The New Method / Substep / Concept Idea Thread

[Shiv3r]

I can do roux but my ZBLL skills are limited

use the 21LL at this page:
lar5.com/cube/270

 

[wir3sandfir3s]

I've got a couple ideas. They're alright (ish? Idk).

We all know that one forgotten page on the wiki in the experimental methods section, COALL... Basically ZB with less Algs. And with ZZ-CT being a thing now, I have an idea. I think. My ideas will hopefully get better with summer and all... Anyway. Solve with CFOP till last slot. Easy? Good. Now, this is weird. Insert the last edge with an oriented corner while applying a 4 flip if needed... If all edges are already oriented, or 2 are oriented then don't do it. Should be 54 Algs. Now do TTLL or some weird type of TTLL for 2 unoriented to finish the solve.
Now it gets weird... You can add some weird form of phasing to this orienting LS, bringing it "only" up to 162 Algs, while drastically lowering Algs for LL... This means that the 2 unoriented edges always end up opposite... You can also add CP or whatever (on its own) and it will still be 162 Algs.
Now something very weird but pretty awesome, like future-of-cubing-altering awesome. What if you could execute 2 Algs at the same time? I have had this idea for a while but I think this is far as I am going to be able to take it. So, CO can be solved 2 gen, and EO can be solved with a combination of M and U moves. In theory, there should only be 3 Algs needed for EO in LL and 27 for WV or 90 something for TSLE (whatever you choose). If, while genning new Algs for this, Algs were optimized for compatibility with each other, and a little practice, RU and MU moves can blend with each other easily. Recog shouldn't be too bad, I think.
Now, this also means that you can have 10 Algs OLL which is interesting...

And my last idea is ZBLL but with a 4 flip. Basically ZBLL but with no edges oriented. Should be same number of Algs and stats, but yeah. That's it. Good if you already know ZBLL and want a higher chance of LL skip in CFOP.

 

[sqAree]

Now something very weird but pretty awesome, like future-of-cubing-altering awesome. What if you could execute 2 Algs at the same time? I have had this idea for a while but I think this is far as I am going to be able to take it. So, CO can be solved 2 gen, and EO can be solved with a combination of M and U moves. In theory, there should only be 3 Algs needed for EO in LL and 27 for WV or 90 something for TSLE (whatever you choose). If, while genning new Algs for this, Algs were optimized for compatibility with each other, and a little practice, RU and MU moves can blend with each other easily. Recog shouldn't be too bad, I think.
Now, this also means that you can have 10 Algs OLL which is interesting...

Oh, can you give an example for that?

 

[Shiv3r]

I've got a couple ideas. They're alright (ish? Idk).

We all know that one forgotten page on the wiki in the experimental methods section, COALL... Basically ZB with less Algs. And with ZZ-CT being a thing now, I have an idea. I think. My ideas will hopefully get better with summer and all... Anyway. Solve with CFOP till last slot. Easy? Good. Now, this is weird. Insert the last edge with an oriented corner while applying a 4 flip if needed... If all edges are already oriented, or 2 are oriented then don't do it. Should be 54 Algs. Now do TTLL or some weird type of TTLL for 2 unoriented to finish the solve.
Now it gets weird... You can add some weird form of phasing to this orienting LS, bringing it "only" up to 162 Algs, while drastically lowering Algs for LL... This means that the 2 unoriented edges always end up opposite... You can also add CP or whatever (on its own) and it will still be 162 Algs.
Now something very weird but pretty awesome, like future-of-cubing-altering awesome. What if you could execute 2 Algs at the same time? I have had this idea for a while but I think this is far as I am going to be able to take it. So, CO can be solved 2 gen, and EO can be solved with a combination of M and U moves. In theory, there should only be 3 Algs needed for EO in LL and 27 for WV or 90 something for TSLE (whatever you choose). If, while genning new Algs for this, Algs were optimized for compatibility with each other, and a little practice, RU and MU moves can blend with each other easily. Recog shouldn't be too bad, I think.
Now, this also means that you can have 10 Algs OLL which is interesting...

And my last idea is ZBLL but with a 4 flip. Basically ZBLL but with no edges oriented. Should be same number of Algs and stats, but yeah. That's it. Good if you already know ZBLL and want a higher chance of LL skip in CFOP.

Lookahead for 4-flip ZBLL would actually be really good, because you could basically look at the corner orientation and the top layer and you could recognize the alg.
recognition would actually be really good, I think.

 

[4Chan]

4-flip ZBLL already exists, it's called anti-ZBLL, and it SUCKS because whenever you're flipping that many pieces, the algs are bad.

6 years ago, a guy learned a huge fraction of it, and honestly, it's not worth it for speed.
But yeah, recog is sick.

 

[wir3sandfir3s]

Oh, can you give an example for that?

Of what (lol)

 

[Shiv3r]

4-flip ZBLL already exists, it's called anti-ZBLL, and it SUCKS because whenever you're flipping that many pieces, the algs are bad.

6 years ago, a guy learned a huge fraction of it, and honestly, it's not worth it for speed.
But yeah, recog is sick.

you know, you could do wonething like 21LL for it... learn 4 lags that 4-flip while permuting edges, then solve L4C(recon at same time)

 

[Daniel Lin]

what's the average movecount for anti-zbll?
not that I'm actually going to learn it

 

[Cale S]

what's the average movecount for anti-zbll?
not that I'm actually going to learn it

looks like about 13.11
http://birdflu.lar5.com/?cop=&oll=&co=&cp=&eo=0&ep=&clicked=#eo&list=positions
but it seems like some cases are counted multiple times so not entirely accurate

 

[sqAree]

Of what (lol)

I meant the thing about executing two algs at the same time as I'm not quite sure what you mean.

 

[Daniel Lin]

Now do TTLL or some weird type of TTLL for 2 unoriented to finish the solve.

wait I'm confused. The last layer corners are not oriented yet, so how can you use TTLL? You just inserted the last slot edge with an oriented corner, so what about the LL corners?

You can also add CP or whatever (on its own) and it will still be 162 Algs

I don't get it.Can you give an example?

 

[wir3sandfir3s]

wait I'm confused. The last layer corners are not oriented yet, so how can you use TTLL? You just inserted the last slot edge with an oriented corner, so what about the LL corners?

I don't get it.Can you give an example?

The LL corners are oriented because when you insert the last F2L edge with an printed corner, you are doing a less restrictive WV (I think it's a subset of TSLE). Did I miss that part? Whatever. Also, you are doing a 4 flip if needed.
For adding CP or phasing to "TSLE", it's literally just TSLE while doing CP or phasing two edges, basically more Algs to use to your advantage.

 

[4Chan]

I'm all for innovation, but TSLE + CP is thousands of algs with ridiculous recognition.
With five corners moving around, you then have to account for 5 cycles, and it gets absolutely absurd with alg count.

Even though TTLL has 5 cycles, everything's oriented so rotational symmetry makes it super small.
If things are misoriented, you can't abuse rotational symmetry.

One of the benefits of TSLE is that it completely ignores permutation of every piece other than a single edge.
By not caring about anything else, that's how the algs are nice.

 

[Shiv3r]

I'm all for innovation, but TSLE + CP is thousands of algs with ridiculous recognition.
With five corners moving around, you then have to account for 5 cycles, and it gets absolutely absurd with alg count.

Even though TTLL has 5 cycles, everything's oriented so rotational symmetry makes it super small.
If things are misoriented, you can't abuse rotational symmetry.

One of the benefits of TSLE is that it completely ignores permutation of every piece other than a single edge.
By not caring about anything else, that's how the algs are nice.

for CP, (I use this for OH BTW), I detect CMLL case and based on if it is a x1, x2, x3, I do the appropriate AUF(I have them all written down for all 6 possible CMLL numbers) and then do a niklas(or 2 for the diagonal case.)

 

[Shiv3r]

I'm all for innovation, but TSLE + CP is thousands of algs with ridiculous recognition.
With five corners moving around, you then have to account for 5 cycles, and it gets absolutely absurd with alg count.

Even though TTLL has 5 cycles, everything's oriented so rotational symmetry makes it super small.
If things are misoriented, you can't abuse rotational symmetry.

One of the benefits of TSLE is that it completely ignores permutation of every piece other than a single edge.
By not caring about anything else, that's how the algs are nice.

TTLL can be reduced to 24 algs(not including 9 PLLs, so 33 in total) with phasing(as in Navi method)

also, I was thinking instead of orienting all/permuting all, would it make sense to make an alg set to solve, say, all the edges and the most convenient corner(EO solved already, as in ZZ/ZB), then you can use an L3C finish? because its not easy to calculate, i think the alg count would be pretty low, as the corner oprions(I think) would be an adjacent cycle/swap, a cycle/swap that twists the corner one of 2 ways. combine that with EPLL and you have a 12 alg 1st look, and a 24(?) alg L3C.

 

[4Chan]

From what I understand, it'd be:

1. Insert edge + edge permutation + 1 correct corner
2. Solve everything

Yep, that'd work with low move count and alg count, but with tough recognition for the first step.
Since you'd have multiple corners to consider.

I'd say give it a try!
.

 

[Shiv3r]

From what I understand, it'd be:

1. Insert edge + edge permutation + 1 correct corner
2. Solve everything

Yep, that'd work with low move count and alg count, but with tough recognition for the first step.
Since you'd have multiple corners to consider.

I'd say give it a try!
.

I would, but what would the real alg count be?

also , if combined with ZZ/petrus and LPELL, you could make it a 3-alg first look because edges are already permuted.

 

[Shiv3r]

As a tradeoff for more algs but easier recog, you could AUF for the EPLL's and then figure out where the corner at FR has to go.

 

[Shiv3r]

I had an interesting idea for a ZZ+ Roux mix.
this is to make the most fluid method ever, because ZZF2L is really fluid, but the Last layer is just as unfluid as CFOP.
I like roux's last steps, CMLL and LSE, but the blockbuilding is a little less fluid(lookahead is not as good as ZZF2L)

so what I did was this:
step 1:EOUnline- this is an EOline but with either A)2 U layer edges instead of line or B) place the UR/UR edges in the line instead, to make a really fast last steps
step 2:ZZF2L- because the line is a psuedoline this looks a lot more like F2B+EO, but treat it like ZZF2L.
step 3:COLL-Either influence into a shorter algset or use full COLL, either way, not too bad alg-wise.
step 4:finish with a kind of LSE-not really an LSE, but based on your EOUnline choices, it could be:
A)4b and 4c
B)do an M2 to solve UL/UR and then do 4c: because of how it starts, you can use BU recog right at this point, right after COLL.

maybe not good movecount wise, But I think that if we are looking for the most fluid method we combine the two most fluid things anyone has come up with yet: ZZF2L and LSE.

 

[shadowslice e]

I had an interesting idea for a ZZ+ Roux mix.
this is to make the most fluid method ever, because ZZF2L is really fluid, but the Last layer is just as unfluid as CFOP.
I like roux's last steps, CMLL and LSE, but the blockbuilding is a little less fluid(lookahead is not as good as ZZF2L)

so what I did was this:
step 1:EOUnline- this is an EOline but with either A)2 U layer edges instead of line or B) place the UR/UR edges in the line instead, to make a really fast last steps
step 2:ZZF2L- because the line is a psuedoline this looks a lot more like F2B+EO, but treat it like ZZF2L.
step 3:COLL-Either influence into a shorter algset or use full COLL, either way, not too bad alg-wise.
step 4:finish with a kind of LSE-not really an LSE, but based on your EOUnline choices, it could be:
A)4b and 4c
B)do an M2 to solve UL/UR and then do 4c: because of how it starts, you can use BU recog right at this point, right after COLL.

maybe not good movecount wise, But I think that if we are looking for the most fluid method we combine the two most fluid things anyone has come up with yet: ZZF2L and LSE.

This won't be fluid as you won't easily be able to easily see the FB. In addition, you may well have edges in the awkward places in FD/BD. As well as this, the movecount will be higher than just Roux F2B which I would consider to be better than ZZF2L as it has better lookahead becuas eyou don't have to blockbuild after a slightly jarring transition after EO.