I don’t know a lot about roux, but I after a method idea I did some research about roux and think I found a pretty good varaint.
1. Solve 2 1x2x3's like normal
2. Solve the top 2 1x1x3's
3. Permute & Oreint the last 4 edges
For step 2:
I would just permute & orient the UL and UR edges intuitively, and then do COLL. COLL algs could probably be shortened by not caring about UF, UB, DF, and DB edges.
The problem with CMLL is it doesn't care about the F, U, and B centers, along with the UL and UR edges.
Something I would find useful is if the DB edge could be solved during step 2. This would eliminated cube rotations and make recognition way easier.
What do you think. Is it viable, it does it still need work?
I don’t know a lot about roux, but I after a method idea I did some research about roux and think I found a pretty good varaint.
1. Solve 2 1x2x3's like normal
2. Solve the top 2 1x1x3's
3. Permute & Oreint the last 4 edges
For step 2:
I would just permute & orient the UL and UR edges intuitively, and then do COLL. COLL algs could probably be shortened by not caring about UF, UB, DF, and DB edges.
The problem with CMLL is it doesn't care about the F, U, and B centers, along with the UL and UR edges.
Something I would find useful is if the DB edge could be solved during step 2. This would eliminated cube rotations and make recognition way easier.
What do you think. Is it viable, it does it still need work?
I don’t know a lot about roux, but I after a method idea I did some research about roux and think I found a pretty good varaint.
1. Solve 2 1x2x3's like normal
2. Solve the top 2 1x1x3's
3. Permute & Oreint the last 4 edges
For step 2:
I would just permute & orient the UL and UR edges intuitively, and then do COLL. COLL algs could probably be shortened by not caring about UF, UB, DF, and DB edges.
The problem with CMLL is it doesn't care about the F, U, and B centers, along with the UL and UR edges.
Something I would find useful is if the DB edge could be solved during step 2. This would eliminated cube rotations and make recognition way easier.
What do you think. Is it viable, it does it still need work?
I don’t know a lot about roux, but I after a method idea I did some research about roux and think I found a pretty good varaint.
1. Solve 2 1x2x3's like normal
2. Solve the top 2 1x1x3's
3. Permute & Oreint the last 4 edges
For step 2:
I would just permute & orient the UL and UR edges intuitively, and then do COLL. COLL algs could probably be shortened by not caring about UF, UB, DF, and DB edges.
The problem with CMLL is it doesn't care about the F, U, and B centers, along with the UL and UR edges.
Something I would find useful is if the DB edge could be solved during step 2. This would eliminated cube rotations and make recognition way easier.
What do you think. Is it viable, it does it still need work?
But that argument doesn’t hold up. Those algs would almost certainly be non- MU gen, which is what make LSE fast. And any MU gen algs for those cases would be very long because you would have to replace the edges you displaced, at which point it would be more convenient to solve with regular LSE.
Darn guys, you kinda just shot down his idea.
Try to be nice and give more explanation why other options are better, but good job ImmolatedMarmoset for telling why it’s bad, thank you.
So, the first two blocks of Roux and your method are the same.
Then, you decide to solve the UL UR edges?
This actually takes about 7 moves on its own, this is from some personal testing, which is pretty inefficient for just placing two pieces, which isn’t great, but if it leads to better steps it could be worth it.
Then you get to CLL, which, is not simply doing COLL but less restrictive, but in fact, more restrictive, so the algorithms would be much longer and worse. I have personally tested this with CE, and CMLL’s that can be solved in 9 moves require 13-14 moves now, because of those edges that you have to preserve.
Finally, we have L4E, which at first, I thought they would be oriented, so then you could solve it with just a simple cycle that would take 4-5 moves and is very quick, but if they actually not oriented which is what I understood, then it would take many more moves, but also, the algorithms would be non-MU. I have personally tested this with CE and verified that the shortest solutions are MFE or MBE moves and such, and about 10 moves of those weird moves, and if you search for only MU algorithms, they are much longer, in fact, as long as the entire LSE on its own.
So, since we know this is bad, let’s tell you an alternative to your method, and hopefully it’ll teach you of what’s bad and good.
After first two blocks, if you want to solve UL UR, you should stock them in D layer, because 1) this always takes less time then solving them directly, it in fact is close to cutting it in half, and 2) the CLL algorithms will be much nicer, because they will not be restricted by the UL UR edges on top.
But, even though this little slight variant is much better than what you proposed, it still leaves you with the same gross L4E as mentioned earlier. So, if we can manage to orient these edges in an earlier step, then the L4E will be super duper nice, infinitely better than it is now. We could do so whilst solving the UL UR edges to D layer, because this would require extra algorithms to be memorized as if we were to orient while doing CLL, and, orienting these four edges while stocking UL UR actually only takes a few extra moves, so it’s a good trade off. Perhaps just three very quick and easy MU moves added here to get rid of eight or so bad and slow moves in the end.
So, this new and improved method I have goes like this:
- First Block
- Second Block
- EO + Stock ULUR (or UFUB)
- CLL (similar to COLL)
- Good L4E (easy M-slice)
Here’s an example solve to prove its good:
Scramble: U B2 R' F2 R2 U2 L2 F2 D L2 F2 D L2 U2 B' F L B2 F' R' U'
(y2)
D R f R D' f // First Block (6/6)
r' y R U R' U S' R' F2 R y' // Second Block (9/15)
U' M U M' U2 M U' M' // EO + ULUR (8/23)
U' R U2 R2 U' R2 U' R2 U2 R // CLL (10/33)
U2 M2 U' M2 // Permute Final Edges (4/37)
Darn guys, you kinda just shot down his idea.
Try to be nice and give more explanation why other options are better, but good job ImmolatedMarmoset for telling why it’s bad, thank you.
So, the first two blocks of Roux and your method are the same.
Then, you decide to solve the UL UR edges?
This actually takes about 7 moves on its own, this is from some personal testing, which is pretty inefficient for just placing two pieces, which isn’t great, but if it leads to better steps it could be worth it.
Then you get to CLL, which, is not simply doing COLL but less restrictive, but in fact, more restrictive, so the algorithms would be much longer and worse. I have personally tested this with CE, and CMLL’s that can be solved in 9 moves require 13-14 moves now, because of those edges that you have to preserve.
Finally, we have L4E, which at first, I thought they would be oriented, so then you could solve it with just a simple cycle that would take 4-5 moves and is very quick, but if they actually not oriented which is what I understood, then it would take many more moves, but also, the algorithms would be non-MU. I have personally tested this with CE and verified that the shortest solutions are MFE or MBE moves and such, and about 10 moves of those weird moves, and if you search for only MU algorithms, they are much longer, in fact, as long as the entire LSE on its own.
So, since we know this is bad, let’s tell you an alternative to your method, and hopefully it’ll teach you of what’s bad and good.
After first two blocks, if you want to solve UL UR, you should stock them in D layer, because 1) this always takes less time then solving them directly, it in fact is close to cutting it in half, and 2) the CLL algorithms will be much nicer, because they will not be restricted by the UL UR edges on top.
But, even though this little slight variant is much better than what you proposed, it still leaves you with the same gross L4E as mentioned earlier. So, if we can manage to orient these edges in an earlier step, then the L4E will be super duper nice, infinitely better than it is now. We could do so whilst solving the UL UR edges to D layer, because this would require extra algorithms to be memorized as if we were to orient while doing CLL, and, orienting these four edges while stocking UL UR actually only takes a few extra moves, so it’s a good trade off. Perhaps just three very quick and easy MU moves added here to get rid of eight or so bad and slow moves in the end.
So, this new and improved method I have goes like this:
- FB
- SB
- EO+ULUR
- CLL
- Good L4E
Here’s an example solve to prove its good:
Scramble: R2 L2 B U' D L F B2 D R2 F' L2 B2 R2 D2 R2 U2 F D2 B L2
It’s really not bad at all. In fact, I like it quite a lot. My apologies to @brododragon for being more than a little belligerent. EO+ULUR feels a little weird, but I know I can get used to this in time. Maybe I’ll do an Ao50 with this today (I average 14.5 with standard Roux). I don’t know COLL, so I’ll have to improvise, but it should be fun!
CFCE CLL doesnt necessarily preserve EO, so you would need COLL. I think in general, if some change benefits the efficiency of an intuitive step, but hurts an algorithmic step, it doesnt matter as much because TPS will always be higher in algorithmic steps. So this might be better.
Darn guys, you kinda just shot down his idea.
Try to be nice and give more explanation why other options are better, but good job ImmolatedMarmoset for telling why it’s bad, thank you.
Let me just start off by saying what you created is almost definitely way better then what I was visualising. I want to see if what I had in mind had any viability anyway, though. The UL and UR edge aspect has viability, but I'm wondering if there's more.
What I had in mind was solving UR, UL, UFL, UFR, UBL, and UBR. The weird L4E, though, kinda ruins it.
The best thing I can think of that follows my vision is this:
1. Solve either the UL or UR edge + EO
2. insert the other edge whilst simultaneously doing COLL.
3. L4E
Note: this is nothing like what I had in mind, just a random idea.
I don't really know why I'm doing these any more; you already had a great idea I can probably not top, but here it goes.
1. F2B
2. Solve the DB and UB edges, along with the UBR and UBL corners. For this, you will probably need to do an x' rotation, but that shouldn't be too slow.
3. Solve the UF and DF edges, along with the UFR and UFL corners.
4. Solve the UL and UR edges
For step 2, before the x' rotation, you should probably take note of what the F face looks like (and the bottom of the DF edge.
Edit: 37 moves?! That's better than a lot of FMC solves.
So recently I had the idea that you could solve some kind of EOLine and EOEdge in one step for ZZ. It didn't really end in a great method, but I thought I'd still share it here.
It's called ZZ-Line and the steps are:
EOLineLine: solves EO and two lines of edges, LF + LB and DF + DB
Left line: solves a triplet (corner + edge + corner) in DL
Right line: solves a triplet in DR
Edge lines: solves two lines of edges, UF + UB and RF + RB (so that all edges except for UL and UR are solved now)
Last 2 Lines (or just ZZLL if that was too much lines for you): Solves two triplets in UL and UR, respectively, finishing the solve
So recently I had the idea that you could solve some kind of EOLine and EOEdge in one step for ZZ. It didn't really end in a great method, but I thought I'd still share it here.
It's called ZZ-Line and the steps are:
EOLineLine: solves EO and two lines of edges, LF + LB and DF + DB
Left line: solves a triplet (corner + edge + corner) in DL
Right line: solves a triplet in DR
Edge lines: solves two lines of edges, UF + UB and RF + RB (so that all edges except for UL and UR are solved now)
Last 2 Lines (or just ZZLL if that was too much lines for you): Solves two triplets in UL and UR, respectively, finishing the solve
Interesting, but I sorta think that if you're going to solve two extra edges during step 1, you might as well just make them the other two cross edges followed by rotationless F2L and ZBLL.
