The WaterRoux 3x3 method thread

[Deleted member 36445]

EDIT: As far as L8E, the only reasonable approach I've seen was EZD. Which requires edges to be oriented and separated to their correct layers first-- and runs into possible parity cases (fixed via M2 U2 M2 although it can be hard to cancel into this sequence).

where can I find the EZD algs?

I found out how to solve any L8EP 2 Look. you need to know EPLL and the "Opposite-Cross ELL" cases where all edges are oriented
https://meep.cubing.net/ocell.html

it's 9 algs total (EPLL+OCELL EP), but you'll need a x2 or z2 rotation

I'm going to publish a document soon

 

[efattah]

The LMCF-H method of solving L8E is pretty good and I strongly doubt something better exists.
If you want to solve by the method you suggest it would be 3 steps:
1. orient 4 U & 4 D edges
2. Separate U & D edges into their respective layers
3. permute last 8 edges

I don't see how you could possibly do that in 2 looks...

 

[Deleted member 36445]

The LMCF-H method of solving L8E is pretty good and I strongly doubt something better exists.
If you want to solve by the method you suggest it would be 3 steps:
1. orient 4 U & 4 D edges
2. Separate U & D edges into their respective layers
3. permute last 8 edges

I don't see how you could possibly do that in 2 looks...

the method I'm developing has the edges pre oriented and separated so you just have to permute on top and on bottom and you have some kind of parity

 

[Silky]

where can I find the EZD algs?

I found out how to solve any L8EP 2 Look. you need to know EPLL and the "Opposite-Cross ELL" cases where all edges are oriented
https://meep.cubing.net/ocell.html

it's 9 algs total (EPLL+OCELL EP), but you'll need a x2 or z2 rotation

I'm going to publish a document soon

EZD algs can be found here here (w/ and w/o parity. To avoid parity just do M2 U2 M2 ). I was re-genning the algs to find some that I like/ to make an SSC mega-doc but I still have to finish it. I'll link it whenever I finish it. I was also working on an EZD+1 algset but I've yet to put much time into it.

 

[Zetko]

Hi!

I am learning WaterRoux. I don't really care about alg count but i still agree with Athefre that with 250+ algs some cases appear really rarely.

Is this method still promising? And if it is how does it work?

What would be the best method for solving L7E or L8E...?

 

[efattah]

I am using LMCF-H which is almost the same as WaterRoux, the only difference is LMCF-H you solve Roux first block minus the DL edge (creating an H on the left side), and then solve all the corners with LEG1/CLL. I have optimized the LEG1 algs to make sure they don't break Roux-FB. If you can one look the first block & corners during inspection, the method has crazy potential. With LMCF-H the one look is easier, only 2 edges to solve plus corners. You can solve 1 edge usually by doing wide moves during face creation for the corners.

Once the H & corners are solved, I use pair & triplet algs from LMCF which have been greatly evolved in the last 6 years since the last document I posted. Move count averages around 44-45. I can post the LEG1 algs for use in WaterRoux if you want.

 

[T_adapledge]

I am using LMCF-H which is almost the same as WaterRoux, the only difference is LMCF-H you solve Roux first block minus the DL edge (creating an H on the left side), and then solve all the corners with LEG1/CLL. I have optimized the LEG1 algs to make sure they don't break Roux-FB. If you can one look the first block & corners during inspection, the method has crazy potential. With LMCF-H the one look is easier, only 2 edges to solve plus corners. You can solve 1 edge usually by doing wide moves during face creation for the corners.

Once the H & corners are solved, I use pair & triplet algs from LMCF which have been greatly evolved in the last 6 years since the last document I posted. Move count averages around 44-45. I can post the LEG1 algs for use in WaterRoux if you want.

Could you post the algs plz

 

[Zetko]

I really want to use WaterRoux as my main and i am prepared to learn to inspect block + corners.

How should i solve l7e than? How far can we go with efficiency in this step?
I can learn a lot of algorithms. It would be nice if there would be more sets with smaller alg count for easier drilling. I can learn like idk 400 algs but just if they are in sets. For example 20, 30 algs per set. Is that possible in this case?

Now i am using Roux. Is there any chance of method neutrality like switching between Roux, WaterRoux, 42... or its best for me to just stick with WaterRoux?

Any tips with switching to WaterRoux?

And for LMCF... I was really interested in this method past months. But i still think ergonomics are questionable (no offense).
Algs like B U M' U' B' or those ones with F and M moves? Are there good ways to fingertrick them (without huge regrip)...?

And in general... WaterRoux is better than LMCF because of ergonomics...?

I read that for LMCF 30 move solution appears everytime in 1 in 500 solves. Whar about WaterRoux? Is move average below 35 common...?

I know there's a LOT of questions, so HUGE thanks to everyone for responding!!!

These questions keep me awake...

 

[Zetko]

In last couple of months i tried to find the best method. I came back to Roux variants...

So may I ask @Athefre :
What is going on woth APB and Nautilus? As you said you use H LMCF.
Do you think methods with Roux 4c for the last step are better than methods with last layer, l5c...

 

[Athefre]

In last couple of months i tried to find the best method. I came back to Roux variants...

So may I ask @Athefre :
What is going on woth APB and Nautilus? As you said you use H LMCF.
Do you think methods with Roux 4c for the last step are better than methods with last layer, l5c...

APB and Nautilus are both fully developed and there for the community to start using. There are two problems. One is that we don't yet have a conclusive answer to the debate of intuition vs algorithms. Both methods are based on the idea that memorized algorithms can have a higher TPS and a more consistent move count than relying upon intuition. So algorithmic L5E in Nautilus vs intuitive LSE in Roux and algorithmic EO + RB in APB vs intuition in Petrus. But this concept is under-tested so we don't yet know. The other problem is that people are resistant to change and are afraid to use things that aren't yet proven. So adoption of new methods is extremely slow.

I'm a Roux user by the way. Always have been. I don't solve with LMCF.

 

[OreKehStrah]

In last couple of months i tried to find the best method. I came back to Roux variants...

So may I ask @Athefre :
What is going on woth APB and Nautilus? As you said you use H LMCF.
Do you think methods with Roux 4c for the last step are better than methods with last layer, l5c...

Ettafah is the lmcf guy.

I really want to use WaterRoux as my main and i am prepared to learn to inspect block + corners.

How should i solve l7e than? How far can we go with efficiency in this step?
I can learn a lot of algorithms. It would be nice if there would be more sets with smaller alg count for easier drilling. I can learn like idk 400 algs but just if they are in sets. For example 20, 30 algs per set. Is that possible in this case?

Now i am using Roux. Is there any chance of method neutrality like switching between Roux, WaterRoux, 42... or its best for me to just stick with WaterRoux?

Any tips with switching to WaterRoux?

And for LMCF... I was really interested in this method past months. But i still think ergonomics are questionable (no offense).
Algs like B U M' U' B' or those ones with F and M moves? Are there good ways to fingertrick them (without huge regrip)...?

And in general... WaterRoux is better than LMCF because of ergonomics...?

I read that for LMCF 30 move solution appears everytime in 1 in 500 solves. Whar about WaterRoux? Is move average below 35 common...?

I know there's a LOT of questions, so HUGE thanks to everyone for responding!!!

These questions keep me awake...

You should look into the L5C + L7E variant of Roux if you want to learn a lot of algs

 

[Silky]

Ettafah is the lmcf guy.

You should look into the L5C + L7E variant of Roux if you want to learn a lot of algs

WaterRoux with L5C also looks like it could be promising.

 

[efattah]

The LMCF document (v4.5) published in 2017 was a very early stab at the method and the algorithms in that document have very poor ergonomics, plus, that method used tons of rotations. In the last 6 years I have developed two paths, the original LMCF with improved ergonomics, and LMCF-H, both methods attempt to reach the columns state as quickly as possible and then use UMD pairs & triplets to reach L6E and then finish with Waterman L6E.

The LMCF-H variant usually has no rotations, or sometimes 1 rotation. I would estimate the present speed potential as approximately the same as Roux; if you compare Roux & LMCF-H, the Roux solve suffers from the pause at CMLL recognition, which LMCF-H does not, because it puts the recognition into the inspection phase. This seems to give LMCF-H an advantage, as it has the same move count as Roux; however, slightly faster TPS on Roux second block makes up for this disadvantage, giving both methods approximately equal speed potential.

 

[Silky]

The LMCF document (v4.5) published in 2017 was a very early stab at the method and the algorithms in that document have very poor ergonomics, plus, that method used tons of rotations. In the last 6 years I have developed two paths, the original LMCF with improved ergonomics, and LMCF-H, both methods attempt to reach the columns state as quickly as possible and then use UMD pairs & triplets to reach L6E and then finish with Waterman L6E.

The LMCF-H variant usually has no rotations, or sometimes 1 rotation. I would estimate the present speed potential as approximately the same as Roux; if you compare Roux & LMCF-H, the Roux solve suffers from the pause at CMLL recognition, which LMCF-H does not, because it puts the recognition into the inspection phase. This seems to give LMCF-H an advantage, as it has the same move count as Roux; however, slightly faster TPS on Roux second block makes up for this disadvantage, giving both methods approximately equal speed potential.

Do you plan to release an updated document?

 

[efattah]

I do hope to release an update, I have had ten or more people ask for it. Unfortunately it is a fairly large task as pretty much all of the 600+ algorithms in the document have been revised and major steps have changed so it is basically re-writing the document from scratch.
Aside from the curiosity & fun of using a 'different' method, if you consider that LMCF-H and Roux presently have similar potential, one could argue the drawback is LMCF-H requires 500 algorithms to master while Roux only needs around 100. However, LMCF-H is not fully developed, and new breakthroughs are being made all the time. If ten people were developing the method at a high level, it would only take a few more breakthroughs to finally surpass Roux. I won't speak at all into surpassing CFOP as Yiheng, Max & Tymon have taken CFOP to such a level that I am starting to wonder if any method can ever overtake it.