Move/look count for L4C/L5C for sandwich/cage

[irontwig]

So I was messing around with the cage method today and it got me wondering how many looks and/or moves the centre stage takes for people who use centre last type method for different cube sizes and also how many different (kinds of) algorithms you use for these. It's seems to me that their are at least three ways to do centres; one colour at a time, seperation first , and of course just freestyling it.

 

[TMOY]

I use separation first for L4C, it seems to me the best compromise between intuitiveness and low number of looks/moves; Note that it's really a L4C method, it doesn't adapt well to L5C.
For 4^3, it's theoretically possible to solve centers in 2 looks (1 for separation and 1 for solving both pars of opposite centers at once) but I take more lke about 4 looks on average. For 5^3 I would say 7-8 looks or something like that on average (once again it's possible to reduce their number with more algs). For bigger cubes my centers solving gets more freestylish so it's hard to count.
The idea behind the method is to use as few niklas as possible by making a wise use of the 4-move comms, but of course the niklas is sometimes unavoidable. So I use mainly these two kinds of algs, plus some more algs which are only combinations of shorter algs of these two kinds (for example an alg like l u2 M' u2 r' is simply two 4-move comms merged together with some moves canceling).

 

[rachmaninovian]

hi. i'm mostly retired now...got too busy =/

anyway, i have an algorithm that i follow losely -
1. pick easiest center to solve, rotate such that it's on the top layer
2. solve pieces from the opposite center (on D, that is), followed by pieces from the front center. For example if the top center is supposed to be red, I would make sure the red *****es from the bottom layer go on top and those on the front to go on top. this is generally the guideline...though sometimes it's good to push something from the top to the front.

3. now, if the top layer is solved, do an x rotation, and repeat step 2.
if it isn't solved, do an x' rotation, and repeat step 2.

4. repeat 3 lol.

of course there are exceptions...and sometimes i cycle between 3 centers. I know just a few random algs that reduce my alg count sometimes...like a total 3 center cycle. and of course, sometimes there aren't pieces from the opposite center to solve, and sometimes there aren't pieces from the front center to solve. just do where it deems fit. =)
anyway i believe this yields a <4.5 look finish. slightly less efficient than TMOY =P but i think it gives me better lookahead. A while ago I made a new system that I never learned that can approximately give 3.5 looks on average. it's really just an approximation...and this current system I use with this loose algorithm to follow was based on that. like between freestyle and the new system.

 

[irontwig]

I use separation first for L4C, it seems to me the best compromise between intuitiveness and low number of looks/moves; Note that it's really a L4C method, it doesn't adapt well to L5C.
For 4^3, it's theoretically possible to solve centers in 2 looks (1 for separation and 1 for solving both pars of opposite centers at once) but I take more lke about 4 looks on average. For 5^3 I would say 7-8 looks or something like that on average (once again it's possible to reduce their number with more algs). For bigger cubes my centers solving gets more freestylish so it's hard to count.
The idea behind the method is to use as few niklas as possible by making a wise use of the 4-move comms, but of course the niklas is sometimes unavoidable. So I use mainly these two kinds of algs, plus some more algs which are only combinations of shorter algs of these two kinds (for example an alg like l u2 M' u2 r' is simply two 4-move comms merged together with some moves canceling).

One con I see with seperation though is that you can't look at two opposite faces without rotating and even though four movers are shorter (in half turn kind of metrics rather than e.g. sqtm) they might not be that much faster than 8 move comms.

 

[mrCage]

I guess you could call my approach freestyling, as i make up the 3-cycle commutators on the fly. I don't count cases really. But mostly i commute on top layer, not on a middle layer. That is ABA'B'. Then B will be an outer layer turn. An example to commute on inner layer would be [r U2 r',u2]. When commuting on outer layer the 3-cycle could involve 3 faces or 2 faces.

The order of which i solve the face centers is a bit random. But mostly i try to complete 3 adjacent faces. This makes the last 3 much easier.

And lastly of course 1 look is possible - how else would blindfold solving be possible? But for speed why limit yourself to 1 or 2 looks?? More important is the number of algs you can see ahead. Not the # of looks to complete the centers. Before i could sometimes see 3 algs ahead. That was 5-6 yrs ago. Now my speed s****:fp

Per

 

[TMOY]

One con I see with seperation though is that you can't look at two opposite faces without rotating and even though four movers are shorter (in half turn kind of metrics rather than e.g. sqtm) they might not be that much faster than 8 move comms.

Of course, but you can usually solve more centers at once on average with 4-movers than with 8-movers; And as Per points out, it's always possible to see 2 algs or more ahead (on bigger cubes it's not even that hard).
-

 

[rachmaninovian]

8 move comms like niklas can be done really fast, if you're looking at speed. especially if they are restricted to <r, l, U>. besides [r2,f2] and [r2,S2] i don't use any more 4 movers

 

[irontwig]

8 move comms like niklas can be done really fast, if you're looking at speed. especially if they are restricted to <r, l, U>. besides [r2,f2] and [r2,S2] i don't use any more 4 movers

Don't you mostly solve 4x4 though ? Speaking of which do you still solve corners and midges first on 5x5?

 

[mrCage]

Of course, but you can usually solve more centers at once on average with 4-movers than with 8-movers; And as Per points out, it's always possible to see 2 algs or more ahead (on bigger cubes it's not even that hard).
-

I dont use 4-move comms for centers last. Yes it can be efficient. But also harder to utilize IMHO. To add to my point about seeing algs ahead. I am also able to see ahead move cancellations between algs. Not really hard when commuting on outer layer ...

Per

 

[somerandomkidmike]

I suppose what I use would be called freestyle. For the 4x4x4, I think it is easily possible to solve in 4 looks every time for L4C, but as Per pointed out, it's more important to see the algorithms ahead of you. Sometimes you can save 2-3 seconds just from doing an algorithm that isn't optimal, rather than looking for the optimal solution. I believe Rachmaninovian solves the 4x4x4 more often. The 5x5x5 for him is something like his 3x3x3 method.

 

[mrCage]

I practise all my cubes equally lilttle these days. If i get the time I will practice a bit for the competition here in Trondheim, Norway in february. I don't want to embarass myself completely. And i have a brand new Shengshou 6x6x6 that needs breaking in and tension adjustment ...

Per

 

[Kirjava]

I'm faster at one piece at a time niklas than two at a time 3cycles.

I don't fit in this thread

 

[mrCage]

I'm faster at one piece at a time niklas than two at a time 3cycles.

I don't fit in this thread

Hmm. Even on V6 and V7 ??

Per

 

[rachmaninovian]

Don't you mostly solve 4x4 though ? Speaking of which do you still solve corners and midges first on 5x5?

the last time i was speedsolving the 5x5...it was basically, as somerandomkidmike has mentioned, my 3x3+4x4 method. first 2 centers, corners, finish tredges on yellow and white, 4x4 midges, 3x3 midges, last 4 centers.

back to topic, maybe I should revive one method I tried to develop...slightly over 3 looks on average. basically I find the easiest center to do, apply an alg that solves pieces from the bottom and front centers, then rotate to a new unsolved center. I made like 32 algs, but I guess being practical, really only 12 algs to learn. however I don't like 4 of them...so maybe just 8. down side is, <U,D> is needed for setup. anyone interested? =) anyway I counted a case like opp niklas + adjacent niklas to be 1 look, since you can basically do them back to back without pausing...and of course, recognition problems - which was the primary reason why I never learned it. LOL!

i'm trying to work on another new system, maybe I should share soon ~~