The New Method / Substep / Concept Idea Thread

[Shiv3r]

Hey, just a bump. I was wondering if anyone used Vandenbergh-harris? Because I know COLL(ish) I end up using it when I try to do CFOP example solves. It seems easier than vanilla CFOP to me, but that may just be because I still don't know full OLL, only full PLL.

 

[obelisk477]

Hey, just a bump. I was wondering if anyone used Vandenbergh-harris? Because I know COLL(ish) I end up using it when I try to do CFOP example solves. It seems easier than vanilla CFOP to me, but that may just be because I still don't know full OLL, only full PLL.

pretty sure @Bindedsa uses many of the VHLS cases since he uses full ZBLL, and almost certainly full VHLS for OH. Not useful IMO unless you know and can use full ZBLL

 

[sqAree]

pretty sure @Bindedsa uses many of the VHLS cases since he uses full ZBLL, and almost certainly full VHLS for OH. Not useful IMO unless you know and can use full ZBLL

Someone that knows ZBLL has a high chance to use VHLS, granted, but bindedsa uses so many other LL cases that it might not be worth it for him.

 

[Shiv3r]

The original VH was COLL/EPLL, and it was to ZB as s ZZ-a is to ZZ-b. Its pretty nice just doing VHLS->COLL->EPLL, and the total alg count is still less than full Vanilla CFOP, around 20 algs less or so.

 

[xyzzy]

Which one is better for speedsolving, the one you were talking about, or the Z4 EOpairing? ZZ-4 is an attempt to optimize Z4 kinda like Yau optimized redux/roux, so of course I kept the EOpairing concepts from it.

I haven't come up with a good way to do EO alone, so I'm guessing EOpairing has more potential. I tried EOpairing-style orientation (EOpairing but without actually pairing edges) to orient two/four edges at once, which is much less reliant on pausing to inspect and also more fingertrickable, but this feels like defeating the purpose. (Imagine if you did ZZ by orienting edges two at a time using F U F' instead of four at a time using F!)

I'm a bit slower with Yau and Hoya than plain reduction, so I can't really give useful comments on whether ZZ-4 is a better or worse version of Z4. (That aside, the naming is also pretty confusing, to be honest.)

 

[Sue Doenim]

Would it be possible to have a two look, or two slot VLS? Sorry if there is already a post on this subject; I'm really not willing to read all 2000 something previous posts. The steps would be as follows:

1) F2L-2
2) VHLS- at this point, I understand that there could be a misoriented edge in the last F2L slot. This would only, I believe, add 6 algorithms. Also, I think that this would be better done from any slot, not a fixed slot such as always from FL, and that some (modified) ZBLS cases could be learned to increase efficiency.
3) Winter variation- unlike at the previous stage, I believe that there should be a fixed slot (FR) for this, owing to the greater amount of complexity. Summer variation, as well as possibly parts of ZZ-c, similarly to ZBLS in the last step, may be a valuable asset.

Let me know what you guys think!

 

[obelisk477]

Would it be possible to have a two look, or two slot VLS? Sorry if there is already a post on this subject; I'm really not willing to read all 2000 something previous posts. The steps would be as follows:

1) F2L-2
2) VHLS- at this point, I understand that there could be a misoriented edge in the last F2L slot. This would only, I believe, add 6 algorithms. Also, I think that this would be better done from any slot, not a fixed slot such as always from FL, and that some (modified) ZBLS cases could be learned to increase efficiency.
3) Winter variation- unlike at the previous stage, I believe that there should be a fixed slot (FR) for this, owing to the greater amount of complexity. Summer variation, as well as possibly parts of ZZ-c, similarly to ZBLS in the last step, may be a valuable asset.

Let me know what you guys think!

For step two, you're not considering the cases where the F2L-3 corner is caught in the F2L-4 slot. That adds alot of algorithms

 

[AlphaSheep]

For step two, you're not considering the cases where the F2L-3 corner is caught in the F2L-4 slot. That adds alot of algorithms

You use VHLS for F2L-3, so the first step would be to bring that piece into LL and setup the VHLS case anyway. It wouldn't need extra algs. The only extra algs needed would be to flip the edge in the last slot (which i think would actually double the alg count).

 

[Smiles]

i've thought of similar ideas before, and here are my thoughts after many years.

in my experience, about 1/3 of WV is good, and the rest is either hard to recognize or its a slow alg.
SV has like a few good cases and the rest are both hard to recognize and bad algs.
and imo VHLS algs are mostly pretty slow for the goal that VHLS accomplishes. like they're not slow algs themselves, but they're pretty slow for just doing eo.

MGLS a 2-look substep that does F2L-1 + OLL, maybe you'd be interested in that.

anyway, if you want to do OLL before OLL then my opinion is that the best practical option is OLS but only the decent cases. of course you can learn them all, but that's hard. for bad cases, you can either do regular OLL or memorize what OLL you'll get and insert + OLL will be almost as fast as knowing the OLS alg.

plus, if you get a terrible VHLS case + terrible WV/SV case, then i assure you that it would have been faster to just do OLL.

 

[mDiPalma]

I think treating VHLS as an algorithm set is the wrong mentality. It's just orienting the edges while inserting a pair. Just do it intuitively. 0 algs - no matter when you do it.

Once you understand how edge control works, you can do it naturally and fluently throughout the solve, not just during the 3rd/4th pair.

However, one interesting idea would be to form the last F2L pair (R U R' or R U' R' inserts) while orienting the LL edges. That would leave you open to LS variants like you suggested.

Or, if all the edges are already oriented, form the last F2L pair (R U R' or R U' R' inserts) while setting up an easy WV/SV case. What characteristic of WV/SV cases makes them easy? 1 misoriented corner? oriented corners adjacent to the pair?

 

[obelisk477]

What characteristic of WV/SV cases makes them easy? 1 misoriented corner? oriented corners adjacent to the pair?

1 or 2 misoriented corners, which are not the UBL corner. Those are the only cases I use anyhow

 

[Sue Doenim]

Thanks for all the answers; however I wasn't really looking into learning it for myself. I was thinking it could be used as possibly an alternative to or gateway into VLS. I think, personally, that LS methods or substeps are the future of cubing. The concept of VHLS applied to when there is one more slot left could however be useful. It could open up users of CFOP and similar methods to the possibility of methods like ZZ-CT, phazing, or as previously mentioned, winter variation. I am in the process of learning ZZ-CT, and I think that many CFOP users would benefit greatly from this style of LL or technically, LL skip, without having to learn EOLine or ZZF2L.

 

[shadowslice e]

Thanks for all the answers; however I wasn't really looking into learning it for myself. I was thinking it could be used as possibly an alternative to or gateway into VLS. I think, personally, that LS methods or substeps are the future of cubing. The concept of VHLS applied to when there is one more slot left could however be useful. It could open up users of CFOP and similar methods to the possibility of methods like ZZ-CT, phazing, or as previously mentioned, winter variation. I am in the process of learning ZZ-CT, and I think that many CFOP users would benefit greatly from this style of LL or technically, LL skip, without having to learn EOLine or ZZF2L.

I sometimes use it so that i can get a mostly 2 gen LS or last 2 slots when i bother using CFOP though i would say that it is still more useful to just use straight multislotting most of the time (hence why feliks-probably the best multislotter- is faster than Mats- one of the few people who uses VLS most of the time- 90% of the time).

 

[Shiv3r]

While we are talking about LS methods, would it be possible to do an L2S/LS method after roux F2B? assume the centers are placed correctly. Inserting the last 2 D edges could be made into a ls-influencing method.

anyway, I was working on a different strategy for columns-first, after building the columns, and I think I came up with a good Idea(completely intuitive for me, because Im rouxer) that is fairly efficient and doesn't require S moves. It is fairly rotationlessness, the only rotations ae during the column-building step. here is an example solve. anyone still interested?
The last steps could be optimized with algorithms, which I Have yet to generate.(one example alg set would be solving UL/UR and DL/DR simultaneously, alg count could be shortened by placing them all in the U layer i think.)
EDIT: in the example solve, the DL/DR edges could also be solved by placing them at UF/UB and then doing D/D' M2 D/D', which would be a better solve ergonomically.

 

[AlphaSheep]

anyway, I was working on a different strategy for columns-first, after building the columns, and I think I came up with a good Idea(completely intuitive for me, because Im rouxer) that is fairly efficient and doesn't require S moves. It is fairly rotationlessness, the only rotations ae during the column-building step. here is an example solve. anyone still interested?
The last steps could be optimized with algorithms, which I Have yet to generate.(one example alg set would be solving UL/UR and DL/DR simultaneously, alg count could be shortened by placing them all in the U layer i think.)
EDIT: in the example solve, the DL/DR edges could also be solved by placing them at UF/UB and then doing D/D' M2 D/D', which would be a better solve ergonomically.

I could follow the pairs, CMLL and EO, but after that I was completely lost. It seems you do a lot of moves and accomplish very little. Solving DR+DL straight away is very easy in this case, then you can do Roux style LSE. My ending to your example solve would be
R2 S' R2 S// DR+DL
M U2 M' U' M2 U' // UR+UL
M U2 M' U2 M2 // L3E
It's 8 STM (11 HTM) shorter. Yes, the R2 S isn't great to execute, but it's no worse than yours. You could do it as z' y U2 M' U2 M y' z, but the R2 S can be done with one regrip whereas rotating needs two.

 

[Shiv3r]

I could follow the pairs, CMLL and EO, but after that I was completely lost. It seems you do a lot of moves and accomplish very little. Solving DR+DL straight away is very easy in this case, then you can do Roux style LSE. My ending to your example solve would be
R2 S' R2 S// DR+DL
M U2 M' U' M2 U' // UR+UL
M U2 M' U2 M2 // L3E
It's 8 STM (11 HTM) shorter. Yes, the R2 S isn't great to execute, but it's no worse than yours. You could do it as z' y U2 M' U2 M y' z, but the R2 S can be done with one regrip whereas rotating needs two.

so here is what my general strategy is:
-EO on both sides, <U,M,D>-gen
-get UL/UR and DL/DR on U layer
-execute an EPLL to solve UL/UR and get DL/DR opposite each other, then do D/D' M2 D'/D to solve UL/UR and DL/DR simultaneously(note to self: make an algset to do this in one step later)
-roux step 4c.

The problem with solving DR/DL and then LSE is that it ie pretty much a shittier roux. This method, however, the last steps are Completely <UMD>-gen so its much easier. This method, if fully color neutral, means that most of the time you already have one pair completed, and pair building is much more efficient there. My method tries to solve UL/UR and DL/DR Simultaneously, so you can just do a simple 4c <M,U> algorithm afterwards. The basis of this Idea was to prevent columns-first from being a redux down to Roux like it is at the moment.

 

[AlphaSheep]

The problem with solving DR/DL and then LSE is that it ie pretty much a shittier roux

Yes, I agree.

The basis of this Idea was to prevent columns-first from being a redux down to Roux like it is at the moment.

Many early methods were corners first, but all the slice moves and rotations were awkward. It is pretty easy to see that solving a few edges while solving the corners is far more efficient. More advanced corner first methods did that. PCMS is one where you solve the 4 E slice edges while solving the bottom corners. Roux is another which solves 6 edges while solving the bottom corners. Roux seems to be what all corner first derivatives tend towards as you try optimise them. I think to do something just to be different to Roux is a mistake. You should do it because it's more efficient or easier. In this case, I think your method is neither.

 

[Shiv3r]

I have two things for roux solvers: one, when you are trting to teach friends, use a sandwich-style corners first method like this because it transitions so nicely to roux(you can switch steps over slowly, and it isnt too bad)
and second, for trying to learn blockbuilding, I think that this is absolutely awesome and totally useful.

 

[Attila]

R2 S' R2 S// DR+DL
M U2 M' U' M2 U' // UR+UL
M U2 M' U2 M2 // L3E
It's 8 STM (11 HTM) shorter.

It’s even shorter:

F2 L2 U S2 U’

B2 R’ r’ F2 R r

 

[Shiv3r]

anyway, what are the general thoughts on ZZ-4? It seems that with a little toying with it could be a nice method. Even if people don't use the method, I still think that Ideas such as fixing parity while pairing last 2 edges and EPLL+parity algs are very useful and should be implemented in other solving methods soon. Anyone have suggestions to improve ZZ-4?