[Help Thread] Roux Discussion and Help

Hey!
I just started learning roux and I frequently stumble upon a case in my 4c that I cannot solve.
Here is the image of the case:
4c_stuck.png
And the scramble to get to it: B' B' U' L' U R' B U L R' U' F' L U' R U'
It seems that this is not one of the standard cases that can be found in Kian's video or other ressources.
Any hints would be appreciated!
 
Hi,
your centers are not good: if your top is green, you should do M moves to have the green or blue center on top.
An exemple: 
Code: 
M' // good center (or M)
U M' U2 M // make arrow
U2 M U' M // all green/blue
U2 M' U2 M U' M2 U // UL/UR
M U2 M' U2 M2
 
narusite said:
Hi,
your centers are not good: if your top is green, you should do M moves to have the green or blue center on top.
An exemple: 
Code: 
M' // good center (or M)
U M' U2 M // make arrow
U2 M U' M // all green/blue
U2 M' U2 M U' M2 U // UL/UR
M U2 M' U2 M2
Click to expand...
That was it! I forgot this step in 4a. Thanks a lot!
 
Some thoughts on Pinkie Pie. It is a proposal for Roux where the UL+UR edges are placed, oriented, at DF+DB after F2B. Then the user does OLLCP and ends with L6EP.

The first step of placing the two L/R edges on the D layer is almost equivalent to the common proposal of combining Roux and CFOP by solving F2B then placing the DF+DB edges. Maybe around a move more efficient. If that Roux+CFOP proposal is seen as not a good way to build F2L and end with OLL and PLL, then maybe Pinkie Pie should be viewed the same. Maybe it should be viewed as better to do a cross with the UL+UR edges at DF+DB then the four F2L pairs and end with OLLCP and L6EP. It is the non-EO version of ZZ-4c, which also isn't highly regarded. More details:

Movecount

Normal Roux:
  • CMLL - .75 AUF + 10.5 alg = 11.25
  • LSE - 14
  • Total - 25.25
Pinkie Pie:
  • LR to DFDB - 4
  • OLLCP - .75 AUF + 12 alg (based on current docs and those developing OLLCP) = 12.75
  • L6EP - 7
  • Total - 23.75
Algorithms

Normal Roux:
  • CMLL - 42
  • LSE - 0, intuitive cases for EOLR
Pinkie Pie
  • OLLCP - 331
Ergonomics

Normal Roux:
  • CMLL is free to disturb six edges and four centers. This makes for a lot of freedom in finding great algorithms.
Pinkie Pie:
  • OLLCP isn't able to disturb much and it has a very specific goal with more than one step combined. This means the algorithms are more restrictive and won't be as ergonomic overall compared to CMLL.
Number of Looks

Normal Roux:
  • CMLL is one look. Some users can predict the OCLL before getting to the step, making CMLL take less time to recognize.
  • LSE is three sub steps but is mostly two looks with one of those looks (M slice EP) usually occurring during lookahead.
Pinkie Pie:
  • LR to DFDB is one look. Much can be seen during F2B.
  • OLLCP is one look. The CLL case or EO can likely be seen during the previous step. Overall a big longer to recognize compared to CMLL.
  • L6EP is one look. Part of the case can be seen during OLLCP.
So overall you would be learning 331 algorithms which are less ergonomic and slower to execute than CMLL just to save maybe 1-2 moves. With OLLCP recognition being slightly longer and the algorithms for that step being worse overall than CMLL, that 1-2 move save is cancelled out or possibly even brought into the negative compared to normal Roux. There are other Roux advancements which I think should be the focus for Roux users. Advancements which save many more moves, have better ergonomics, or require fewer algorithms to memorize.
 
Last edited:
Athefre said:
Some thoughts on Pinkie Pie. It is a proposal for Roux where the UL+UR edges are placed, oriented, at DF+DB after F2B. Then the user does OLLCP and ends with L6EP.

The first step of placing the two L/R edges on the D layer is almost equivalent to the common proposal of combining Roux and CFOP by solving F2B then placing the DF+DB edges. Maybe around a move more efficient. If that Roux+CFOP proposal is seen as not a good way to build F2L and end with OLL and PLL, then maybe Pinkie Pie should be viewed the same. Maybe it should be viewed as better to do a cross with the UL+UR edges at DF+DB then the four F2L pairs and end with OLLCP and L6EP. It is the non-EO version of ZZ-4c, which also isn't highly regarded. More details:

Movecount

Normal Roux:
  • CMLL - .75 AUF + 10.5 alg = 11.25
  • LSE - 14
  • Total - 25.25
Pinkie Pie:
  • LR to DFDB - 4
  • OLLCP - .75 AUF + 12 alg (based on current docs and those developing OLLCP) = 12.75
  • L6EP - 7
  • Total - 23.75
Algorithms

Normal Roux:
  • CMLL - 42
  • LSE - 0, intuitive cases for EOLR
Pinkie Pie
  • OLLCP - 331
Ergonomics

Normal Roux:
  • CMLL is free to disturb six edges and four centers. This makes for a lot of freedom in finding great algorithms.
Pinkie Pie:
  • OLLCP isn't able to disturb much and it has a very specific goal with more than one step combined. This means the algorithms are more restrictive and won't be as ergonomic overall compared to CMLL.
Number of Looks

Normal Roux:
  • CMLL is one look. Some users can predict the OCLL before getting to the step, making CMLL take less time to recognize.
  • LSE is three sub steps but is pretty much one look and can be executed almost as fast as if the step was a single algorithm. EO can be seen during execution of CMLL.
Pinkie Pie:
  • LR to DFDB is one look. Much can be seen during F2B.
  • OLLCP is one look. The CLL case or EO can likely be seen during the previous step. Overall a big longer to recognize compared to CMLL.
  • L6EP is one look. Part of the case can be seen during OLLCP.
So overall you would be learning 331 algorithms which are less ergonomic and slower to execute than CMLL just to save maybe 1-2 moves. With OLLCP recognition being slightly longer and the algorithms for that step being worse overall than CMLL, that 1-2 move save is cancelled out or possibly even brought into the negative compared to normal Roux. There are other Roux advancements which I think should be the focus for Roux users. Advancements which save many more moves, have better ergonomics, or require fewer algorithms to memorize.
Click to expand...
Yeah I completely agree. Tcmll and Acmll are definitely more useful.Plus LSE move count can easily be reduced by using MC, UFUB and EOLR.
 
Roux LSE to L5E Reduction System

I recently developed an alternative to Roux LSE. It is an LSE to L5E reduction system. It consists of a basic version and an advanced version.

Basic
  1. Orient all edges while solving any edge. This means either slotting a U layer edge or solving the DF or DB edge. If slotting a U layer edge, it is completely intuitive and can be done during EO. Algs can also be generated for the rare situations where the typical EO doesn't solve an edge. If solving a D layer edge during EO, algs have been generated for this set.
  2. Permute the last five edges (L5EP). If in the first step a D layer edge was placed, L5EP will be the four edges on the U layer and the remaining edge on the D layer. If a U layer edge was solved in the first step, then L5EP will be the three edges on the U layer and the two edges on the D layer. U edge solved L5EP is 28 cases. D edge solved L5EP is 16 cases.
Advanced
  1. Solve any edge. So either slot a U layer edge or solve the DF or DB edge. Just like the basic version of the system but without the requirement of EO.
  2. Solve the last five edges (L5E). If in the first step a D layer edge was placed, L5E is the four edges on the U layer and the edge on the D layer. If a U layer edge was solved, L5E is the three edges on the U layer and the two edges on the D layer. D edge solved L5E is 245 cases and U edge solved L5E is ~400 cases.
For move-counts within the system, the U edge versions beat the D edge versions. EO+U Edge then L5EP has an average of 13.53 moves. This means that after learning just 28 algs, the move-count is better than the typical EO -> L+R -> M slice method. It is also shorter than EOLR and has fewer cases. L5EP is only 28 cases and EOLR is ~60.

The advanced U Edge (no EO) then L5E has an average of 11-12 moves. That is comparable to EOLRb, except L5E again has fewer cases. This advanced non-EO L5E set is for when people feel that they have reached the limit of how fast they can execute the EO+U Edge -> L5EP set.

Most algs have been generated. All that's left is L5E for 3U1D, which is a lot of cases, and EO + Any U edge for those that want to learn algs.
 
Last edited:
Not sure if this is good or not, but for dots, instead of rotating to do the E2 M' E2 M', what if you did an R' to get into a better grip without having to regrip?

For example, for a pure dots case, you could do Rw' E2 M' E2 R.

And then, for pure dots + M2, you could do R' E2 M' E2 Rw.

Of course, there are two other cases for when you misoriented centers:
R' E2 M' E2 R
R' E2 M E2 R

Doing this may not be worth it for all of the cases, since regripping once isn't actually that bad. This is just an idea that I came up with.
 
Brouxt Force said:
Not sure if this is good or not, but for dots, instead of rotating to do the E2 M' E2 M', what if you did an R' to get into a better grip without having to regrip?

For example, for a pure dots case, you could do Rw' E2 M' E2 R.

And then, for pure dots + M2, you could do R' E2 M' E2 Rw.

Of course, there are two other cases for when you misoriented centers:
R' E2 M' E2 R
R' E2 M E2 R

Doing this may not be worth it for all of the cases, since regripping once isn't actually that bad. This is just an idea that I came up with.
Click to expand...
people usually cancel a move and do the U2 M2 solution its way better than regripping for EM dots
 
I have a problem finding the edge and corner pieces during the blockbuilding. This is easily what takes the most time out of my solves; with CFOP this was way easier
Any recommendations?
 
Pax said:
I have a problem finding the edge and corner pieces during the blockbuilding. This is easily what takes the most time out of my solves; with CFOP this was way easier
Any recommendations?
Click to expand...
It's just different. It is normal to struggle with it just like when you started learning CFOP. I don't have any tips other than just keep trying. All I can say is this, remember where pieces were and track them. Keep looking for pieces. It isn't harder than CFOP, it is just different.
 
Pax said:
I have a problem finding the edge and corner pieces during the blockbuilding. This is easily what takes the most time out of my solves; with CFOP this was way easier
Any recommendations?
Click to expand...
Use inspection as much as possible. I also suspect that you may not be used to the sort of weird color neutrality of Roux. Most (read: almost all) solvers will only use yellow or white as the bottom color, but be able to use any of the other colors as the first and second block colors. Make sure that you know what the left color of the first block and the right color of the second block is before you start solving. That way you know what pieces you have to search for.

Let me know if this helps.
 
Im trying my go with roux to see if i can catch up if not, surpass my avg with cfop but im having trouble with block building since i keep using my cfop habits and wasting a lot of time with my first and second block. Does anyone know a good in-dept video that teaches good block building strategies and how to improve first and second block?

Im trying my go at roux to see if i can surpass my avg of cfop and im trying to transition from using beginner block build to intermediate to advance. I went through a few videos but they don't really go in-dept. with explanation and i already watch Kian's video on block building.
 
Hey everyone, so I’m trying to figure out the Roux method since I have just not been getting better at CFOP but have yet been staying at sub 25, it’s been a month since I’ve been sitting around these times so I thought I should find a different method that suites me better. Please don’t link me Jperm’s tutorial because he goes to fast and it’s hard for me to understand. Don’t get me wrong I love JPerm though.
 
I would recommend focusing on block building. Experiment with different solutions. Additionally, I would recommend to stay away from the timer for most of your practice. The timer creates pressure on you to go faster, which will lower the quality of your block building.

To practice block building, you can use a block trainer, starting off with 1 or 2 moves (The "Level" is the movecount. Also, make sure you select one of the Roux options, as you're not practicing Petrus block building. I'm only saying this because I've made this mistake before). Sometimes, even with those extremely short first blocks, you might miss the optimal solution. Block building can be difficult at first, but with practice, you will get much more familiar with it.
 
You must log in or register to reply here.

Similar threads

pcwiz
[Help Thread] CFOP Discussion
144 145 146
Replies
2,918
Views
770,735
abunickabhi
abunickabhi
E
  • Poll Poll
Roux CN discussion (x2/y vs full CN)
2
Replies
22
Views
3,814
abunickabhi
abunickabhi
C
[Free App] Cube Trainer — CFOP + Roux algorithm trainer with timer
Replies
0
Views
117
cellonty
C
T
New method idea: Using Pinkie Pie roux variant with CFOP instead
Replies
4
Views
408
theail03
T
Roux
Should i become full CN with Roux?
Replies
4
Views
536
Roux
Roux
Back
Top