[Help Thread] Roux Discussion and Help

[mafergut]

You should definitely learn 2 look cmll first. Its only 9 algs, and 3 of them are almost the same (f sexy f', f 2 sexy f', f 3sexy f'). Sune and antisune are mirror, so basically its just 6 unique algs.

Hi, Only so that you know I searched the forum before posting my question
This post by PenguinsDontFly was responding to a question by kylan back in December, where he asked whether to go from DeeDubb's basic (Niklas+Sune) CMLL approach directly to full / 1-look CMLL or go through 2-look CMLL first. My question and story have a lot to do with that. Here I go.

I am a CFOP solver but after reading many people talk about Roux (and ZZ, and Petrus...) I decided to take a look at the method. I watched the beginners videos by DeeDubb and I got interested but, of course, once you go past beginner stage, even a more intuitive method (comparing to CFOP) like Roux involves some hard alg learning.

As a CFOP solver I just know the 57 OLL cases and 21 PLLs (well, some additional algs for other PLL orientations, as well, but only a handful). What has turned me away from getting into more advanced stuff like ZBLL, etc is the sheer amount of algs needed.

And now, I see that, even for Roux I would need at least like 40-something new algs (or 20-sth depending if you count mirrors or inverses or ...?) for CMLL and some others (okay, easy ones) for OE & LSE. I checked Waffo's tutorial. I found it to be an excelent tutorial, at least for what I was looking for, that is CMLL, EO & LSE but I was hoping the amount of algs would be much less for Roux.

So, is this what it takes to get within the 20-30 second realm with Roux? Or should I just start with 2-look CMLL? Penguins seemed to recommend going through 2L cmll first.

I assume, apart from the algs given by Waffo, using any normal OLLs + J/L/T-perm for adj swap + Y-perm for diag swap should be okay, right? Is there anything wrong with that approach, even if not as efficient? Of course the OLLs can be improved disregarding the edges, like you would do to to use OLLs in a 2x2, right? Which basically takes me to Waffo's guide anyway

By the way, when Waffo says: Diag Swap/NOT N-Perm, what does he mean by that?

 

[supercavitation]

Hi, Only so that you know I searched the forum before posting my question
This post by PenguinsDontFly was responding to a question by kylan back in December, where he asked whether to go from DeeDubb's basic (Niklas+Sune) CMLL approach directly to full / 1-look CMLL or go through 2-look CMLL first. My question and story have a lot to do with that. Here I go.

I am a CFOP solver but after reading many people talk about Roux (and ZZ, and Petrus...) I decided to take a look at the method. I watched the beginners videos by DeeDubb and I got interested but, of course, once you go past beginner stage, even a more intuitive method (comparing to CFOP) like Roux involves some hard alg learning.

As a CFOP solver I just know the 57 OLL cases and 21 PLLs (well, some additional algs for other PLL orientations, as well, but only a handful). What has turned me away from getting into more advanced stuff like ZBLL, etc is the sheer amount of algs needed.

And now, I see that, even for Roux I would need at least like 40-something new algs (or 20-sth depending if you count mirrors or inverses or ...?) for CMLL and some others (okay, easy ones) for OE & LSE. I checked Waffo's tutorial. I found it to be an excelent tutorial, at least for what I was looking for, that is CMLL, EO & LSE but I was hoping the amount of algs would be much less for Roux.

So, is this what it takes to get within the 20-30 second realm with Roux? Or should I just start with 2-look CMLL? Penguins seemed to recommend going through 2L cmll first.

I assume, apart from the algs given by Waffo, using any normal OLLs + J/L/T-perm for adj swap + Y-perm for diag swap should be okay, right? Is there anything wrong with that approach, even if not as efficient? Of course the OLLs can be improved disregarding the edges, like you would do to to use OLLs in a 2x2, right? Which basically takes me to Waffo's guide anyway

By the way, when Waffo says: Diag Swap/NOT N-Perm, what does he mean by that?

He means that it'll swap the corners diagonally like an N perm, but it isn't an N-perm, and you shouldn't think it is.

 

[mafergut]

He means that it'll swap the corners diagonally like an N perm, but it isn't an N-perm, and you shouldn't think it is.

Well, yeah, I have tried his algs and already realized they do not preserve the M-slice so, answering my other question, if comfortable with Y-perm or similar I can, for now, use that instead until perhaps I want to go all the way to 1-look CMLL.

 

[TheOneOnTheLeft]

You can use Y-perm for 2-look CMLL - that's what I always used. Also, try going through your OLLs to see what corner cases they solve (that is, what permutation, not just what orientation) and chances are you know algs for about half of the 1-look CMLL cases anyway. Whether you want to 2-look the remaining cases or learn algs for them is up to you at that point.

 

[mafergut]

You can use Y-perm for 2-look CMLL - that's what I always used. Also, try going through your OLLs to see what corner cases they solve (that is, what permutation, not just what orientation) and chances are you know algs for about half of the 1-look CMLL cases anyway. Whether you want to 2-look the remaining cases or learn algs for them is up to you at that point.

Thanks for the great tip. I will definitely look into that. To be honest, what made me get interested in Roux was, among other things that I'm getting a bit bored of just trying to get faster and faster with CFOP (without much success, I'm probably close to my limit) and, in order to remain interested in cubing I needed to get involved in new things that spark my curiosity.

Only problem I see with 1-look CMLL (and, hence, with your suggestion) is that I find it terribly difficult to recognize permutation cases when corners are still not oriented. Those sticker-color patterns make me mad.

 

[TDM]

Only problem I see with 1-look CMLL (and, hence, with your suggestion) is that I find it terribly difficult to recognize permutation cases when corners are still not oriented. Those sticker-color patterns make me mad.

Start with H and Pi, they aren't too bad. Neither are T or U. Recognising them is actually quite easy. L, S and A are the harder sets to recognise, but using 2-look for S/A isn't as slow as using 2-look for other subsets because their OCLL algs are fast, so you can leave them until later.

 

[mafergut]

Start with H and Pi, they aren't too bad. Neither are T or U. Recognising them is actually quite easy. L, S and A are the harder sets to recognise, but using 2-look for S/A isn't as slow as using 2-look for other subsets because their OCLL algs are fast, so you can leave them until later.

Thanks! I will also try to follow your advice, but I guess I'll wait until I can get sub-30 or -35 with 2-look CMLL.

Also, clarifying what TheOneOnTheLeft told me about analysing the effect of my OLLs on the corner permutation, what you mean is that, e.g.: the T-case OLL F sexy F', permutes the front corner pair and also the back corner pair, which, for all effects and purposes in CMLL means a diag swap, right? So, If I get the headlights case with a diag swap, the T OLL will do 1-look CMLL. Now I just should go to check the tutorial and test my reasoning, hehehe, because I am sure this is the alg for that case, as there cannot be another one that's faster.

EDIT: I just checked and, my assumption was correct!!! It is E6 case. Hooray, now I feel that I am starting to get it. Thanks all for your help. Now the decision is between pushing for sub-20 CFOP or getting serious with Roux (sub-minute for now, I'm not too ambitious)

 

[TheOneOnTheLeft]

There are also a surprising number of cases where a combination of Sune and Niklas will result in a good alg. A lot of the 2-gen cases are sune combos, but other examples are G2, G4 and G5, as well as D4 being basically a weird Niklas (l' U' L' U R U' L U x'). Learning these easy cases will give you a pretty good head start on full CMLL and then you can add in the other ones whenever you feel like it. I didn't finish CMLL until about a 17 second average anyway.

 

[Ross The Boss]

current avg100 = 13.99 (σ = 1.35)
ahh, has been a while since ive seen this.

(18.14), 17.35, (11.70), 13.55, (20.14), 13.60, 14.47, 15.63, 13.49, 16.52+, 12.94, 16.51, 15.15, 14.53, 15.20, 12.93, 12.98, 16.02, 16.29, 16.07, 13.36, 14.19, 14.00, 13.79, 17.24, 13.08, 13.62, (10.64), 16.16, 14.12, 14.47, 14.07, 13.88, 13.05, 13.46, 14.13, 13.70, 13.35, 13.28, 13.05, 16.38, 12.19, 12.43, 13.11, 16.33, (11.46), 13.02, 11.90, (17.66), 14.43, 13.51, (17.68), 12.89, 14.01, 12.65, 13.74, 12.06, (11.77), 16.50, 15.65, 13.84, 12.71, 13.10, 14.89, 16.19, 13.19, 13.58, 12.38, 11.88, 14.39, (19.57), 14.80, 13.70, 12.10, 15.34, 15.17, 15.97, 13.27, 12.86, 12.53, 14.57, 13.93, 14.89, 13.36, 11.97, 12.19, 12.45, 15.18, 14.15, 14.98, 13.65, 14.17, 14.71, 12.79, 13.67, 12.20, 13.21, (10.28), 12.19, 12.99
__________________________________________

I think the issue is that one person posted his accomplishment so others could celebrate/encourage his personal progress, and PDF (much simpler and more obvious abbreviation of penguinsdontfly....) used it as an opportunity to engage in interpersonal competition, which was taken by others as an attack on his personal achievement. So the issue is whether posts about personal achievements should be taken as invitations for others to use said achievements as goals to beat and then subsequently post those personal accomplishments for others to encourage. I remain neutral on the topic.

ya, it is. i think that it's silly to be offended by something like this, but also perfectly okay if you are. be competitive if you want, but beware retaliation. the only way to offend nobody is to do nothing, but thats boring.

 

[rybaby]

Ross, bear in mind that I didn't post at all in the first place. PDF posted about a YouTube video I uploaded and used that post to talk about himself, as he is likely to do. If you need further clarification, refer back to my original post. anyway, I've got other things to do than to sit by this thread. Later-

Ryan

 

[TDM]

Got my second-best Roux single earlier:

253. 8.07 L2 D' F2 U2 F2 R2 D2 U' R2 D' L2 B' L' D' B2 U F2 U2 F L B D

x // Inspection
D2 L' D' // F2B-1 (3/3)
R' U2 R U' B // F2B-2 (5/8)
U R' U R U' R' U2 R U R // F2B-3 (10/18)
R U' R' U2 R U' r' // F2B-4 (7/25) (CMLL skip)
M' U' M U M' U' M' // EO (7/32)
U' M' U2 M U' M2 // ULUR (6/38)
U' M U2 M // EP (4/42)

5.20 TPS.

 

[Ross The Boss]

Got my second-best Roux single earlier:

253. 8.07 L2 D' F2 U2 F2 R2 D2 U' R2 D' L2 B' L' D' B2 U F2 U2 F L B D

x // Inspection
D2 L' D' // F2B-1 (3/3)
R' U2 R U' B // F2B-2 (5/8)
U R' U R U' R' U2 R U R // F2B-3 (10/18)
R U' R' U2 R U' r' // F2B-4 (7/25) (CMLL skip)
M' U' M U M' U' M' // EO (7/32)
U' M' U2 M U' M2 // ULUR (6/38)
U' M U2 M // EP (4/42)

5.20 TPS.

Speedy. Good one! What's up with the splits though? If I'm not wrong, FB is supposed to be written all on one line..?
-----------------

Also: "CMLL.". Why? the corners solved with these algorithms are not of the 'last layer.' in fact, the actual last layer has no corners. Ita not awful, but it bothers me.

 

[TDM]

Speedy. Good one! What's up with the splits though? If I'm not wrong, FB is supposed to be written all on one line..?
-----------------

Also: "CMLL.". Why? the corners solved with these algorithms are not of the 'last layer.' in fact, the actual last layer has no corners. Ita not awful, but it bothers me.

I don't know, I write each step on a separate line. I don't usually look ahead as far as the second part of the FB, so they're more like separate steps to me.

I think the reason why it's called CMLL is that most other methods that have a step involving the corners has the U layer as the last layer: CLL, COLL etc. I think CMLL was just used to be consistent. But now you point it out, yeah, it is a bad name. I've never noticed that before. Do you have any idea what a better name would be? I can't really think of anything... my best idea was: 1a, 1b, 2a, 2b, 3, 4a, 4b, 4c. Which isn't a good idea at all

 

[Phaint]

About a week ago I was looking to learn a new method as I was finding CFOP to getting repetitive and boring with so many algs and simple execution so I looked at more efficient methods like Heise, Snyder, and Petrus, two days ago I started with Roux found it very fun and I'm now sub 40. The biggest problem I have been having is the first block (F2L form CFOP has created many bad habits) does anyone any good sources for example solves for the first block. I have tried memorizing the cube and doing it blindfolded but although this helps I still have trouble seeing more efficent ways to build the first block I know the first block can usually be done in 8-10 moves but have trouble doing it in under 14 with 30 seconds inspection. I was going to read through this whole thread but after 50 pages and an hour-ish of external research I decided to just ask. Sorry for newby question I've been doing un-timed slow solves and find it has not helped mush with block building any links to solve examples or some tricks to making corner edge pairs and such would be greatly appreciated.

On another note I what kind a progression can you expect with this method I was sub 20 with CFOP before and was wondering how much work it will take to get there again? Thanks again for the help!

 

[GuRoux]

About a week ago I was looking to learn a new method as I was finding CFOP to getting repetitive and boring with so many algs and simple execution so I looked at more efficient methods like Heise, Snyder, and Petrus, two days ago I started with Roux found it very fun and I'm now sub 40. The biggest problem I have been having is the first block (F2L form CFOP has created many bad habits) does anyone any good sources for example solves for the first block. I have tried memorizing the cube and doing it blindfolded but although this helps I still have trouble seeing more efficent ways to build the first block I know the first block can usually be done in 8-10 moves but have trouble doing it in under 14 with 30 seconds inspection. I was going to read through this whole thread but after 50 pages and an hour-ish of external research I decided to just ask. Sorry for newby question I've been doing un-timed slow solves and find it has not helped mush with block building any links to solve examples or some tricks to making corner edge pairs and such would be greatly appreciated.

On another note I what kind a progression can you expect with this method I was sub 20 with CFOP before and was wondering how much work it will take to get there again? Thanks again for the help!

roux example solve thread should be very helpful. it shouldn't bee too hard to get back to your cfop times. the main things you need to work on is overall move efficiency. full cmll should be a good idea. usually when people start out, their lse is very ineficient. try to get sub 18 moves for it most of the time.

 

[Ross The Boss]

The biggest problem I have been having is the first block (F2L form CFOP has created many bad habits) does anyone any good sources for example solves for the first block.

here you are, my good sir! https://www.speedsolving.com/forum/...The-Roux-quot-Example-Solve-quot-Game!/page66


______________________________--

and @TDM, an alternative name for CMLL could be CUF. you can pronounce it in one syllable and it sounds pretty cool. it would stand for Corners (of) Upper Face. there isnt really anything about the name that would suggest it's different from any of the cxll sets, as they all solve the U corners, but the fact that it isn't making reference to the LL is -to me- a good enough distinguishing feature.

 

[Phaint]

roux example solve thread should be very helpful. it shouldn't bee too hard to get back to your cfop times. the main things you need to work on is overall move efficiency. full cmll should be a good idea. usually when people start out, their lse is very ineficient. try to get sub 18 moves for it most of the time.

here you are, my good sir! https://www.speedsolving.com/forum/...The-Roux-quot-Example-Solve-quot-Game!/page66


______________________________--

and @TDM, an alternative name for CMLL could be CUF. you can pronounce it in one syllable and it sounds pretty cool. it would stand for Corners (of) Upper Face. there isnt really anything about the name that would suggest it's different from any of the cxll sets, as they all solve the U corners, but the fact that it isn't making reference to the LL is -to me- a good enough distinguishing feature.

Thanks you good sirs definitely will be learning full CMLL kinda forgot about it. (I may also have to start calling it CUF, CMLL certainly doesn't roll off the tongue as well)

 

[TDM]

an alternative name for CMLL could be CUF. you can pronounce it in one syllable and it sounds pretty cool. it would stand for Corners (of) Upper Face. there isnt really anything about the name that would suggest it's different from any of the cxll sets, as they all solve the U corners, but the fact that it isn't making reference to the LL is -to me- a good enough distinguishing feature.

Maybe CUL (Corners of the U Layer)? To me, 'face' doesn't mean permuted, especially for stuff like 2x2, where a face and a layer have different meanings. I know this isn't a 2x2, but it is just corners - and anyway, everyone else talks about Layers, so it would be sensible/consistent to also use L.

Spoiler

or how about just C ( = Corners)?

 

[mafergut]

Oh, well. I have started to put it all together, after the encouragement I got in this thread some days ago. My times are still horrible (over a minute) but I'm making progress:
- Block building is still very inefficient but I more or less get it.
- CUF / CUL, (a.k.a. CMLL) is slow but okay with 2-look with already known OLLs + T/Y-perm
- EO... see below
- ULUR is quite intuitive with just put them diagonally opposed, flip the upper one so that they both are down, adjust U and put them up, with some evident shortcuts depending on the case.
- The L4E or M-slice or whatever you call it is also quite intuitive and I can mostly do it without any algs, except for the dots case E2 M E2 M', which is also easy to remember. Well, there's a case that takes a bit more fidling around and sometimes takes me a couple of tries but the others are all just some M/M' here some U2 there and done

You will see that there's something I left blank in the middle. Damn EO is what, right now, is giving me a headache. I know the "arrow case" to solve the EO but I don't know anything else and I can't, for now, see easy intuitive ways of going from all other cases to the arrow so I basically fiddle around until I get the arrow, which at times can be like more than 6 tries

So, my question is, do you memorize the EO cases as algorithms or, at least do you memorize how to cycle from one case to another until arrow and then solved? I assume with time it will get automated like F2L insertion cases in CFOP but, right now I don't see it is all that intuitive. What am I missing?

Thanks

 

[Ross The Boss]

Oh, well. I have started to put it all together, after the encouragement I got in this thread some days ago. My times are still horrible (over a minute) but I'm making progress:
- Block building is still very inefficient but I more or less get it.
- CUF / CUL, (a.k.a. CMLL) is slow but okay with 2-look with already known OLLs + T/Y-perm
- EO... see below
- ULUR is quite intuitive with just put them diagonally opposed, flip the upper one so that they both are down, adjust U and put them up, with some evident shortcuts depending on the case.
- The L4E or M-slice or whatever you call it is also quite intuitive and I can mostly do it without any algs, except for the dots case E2 M E2 M', which is also easy to remember. Well, there's a case that takes a bit more fidling around and sometimes takes me a couple of tries but the others are all just some M/M' here some U2 there and done

You will see that there's something I left blank in the middle. Damn EO is what, right now, is giving me a headache. I know the "arrow case" to solve the EO but I don't know anything else and I can't, for now, see easy intuitive ways of going from all other cases to the arrow so I basically fiddle around until I get the arrow, which at times can be like more than 6 tries

So, my question is, do you memorize the EO cases as algorithms or, at least do you memorize how to cycle from one case to another until arrow and then solved? I assume with time it will get automated like F2L insertion cases in CFOP but, right now I don't see it is all that intuitive. What am I missing?

Thanks

I learnded how to do eo by solving directly to the arrow and then the final MUM to finish it off. That really matters s the best way at to do it. but should you figure it out on your own, or through learnt algs? I don't think that it really matters, as long as you end up under standing the process. Understanding helps with optimization.

A tip for you: if you are having problems visualizing a way to solve EO, take one of the bad edges from d (if there are none, do an M2 to put one or two there), and then put one of the good edges on u directly above it, and then do a M/M' to bring that bad edge to u. You will have misoriebted the centers, but you will usually be a little closer to the arrow.



And yes, TDM. That is a better argument. CUL is more accurate than CUF