The New Method / Substep / Concept Idea Thread

[Neuro]

Theory: There are some Roux solvers who can plan out FB and DR edge. That's solving 6 pieces. Most can plan out FB, which is 5. In Petrus, you solve 4 and then 3, so 7. What if instead of extending 2x2x2 to a 2x2x3 in Petrus, your first step solved 2x2x2+1 cross edge? Theoretically you could plan it in inspection, as it's 5 pieces being solved (hence Roux FB) And from there you can finish the solve in a ton of different ways while improving the ergonomics of Petrus. Here's a solve of it. Any thoughts? (Used OLLCP because it was nice)

 

[Shiv3r]

or you could just use petrus-ish theory

pootris is better tbh. Pootris is FB->DFDB+centers, and while solving DFDB if you can solve EO sometimes, thats great. But at least tracking DFDB is super easy.

 

[AlphaSheep]

Theory: There are some Roux solvers who can plan out FB and DR edge. That's solving 6 pieces. Most can plan out FB, which is 5. In Petrus, you solve 4 and then 3, so 7. What if instead of extending 2x2x2 to a 2x2x3 in Petrus, your first step solved 2x2x2+1 cross edge? Theoretically you could plan it in inspection, as it's 5 pieces being solved (hence Roux FB) And from there you can finish the solve in a ton of different ways while improving the ergonomics of Petrus. Here's a solve of it. Any thoughts? (Used OLLCP because it was nice)

There are a handful of people who can plan the entire 2x2x3, but what I think some Petrus solvers do after planning the 2x2x2 is to trace the corners and predict where useful pairs will land up (possibly inserting moves or adapting the 2x2x2 solution to make it easier). This leaves just one edge to find and insert to finish the 2x2x3.

 

[Shiv3r]

Hey guys. Here are some things I am working on right now. Most of these are for big cubes.
1.(4x4)I am working on a Spreadsheet of CMLL while solving OLL parity at the same time for Meyer. all of the algs so far(and most of the algs overall probably) Are in the form *setup* Lucasparity/Pure Parity *postmoves*, where either the setup or the postmoves are optional. the solves where I use the OP+CMLL I have learned so far are usually my faster parity solves. description and rant about it here
2.(4x4) just a random headcannon, but @Teoidus among others use CF3L, doing Hoyacross then insert corners and F3L is fun.
3.(5x5) This is something I am totally using right now. I call it Yau5 Shiv3r's variation. Basically after doing L4C instead of solving the 4th cross edge, then after freestyle first 4 edges and then some weird insert 2 edges and then do AvG, like you do in Yau5, you go straight into edge pairing after the L4C with only 3 cross dedges. Place the first or second cross tredge you solve in the cross slot using these two algs(missing spot at DR):
tredge in FR: R U R U' R2
tredge in RB: R' U R' U' R2
then when you get to L4E, you can do the same L4E in the M slice as you would Freeslice, by bringing up edges with U2's(in this variant of YAu5 you gotta bring it up to the missing cross tredge spot first). Then you solve the 4th cross tredge(finding it when finishing up L4E is nice), then do CFOP. here's a video explaining it
4.(Skewb) This "corners first" method is a convenience method like TCLL for 2x2, where you use it only when useful. basically you solve 4 corners relative to each other, then do Sarah's intermediate(make sure the center isn't in the D layer) then do L5C(I 2look it, but probably a good Idea to learn the L5C)

EDIT: please tell me what you think of all of these things.

 

[Neuro]

Weird idea: HF for OH? It actually doesn't seem too bad. Roux solvers won't have a problem making the hexagon, and F2L is pseudo 2 gen. The only thing from there is how to finish the solve. Maybe Hexagon, F2L -1, insert last cross edge while orienting all other edges (pseudo 2 gen or MU), TSLE and TTLL?

EDIT: Here's a solve I did. I am a right handed OH solver. Probably could've been more efficient but it should give you the idea. Algs for EO+DF need to be made, on this solve I just inserted and used an OLLCP. Another thing you can do is insert LS+DF and finish with OLL/PLL. I'd probably do the OG variant because it won't mess with 2 gen solves, is theoretically better to fingertrick, and should have zero rotations.

 

[Thermex]

I feel like this would be a SUPER good replacement to ZBLL for people who dont want to learn 500 algs, it uses only a little over 100 algs and you can 3-look the last slot+last layer in about 25-26 moves with way better regocnition than ZBLL.

1. Insert your last slot corner in a random orientation and orient the edges (~6 moves with about 16 algs)
2. Twist the corner in it's slot so that it's solved and orient the rest of the corners (~8 moves with about 16 algs)
3. ZZ-HW last layer (~11-12 moves with 72 algorithms)

I really think this has potential, but hopefully someone out there has a document of ZZ-HW algs, that would help a lot. This could also have lots of potential for HF OH:

1. Hexagon+3 e-layer edges (every e-layer edge-FR)
2. Orient all six remaining edges while putting the LS corner in its slot in a random orientation (probably around 20-30 algs)
3. Permute the last 5 corners (TCOLL, only 86 additional algs)
4. L6EP (completely 2-gen, 86 algorithms)

I feel like this method has serious potential, anyone want to work with me on it/give some feedback on what I just said?

 

[Shiv3r]

I feel like this would be a SUPER good replacement to ZBLL for people who dont want to learn 500 algs, it uses only a little over 100 algs and you can 3-look the last slot+last layer in about 25-26 moves with way better regocnition than ZBLL.

1. Insert your last slot corner in a random orientation and orient the edges (~7 moves with about 24 algs)
2. Twist the corner in it's slot so that it's solved and orient the rest of the edges (~8 moves with about 16 algs)
3. ZZ-HW last layer (~11-12 moves with 72 algorithms)

I really think this has potential, but hopefully someone out there has a document of ZZ-HW algs, that would help a lot. This could also have lots of potential for HF OH:

1. Hexagon+3 e-layer edges (every e-layer edge-FR)
2. Orient all six remaining edges while putting the LS corner in its slot in a random orientation (probably around 40-50 algs)
3. Permute the last 5 corners (TCOLL, only 42 algs)
4. L6EP (completely 2-gen, 86 algorithms)

I feel like this method has serious potential, anyone want to work with me on it/give some feedback on what I just said?

ZZ-HW was the precursor to ZZ-CT, and is much more inefficient alg-wise iirc.

 

[Neuro]

I feel like this would be a SUPER good replacement to ZBLL for people who dont want to learn 500 algs, it uses only a little over 100 algs and you can 3-look the last slot+last layer in about 25-26 moves with way better regocnition than ZBLL.

1. Insert your last slot corner in a random orientation and orient the edges (~7 moves with about 24 algs)
2. Twist the corner in it's slot so that it's solved and orient the rest of the edges (~8 moves with about 16 algs)
3. ZZ-HW last layer (~11-12 moves with 72 algorithms)

I really think this has potential, but hopefully someone out there has a document of ZZ-HW algs, that would help a lot. This could also have lots of potential for HF OH:

1. Hexagon+3 e-layer edges (every e-layer edge-FR)
2. Orient all six remaining edges while putting the LS corner in its slot in a random orientation (probably around 40-50 algs)
3. Permute the last 5 corners (TCOLL, only 42 algs)
4. L6EP (completely 2-gen, 86 algorithms)

I feel like this method has serious potential, anyone want to work with me on it/give some feedback on what I just said?

1st proposal:I assume this is used with HF? And in step 2 the orientation shouldn't be necessary b/c it was done in step1? ZZ-HW as far as I know has no official documentation, and I think that it had a few really bad cases along with poor recog so you may not wish to pursue it.
2nd proposal:Orienting all edges doesn't have that many cases, but I'd recommend a specified place for your corner to go. Assuming you keep the corner in a constant spot, you'd only have 32 cases. Also, TCOLL would have far more than 42 algs assuming I am reading it correctly. COLL has 42, TCLL+/- have 43 each. L6EP won't have 86 algs though, it'd only have like 18. You'd only have the 7 EPLL's and 11 L5EP's.

Oh and for DF+EO, if you keep the edge in a specific slot (preferrably in FL/R), it takes ~3 moves to set up and case recog improves drastically. Alternatively you can hold it in DF/DB, but it may increase movecount. Not sure though, still need to gen algs.

 

[Thermex]

1st proposal:I assume this is used with HF? And in step 2 the orientation shouldn't be necessary b/c it was done in step1? ZZ-HW as far as I know has no official documentation, and I think that it had a few really bad cases along with poor recog so you may not wish to pursue it.
2nd proposal:Orienting all edges doesn't have that many cases, but I'd recommend a specified place for your corner to go. Assuming you keep the corner in a constant spot, you'd only have 32 cases. Also, TCOLL would have far more than 42 algs assuming I am reading it correctly. COLL has 42, TCLL+/- have 43 each. L6EP won't have 86 algs though, it'd only have like 18. You'd only have the 7 EPLL's and 11 L5EP's.

Oh and for DF+EO, if you keep the edge in a specific slot (preferrably in FL/R), it takes ~3 moves to set up and case recog improves drastically. Alternatively you can hold it in DF/DB, but it may increase movecount. Not sure though, still need to gen algs.

I screwed up a lot of stuff there, step 2 I meant to say orient the rest of the corners, also TCOLL is 125 if you include COLL with it. I was also thinking all of this would be done with the FR slot. Wait but why would L6EP only be 18 algs now? That doesn't make sense. I also think ZZHW should be explored more for this to work.

 

[Neuro]

Oh wait I'm sorry! I forgot that DF isn't solved! So yes, it's still the normal L6EP alg count. As for ZZHW, you'd need to make the algs for that as I don't think any of the algs were released. You should probably start with the conjugated PLL's though.

 

[Thermex]

Oh wait I'm sorry! I forgot that DF isn't solved! So yes, it's still the normal L6EP alg count. As for ZZHW, you'd need to make the algs for that as I don't think any of the algs were released. You should probably start with the conjugated PLL's though.

I know I'm gonna sound really stupid but I kinda don't get what a conjugated PLL is ;( also Im gonna need some help generating HW cases, I'm already pretty busy with the TEG algs.

 

[Neuro]

Ok so basically it'll look something like this. It's an EPLL, but you start off by doing a "setup" move (let's call this move "A"). From there, you perform an alg (let's call this "B") and finally, you reverse your setup move and you should be able to solve the rest very easily. Basically if we look at it from the letter standpoint, it looks something like this: A, B, A'; where A' represents a reversal of the initial A move/setup. You can do multiple moves to setup, but starting with those that only have 1 move will be a lot easier overall.

It gets a little more complicated than this, but this should at least give you a basic idea of how it works. Ryan Heise has a lot of good information on commutators/conjugates on his website.

 

[Thermex]

Oh ok thats really simple, I do that all the time, I just didn't know what it's called. I'll start making those.

 

[Neuro]

Here's a solve proving the true potential of WaterRoux. The solve was fairly average and still got under 40 moves while over half of the solve was pseudo 2 gen and over 75% of the solve was simply algorithmic and/or intuitive.
U R' D2 L B L' y//FB (6/6)
R2 U y' U R' U2 R' F R F' R y//Corners (10/16)
M U2 M2 U R2 U M2 U'//ERL (8/24)
r U' M U R2 M U M U2 M U' M2 U2 R'//L7E (14/38)

 

[Zanaso]

Here's a solve proving the true potential of WaterRoux. The solve was fairly average and still got under 40 moves while over half of the solve was pseudo 2 gen and over 75% of the solve was simply algorithmic and/or intuitive.
U R' D2 L B L' y//FB (6/6)
R2 U y' U R' U2 R' F R F' R y//Corners (10/16)
M U2 M2 U R2 U M2 U'//ERL (8/24)
r U' M U R2 M U M U2 M U' M2 U2 R'//L7E (14/38)

wtf is the other 25%? surely 100% of all solves are intuitive and algorithmic?

 

[Thermex]

wtf is the other 25%? surely 100% of all solves are intuitive and algorithmic?

Lol I was thinking the EXACT same thing when I read that

 

[Gold Cuber]

i did once create a corners first method, but it takes about 10 minutes to solve. if you want to see how to do it, (2 algs) reply telling me

 

[Neuro]

Alright alright I misspoke over 75% was algorithmic

 

[Thermex]

@Neuro that was a pretty funny mistake, it's fine though for WaterRoux L9E do you think there could be some way of solving the last 9 edges similar to Crafto's method? Also I still like the idea for HF I came up with and think it's the best way to end an HF solve invented so far, but I was thinking for CFOP maybe another ZBLL replacement could be:

1. Cross+1 corner (adds one or two additional moves to the cross and forms a fish shape on the d-layer)
2. Last 3 pairs
3. Holding the unsolved slot in front, do a step where you orient all pieces including the edge in the slot with the corner you solved with the cross (114 algorithms since FR edge can be flipped one of two ways and there are 57 OLL cases)
4. ZZ-HW last layer (again, algorithms need to be generated for this and it would probably be ~12 moves per alg)

 

[Hazel]

New way to solve AUFs:
Once you solve PLL, if you have any AUF, just do this until the AUF is solved:
Y perm, E perm, Ua perm
Much better than regular AUF methods