remember. I am currently learning the advanced set and 50% of the algorithms are just mirrored cases( especially L5E and E2L pairs )! Got a little problem with EG-1/CLL, but I still use Ortega for the corners and my PB is on 20.83 right now ( I solved the corners in 6 seconds ).. so even if you do not know a lot of algorithms, you can get pretty fast too, I definitely recommend you trying it

efattah

Member
Juqe is right, you can get really fast with just the LMCF basic algorithm set (around 25-30 algorithms), you can get sub-10 singles with just that. I believe that makes it the fastest existing method for such a low algorithm count...?

Juqe

Member
I believe that makes it the fastest existing method for such a low algorithm count...?
From my own experience I can definitely agree to that, but rather singles than average.

Juqe

Member
To be fair Roux is still more developed than LMCF. I would say a Roux solver who knows EOLR and multiple CMLL's to skip bad LSE cases has an advantage over an LMCF solver... but not for long. The progress I am making improving LMCF is so rapid that I'm not sure that Roux will have an advantage for much longer. LMCF has unparalleled flexibility for innovations, speed short cuts and on-going developments which will continue to speed it up over the next while. Even before any of that, I am also re-working the algorithms for better ergonomics & speed and even lower movecount, as well as improving the ergonomics and lookahead in the E2L phase.
By the way, when will the new document be published? I think I heard someone saying something about E2L Quadruplets ?

Reed Merrill

Member
Juqe is right, you can get really fast with just the LMCF basic algorithm set (around 25-30 algorithms), you can get sub-10 singles with just that. I believe that makes it the fastest existing method for such a low algorithm count...?
Wouldn't someone who gets sub-10 using ZZ-OCLL/PLL be using just as few algs? If I'm not mistaken, this is not to uncommon (I'm pretty sure Phil Yu does it all the time).

efattah

Member
By the way, when will the new document be published? I think I heard someone saying something about E2L Quadruplets ?

It is still some time away. There are huge amounts of work to be done. In addition to E2L quadruplets, there is also a variant where the L face has two or more edges in place that are oriented but not permuted. In which case you can finish the solve in a different way. I am still optimizing the waterman set 2 algorithms, there are many more sets to optimize.

Wouldn't someone who gets sub-10 using ZZ-OCLL/PLL be using just as few algs? If I'm not mistaken, this is not to uncommon (I'm pretty sure Phil Yu does it all the time).
7 corner orientations then 21 PLL's, plus F2L algorithms I suppose? So 28 algorithms plus F2L algorithms. A very low algorithm count, similar to LMCF basic.

Tao Yu

Member
I think you could average sub 8/9 with roux and two look CMLL (9 algs).

I know that Kian didn't know any EOLR until he was around sub 7 or 8, so I think 2 look CMLL would only add about 1-2 seconds to his time. You could also get some nice singles when you get a CO or CP skip.

Last edited:

Reed Merrill

Member
7 corner orientations then 21 PLL's, plus F2L algorithms I suppose? So 28 algorithms plus F2L algorithms. A very low algorithm count, similar to LMCF basic.

Gotcha! So LMCF can be both one of the most or least algorithmic methods, deepending on how much one is willing to learn? If so, that's pretty cool.