Middle layer first method

[kpcube]

So I had and inquirey as to the possibility of a middle layer first method. Does one exist? Is it viable to create one ect...

 

[vcuber13]

well theres the [wiki]belt[/wiki] method

 

[Chrish]

I played with doing the middle layer first, just to solve it a way I had never seem before. The way I did it was very inefficient, but I doubt there is a fast way to do it that would be comparable to fridrich.

 

[Kynit]

You could orient both layers, solve the middle, and solve it like a domino. It's still really not an efficient way.

 

[NeedReality]

well theres the [wiki]belt[/wiki] method

I love how the wiki just bluntly states "this method has no use."

As has been previously mentioned, a middle layer first method just doesn't seem viable or efficient enough for speed cubing.

 

[towwdso]

I do it for fun sometimes, figured it myself too.
Make a belt, separates pieces, orient pieces, permute pieces within their layers
its fun

 

[whauk]

i just made some solves with this..
1. "belt"
2. R2 U R2 for edge seperation trying to get as many corners seperated as well
3. 3* sexy move for remaining CS
4. correct edge parity (just a 2flip)
5. OLL
6. when a corner is still twisted z2 and the fridrich case (if not i just finish PLL and make z2 for solving the other LL)
7. OLL on the other side
8. PLL z2 PLL (sometimes u have to do M2 U2 M2)

number of times: 33/33
best time: 17.67 (OLL skip leaving the M2 U2 M2 case on one D)
worst time: 46.42
best avg5: 24.38 (σ = 1.19)
best avg12: 26.35 (σ = 2.49)
session avg: 29.47 (σ = 4.79)

so i think sub20 is absolutely possible. if you learn some seperation cases and train recognizing fast you can get sub15 i think

 

[towwdso]

Its like solving square-1.

i do:
1 - Belt
2 - ES (intuitively, algin 2 edges and do R2, like on square-1)
3 - CS (one at a time, using RUR'U' or RFR'F')
4A - OLL top, then bottom
4B - fix OLL parity ( (M'U)*3 U (M'U)*3 M' for edge, sexy method technique for corners
5A - PLL parity (U2 M2 U2 M2)
5B - PLL top, then bottom

turn out to be just like whauk. =P

 

[oll+phase+sync]

1. orient an 1x2x3 block i n firstlayer : 6-8 moves very funny
2. place 2 belt edges : 4 -8 moves. : very easy but sometimes quite a lot of moves
3. build an orinted 1x1x2 first layer corner edge pair and place it together with anoter belt edge (can be a shifted block although) 6- 9 meoves
4. last belt edge but keep last 1rst layer corner in D layer 3-5 moves
5. get last corner up into first layer oriented 7-8 moves just 3 cases like RUR'DRUR' or R2UR2UR2U2R2
7. OLL 9moves
8 if current case is no PLL M2U2M2
9 PLL xx PLL

my first concept was not to orient 1st layer but just to enforce a valid OLL case or OFL but movecount was not that much lower and lookup realy worse
maybe just an PLL training methode

 

[Cubezz]

Here's how I solve middle layer first:
Solve M slice (easy, put two opposite cross pieces in, and put the other 2 edges in. Not hard and pretty fast.)
Make cross on L and R face (hardist part of solve, some edge PLLs and OLLs help. Solves all edges. Not that hard though.)
Use Pochmann BLD corner method to solve corners, and finish cube.
Pretty easy, but not too fast.

 

[Cyrus C.]

Doing the middle layer first restricts the cube a lot. Parities aren't fun either. Recognition is meh since you have to check the bottom and top layer. Right now it isn't too good, and probably never will be.

 

[vcuber13]

i do the belt the oll twice then you can separate like l2c on 5x5 then pll twice

 

[oll+phase+sync]

i do the belt the oll twice then you can separate like l2c on 5x5 then pll twice

How do you handle OLL - parity ?

For Edges alone M'UM will fix OLL parity? But for Corners R2 D/D'/D2 R2 ?

 

[vcuber13]

just do like a z then twist 2 corners