The New Method / Substep / Concept Idea Thread

[Miika T]

I'm trying to compile an alternate to the standard 2 Look OLL (for CFOP).

Instead of having three algorithms for orienting edges and then seven different ones for the corners, I wanted to investigate if allowing other possible states after the first algorithm could lead to reusing some of the same moves. For example, some of the dot OLLs can be solved by applying a P-OLL algorithm twice. (The downside is that you must remember exactly from which orientation to start.) My reasons for this where to possibly cut down the number of algorithms needed, trying to shorten the average moves needed to reach the PLL step without actually doing a full OLL, and to have fun attempting to explore this approach.

Has something like this been proposed somewhere? I don't know if it will end up being a useful method for anyone, but at least I'm discovering my own order of learning the OLLs. Also, on the theoretical side, it would be interesting to know if the number of algorithms could be minimized from the standard 2L OLL.

 

[Shiv3r]

This is the thread that was posted a couple of months ago

What does the "OLL" look like? I don't think the algs will be very nice...

I need someone to help me with the case number for OLL. how many cases are there for CSLL? and what about S1PLL (21 without parity, maybe more with parity?)

 

[genericcuber666]

anyone else think that cryoos method has potential im not good at roux but im learning zzct and i think i could get sub 20s with it

 

[Daniel Lin]

I have an idea for Roux that is kinda similar to Pinky Pie but different

If all your edges are oriented when you get to CMLL, put your LR edges on DF and DB. Then do a ZBLL alg to change the edge permutation on the top so that when you place your LR edges in you get a 4c "skip".
i'll include an example later

 

[Berd]

I have an idea for Roux that is kinda similar to Pinky Pie but different

If all your edges are oriented when you get to CMLL, put your LR edges on DF and DB. Then do a ZBLL alg to change the edge permutation on the top so that when you place your LR edges in you get a 4c "skip".
i'll include an example later

Isn't a eo skip quite rare, rarer than to justify learning full zbll?

 

[Daniel Lin]

Isn't a eo skip quite rare, rarer than to justify learning full zbll?

oh yeah true. It's 1/10 (i think)
maybe with tricks you can make the chance higher?

 

[shadowslice e]

oh yeah true. It's 1/10 (i think)
maybe with tricks you can make the chance higher?

Well, the thing is, you can already force a 4c skip maybe 10-15% of the time if you really try or keep the 3 movers about 50-60%, likely more if you really wanted to so I don't think the extra moves warrented by ZBLL (not to mention the alg increase) would also not help. The extra recog time isn't great either. The only practical ways to increase the EO skip is either ZBLS (or VHLS but less so as it is much more inefficient) or Eo before FB (which increases movecount for FB and SB as well as reducing lookahead).

 

[obelisk477]

I'm trying to compile an alternate to the standard 2 Look OLL (for CFOP).

Instead of having three algorithms for orienting edges and then seven different ones for the corners, I wanted to investigate if allowing other possible states after the first algorithm could lead to reusing some of the same moves. For example, some of the dot OLLs can be solved by applying a P-OLL algorithm twice. (The downside is that you must remember exactly from which orientation to start.) My reasons for this where to possibly cut down the number of algorithms needed, trying to shorten the average moves needed to reach the PLL step without actually doing a full OLL, and to have fun attempting to explore this approach.

Has something like this been proposed somewhere? I don't know if it will end up being a useful method for anyone, but at least I'm discovering my own order of learning the OLLs. Also, on the theoretical side, it would be interesting to know if the number of algorithms could be minimized from the standard 2L OLL.

There's this:
https://www.speedsolving.com/forum/threads/suneoll.23222/

 

[TDM]

oh yeah true. It's 1/10 (i think)
maybe with tricks you can make the chance higher?

1/32. I was trying to think of tricks to make the chances higher, but then it turned into Pinkie Pie. I think you might as well just use that.

 

[Miika T]

There's this:
https://www.speedsolving.com/forum/threads/suneoll.23222/

Thanks! That's pretty much what I was looking for My alternate 2 look OLL approach starts from some different algorithms and the objective is a bit more aimed for a beginner, but the basic idea is similar.

I found that using basically three algorithms and their reverses (and some mirrors too, but would have to go through again to check which ones were required) were enough to handle every OLL case. Remembering which one to use is the hard part, but this is helped by 46 cases starting with either the P-case algorithm or its reverse (T), and then just reacting to what comes up afterwards. The other algorithms I chose were the sune and couch algorithm, so they are all pretty simple and a combination of two will never be very long.

If I write this up for others, is this the right place to post it or would some other place be better? Also, what are the chances that this is actually interesting to anyone else besides myself?

 

[jjone fiffier]

Looks cool, how many cases would it be if you did CP+ bottom 2 edges in one go instead of doing CP+M2 setup?

It's 26 I think.
But 1. I have no idea how to generate sq1 algs and 2. I'm fine with averaging high16.
Bump btw.

Gesendet von meinem LG-D331 mit Tapatalk

 

[GenTheSnail]

ZZ Megaminx

1. F2L done with block building and "normal" cfop pairs.
2. Two adjacent balint blocks in the front + the pair in between
3. A balint block in the U-layer that is over the pair
3a. x' so the U-layer balint block is the the front
5. EO of the last three sides
5. Blockbuild the two R/L sides
6. 3LL

I did a couple solves with this, and EO recog isn't that hard, though it might be overly time consuming to be worth using.
Thoughts?

[yes, I saw that shadowslice proposed an EO mega method a year ago, but it didn't sound anything like this]

 

[shadowslice e]

ZZ Megaminx

1. F2L done with block building and "normal" cfop pairs.
2. Two adjacent balint blocks in the front + the pair in between
3. A balint block in the U-layer that is over the pair
3a. x' so the U-layer balint block is the the front
5. EO of the last three sides
5. Blockbuild the two R/L sides
6. 3LL

I did a couple solves with this, and EO recog isn't that hard, though it might be overly time consuming to be worth using.
Thoughts?

[yes, I saw that shadowslice proposed an EO mega method a year ago, but it didn't sound anything like this]

It seems similar to one of the variants I proposed but after some laying around I don't think Eo skip in LL is worth the extra time especially with partial edge control.

 

[GenTheSnail]

It seems similar to one of the variants I proposed but after some [p]laying around I don't think Eo skip in LL is worth the extra time especially with partial edge control.

Hmm, that makes sense.

1. F2L done with block building and "normal" cfop pairs
2. One balint block in the front + the "Spike" edge that connects it to the U layer
3. x' so the U-layer balint block is the the front
4. EO of the last 4 sides [When doing eo, you can't use the spike face, because that 5th side can change eo too]
5. Blockbuild half of one of the bottom R/L sides
5a. z(') so there is a top layer
6. blockbuild R/L sides
7. 3LL

With this variation, you get more out of EO, but recognition is so much weirder.
When I figure out the rules, I'll post them here.

EDIT:
So they seem to be basic EO laws, but just slightly more confusing to recognize.
Faces are labeled 1, 2, 3, 4 from left to right.

If an odd numbered colour is touching an odd numbered face, it is good. If it is touching an even numbered face, it is bad.

If an even numbered colour is touching an even numbered face, it is good. If it is touching an odd numbered face, it is bad.

That's it.

E4: Well, a little bit more,

If an odd numbered colour is on an even numbered face, but they are not touching, it is good.

If an even numbered colour is on an odd numbered face, but they are not touching, it is good.

If an even numbered colour is on an even numbered face, and they are not touching, it is bad

If an odd numbered colour is on an odd numbered face, and they are not touching, it is bad

EDIT2: I did some solves with the first method and my best was like 2:20. I average just at ~2:10 with pure balint.
I'll try out the v2 method and see if I can get a sub-2 after I get used to EO. I think it would be cool If I could actually compete with it at dixon.

E3: I'm calling this ZZ-Spike

 

[Shiv3r]

Hmm, that makes sense.

1. F2L done with block building and "normal" cfop pairs
2. One balint block in the front + the edge that conects it to the U layer
3. x' so the U-layer balint block is the the front
5. EO of the last 4 sides
5. Blockbuild half of one of the bottom R/L sides
5a. z(') so there is a top layer
6. blockbuild R/L sides
6. 3LL

With this variation, you get more out of EO, but recognition is so much weirder.
When I figure out the rules, I'll post them here.

If I had any reason to learn 3LLL, I would totally learn this.

 

[GenTheSnail]

Hmm, that makes sense.

1. F2L done with block building and "normal" cfop pairs
2. One balint block in the front + the "Spike" edge that connects it to the U layer
3. x' so the U-layer balint block is the the front
4. EO of the last 4 sides [When doing eo, you can't use the spike face, because that 5th side can change eo too]
5. Blockbuild half of one of the bottom R/L sides
5a. z(') so there is a top layer
6. blockbuild R/L sides
7. 3LL

With this variation, you get more out of EO, but recognition is so much weirder.
When I figure out the rules, I'll post them here.

EDIT:
So they seem to be basic EO laws, but just slightly more confusing to recognize.
Faces are labeled 1, 2, 3, 4 from left to right.

If an odd numbered colour is touching an odd numbered face, it is good. If it is touching an even numbered face, it is bad.

If an even numbered colour is touching an even numbered face, it is good. If it is touching an odd numbered face, it is bad.

That's it.

E4: Well, a little bit more,

If an odd numbered colour is on an even numbered face, but they are not touching, it is good.

If an even numbered colour is on an odd numbered face, but they are not touching, it is good.

If an even numbered colour is on an even numbered face, and they are not touching, it is bad

If an odd numbered colour is on an odd numbered face, and they are not touching, it is bad

EDIT2: I did some solves with the first method and my best was like 2:20. I average just at ~2:10 with pure balint.
I'll try out the v2 method and see if I can get a sub-2 after I get used to EO. I think it would be cool If I could actually compete with it at dixon.

E3: I'm calling this ZZ-Spike

So I've done a couple solves with this method, and kinda timed them. I realized that if the ZZ part wasn't faster, or at least not considerably slower, than balint S2L then this would be pretty pointless.
So the first five solves are with ZZ-Spike, and I timed the splits.
First split is Blockbuilt F2L + the spike (which I always do as green).
Second split I didn't care about going fast on was EO, I just wanted to see how S2L compared to balint. I went really slowly and made sure all edges were oriented.
Last split is the Last 4 Layers L4F. I realized that (I) should always do the left bottom side and then do a z', which will leave me with cream as LL, but for more colour neutral solvers, they could end up with any of the sides as their LL.

49.14--2:33.82--1:15.80
53.36--1:05.47--1:38.08
54.56--1:47.03--1:15.35
48.15----36.38--1:15.77
47.44----52.33--1:11.83

------------------1:07.62
------------------1:23.12
------------------1:10.97
------------------1:09.76
------------------1:25.33
These last stats are just some pure S2L balint solves + one pre-made balint block to make it fair.

EDIT: I did an ao12 and timed the splits.
Also, I got a sub-2!

Spoiler: Ao12+Splits

Generated By csTimer on 2016-9-14
solves/total: 12/12

single
best: 1:59.46
worst: 2:35.70

mean of 3
current: 2:14.64 (σ = 12.88)
best: 2:07.18 (σ = 7.31)

avg of 5
current: 2:14.01 (σ = 6.24)
best: 2:12.94 (σ = 5.39)

avg of 12
current: 2:14.17 (σ = 7.23)
best: 2:14.17 (σ = 7.23)

Average: 2:14.17 (σ = 7.23)
Mean: 2:14.74

Time List:
---Time---F2L+Spike--EO-----LFL----LL---
1. 1:59.46 - 39.97 / 20.35 / 45.04 / 14.08
2. 2:13.99 - 39.05 / 21.53 / 58.61 / 14.78
3. 2:08.08 - 46.09 / 12.21 / 51.23 / 18.54
4. 2:17.41 - 43.23 / 16.93 / 1:02.65 / 14.58
5. 2:19.49 - 47.80 / 26.04 / 38.84 / 26.79
6. 2:35.70 - 43.14 / 15.92 / 1:19.55 / 17.08
7. 2:14.96 - 48.56 / 25.50 / 46.41 / 14.47
8. 2:17.02 - 49.91 / 29.25 / 44.29 / 13.55
9. 2:06.82 - 46.97 / 19.58 / 46.22 / 14.04
10. 2:00.36 - 40.66 / 24.26 / 35.68 / 19.74
11. 2:25.38 - 49.89 / 23.95 / 55.59 / 15.94
12. 2:18.17 - 50.06 / 16.46 / 56.30 / 15.33

Mean. 2:14.74 - 45.44 / 21.00 / 51.71 / 16.58
Best possible time. 1:40.49 - 39.05 / 12.21 / 35.68 / 13.55

Do you guys think this is a viable method?
Do you want me to make a walk through solve video?

 

[weatherman223]

So I've done a couple solves with this method, and kinda timed them. I realized that if the ZZ part wasn't faster, or at least not considerably slower, than balint S2L then this would be pretty pointless.
So the first five solves are with ZZ-Spike, and I timed the splits.
First split is Blockbuilt F2L + the spike (which I always do as green).
Second split I didn't care about going fast on was EO, I just wanted to see how S2L compared to balint. I went really slowly and made sure all edges were oriented.
Last split is the Last 4 Layers L4F. I realized that (I) should always do the left bottom side and then do a z', which will leave me with cream as LL, but for more colour neutral solvers, they could end up with any of the sides as their LL.

49.14--2:33.82--1:15.80
53.36--1:05.47--1:38.08
54.56--1:47.03--1:15.35
48.15----36.38--1:15.77
47.44----52.33--1:11.83

------------------1:07.62
------------------1:23.12
------------------1:10.97
------------------1:09.76
------------------1:25.33
These last stats are just some pure S2L balint solves + one pre-made balint block to make it fair.

EDIT: I did an ao12 and timed the splits.
Also, I got a sub-2!

Spoiler: Ao12+Splits

Generated By csTimer on 2016-9-14
solves/total: 12/12

single
best: 1:59.46
worst: 2:35.70

mean of 3
current: 2:14.64 (σ = 12.88)
best: 2:07.18 (σ = 7.31)

avg of 5
current: 2:14.01 (σ = 6.24)
best: 2:12.94 (σ = 5.39)

avg of 12
current: 2:14.17 (σ = 7.23)
best: 2:14.17 (σ = 7.23)

Average: 2:14.17 (σ = 7.23)
Mean: 2:14.74

Time List:
---Time---F2L+Spike--EO-----LFL----LL---
1. 1:59.46 - 39.97 / 20.35 / 45.04 / 14.08
2. 2:13.99 - 39.05 / 21.53 / 58.61 / 14.78
3. 2:08.08 - 46.09 / 12.21 / 51.23 / 18.54
4. 2:17.41 - 43.23 / 16.93 / 1:02.65 / 14.58
5. 2:19.49 - 47.80 / 26.04 / 38.84 / 26.79
6. 2:35.70 - 43.14 / 15.92 / 1:19.55 / 17.08
7. 2:14.96 - 48.56 / 25.50 / 46.41 / 14.47
8. 2:17.02 - 49.91 / 29.25 / 44.29 / 13.55
9. 2:06.82 - 46.97 / 19.58 / 46.22 / 14.04
10. 2:00.36 - 40.66 / 24.26 / 35.68 / 19.74
11. 2:25.38 - 49.89 / 23.95 / 55.59 / 15.94
12. 2:18.17 - 50.06 / 16.46 / 56.30 / 15.33

Mean. 2:14.74 - 45.44 / 21.00 / 51.71 / 16.58
Best possible time. 1:40.49 - 39.05 / 12.21 / 35.68 / 13.55

Do you guys think this is a viable method?
Do you want me to make a walk through solve video?

Yes! Please make a walkthrough! This is very interesting and I would like to see the actual steps being done since I'm not a ZZ solver.

 

[Sion]

I was dabbling with my Square one, and I thought of the inneficciency of common methods, so I came up with my own solution. My method idea tries to put a balance between using cubeshape first and cubeshape last. I call it OCP2.

1: Orient corners and edges in cubeshape (all white on one side, all yellow on the other)

2: Solve cubeshape while preserving the orientation.

3: Do two plls in stead of CP/ EP. Algorithms already exist for this.

 

[Teoidus]

This doesn't seem advantageous over that one thing where you do cubeshape then intuitive separation then permutation

 

[Shiv3r]

I think that with some pre-edgepairing magic like in Yau/Meyer/Hoya, Z4 can be a big competitor to the other 4x4 methods.
Here's my proposal(just a rough idea, if you have suggestions please contribute them):
1.2 opposite centers on L/R
2.orient 4 edges, put 3 on the L side(FR, DR, BR, and DL)
3.solve last 4 centers
4.pair the line edges
5.rest of Z4 edgepairing and solving

pros:
-no rotations after 2 opposite centers
-only RUL moves for ZZF2L after edgepairing
-with parity-included EPLL, permutation parity is directly solved, therefore no extra parity step
cons:
-step 2 can be hard to wrap your head around
-See Z4 for Lookahead and other problems

how do you like it?