One of the biggest drawbacks of Roux

[Spencer131]

CMLL recognition is definitely one of the biggest drawbacks of the Roux method. From analyzing my solves I figured out that CMLL recognition takes me over a second on average. This is probably because I can only recognize most cases from one angle, and it would probably be a good place to start to learn more angles. Even Kian, who recognizes from multiple angles and uses prediction sometimes, eats up a lot of time just by recognizing CMLL. I analyzed one his Ao5 videos on youtube, and his recognition was .53 seconds on average. It seems a bit silly to me that a 6.5 second solver should spend half a second doing nothing!

My point is, maybe we should come up with a way to eliminate CMLL recognition time. Or if there is some method I'm not aware of, that should become way more popular.

 

[Mastermind2368]

I personally don't find COLL rec very hard and I guess its the same for Roux. The only way to remove time would to be predicting the CMLL. I can predict the OCLL, and from that Its much easier to have it narrowed down to 7 algs.

 

[shadowslice e]

CMLL recognition is definitely one of the biggest drawbacks of the Roux method. From analyzing my solves I figured out that CMLL recognition takes me over a second on average. This is probably because I can only recognize most cases from one angle, and it would probably be a good place to start to learn more angles. Even Kian, who recognizes from multiple angles and uses prediction sometimes, eats up a lot of time just by recognizing CMLL. I analyzed one his Ao5 videos on youtube, and his recognition was .53 seconds on average. It seems a bit silly to me that a 6.5 second solver should spend half a second doing nothing!

My point is, maybe we should come up with a way to eliminate CMLL recognition time. Or if there is some method I'm not aware of, that should become way more popular.

42?

Or just more practise. How does recog in other methods compare? Also, which Ao5 and how many of them and how did you get that number?

 

[The Pocket Cuber]

42?

Or just more practise. How does recog in other methods compare? Also, which Ao5 and how many of them and how did you get that number?

Recognition of CMLL is a lot like OLL recognition, which you can get under a sec with practice

 

[Sue Doenim]

That's not really a drawback. If you think about it, it's arguably better than other methods. In CFOP, you have to recognize OLL and PLL, and in ZZ, you have to recognize at least 2 different things, unless you're crazy and do ZBLL, which is definitely slower to recognize.

 

[Underwatercuber]

That's not really a drawback. If you think about it, it's arguably better than other methods. In CFOP, you have to recognize OLL and PLL, and in ZZ, you have to recognize at least 2 different things, unless you're crazy and do ZBLL, which is definitely slower to recognize.

You make it sound as though it’s better since you are need to recognize less steps in LL. Roux you need to recognize CMLL and L6E which means you have more pieces to regognize and keep track of (even compared to zbll)

 

[efattah]

Eliminating the CMLL recognition time has been the main idea behind several methods such as LMCF and WaterRoux. By eliminating CMLL recognition and putting it into the inspection, you gain a great deal, but you also lose. For example in LMCF you are unable to avoid bad LSE cases like you can in Roux if you use multiple CMLL's.

 

[Thom S.]

Well, compared to OLL/PLL, CMLL/LSE is actually quite fast.
If your point is to eliminate COLL Recognition, ZBroux/LLOB is your way to go, as EO+ DFDB is super fast and you can recognise your COLL on the fly

 

[efattah]

Well, compared to OLL/PLL, CMLL/LSE is actually quite fast.
If your point is to eliminate COLL Recognition, ZBroux/LLOB is your way to go, as EO+ DFDB is super fast and you can recognise your COLL on the fly

Yes, excellent point, ZBRoux/LLOB does solve the recognition problem without introducing other problems other than lots of algorithms.

 

[Sue Doenim]

You make it sound as though it’s better since you are need to recognize less steps in LL. Roux you need to recognize CMLL and L6E which means you have more pieces to regognize and keep track of (even compared to zbll)

Fair point. I guess I was thinking of L6E more like F2L where you're tracking pieces and turning continuously instead of recognizing cases. Ideally, it would be, but not quite. I still don't think it's really exactly the same.

 

[Thom S.]

Yes, excellent point, ZBRoux/LLOB does solve the recognition problem without introducing other problems other than lots of algorithms.

Yes, ZBLL is quite a beast(although it's one of the reasons I use ZBroux), but due to EO recognition, with COLL, you only have one pause for EPLL

 

[Roussell]

I was once a cfop user before I started useing roux and I could predict my oll 50% of the time I'm realy bad at it but this is how I "look ahead" to cmll

cmll is actually pretty easy to look ahead to corners because you can predict the oll from tracking 3 corner during last pair incertion. Now there are only 42 cmlls and if I can predict the oll that number gets narrowed down to 1 of 6 algs and if you predict you can immediately auf to recog - alg
Now I'm practicing this I'm no where close to mastering it I get the oll right 80% of the time but if I mix up the oll i pause longer.
So that's how I'm practicing to reduce cmll recog time am I doibg it right or wrong I'd love to hear opinions