Last Layer Optimized Blockbuilding [LLOB Method]

[Alex B71]

I felt the need to add an opinion to this.

I first heard of this approach from a video by Tao Yu and since then i have been messing around with this method on and off, my first approach was to mix it with TSLE+TTLL but i would not recommend that as it is extremely inefficient.
I then took the proposed approach of EO+DFDB and then ZBLL and i can say that i love it. My previous main methods were CFOP and ZZ (Both sub-12 but CFOP was faster on average with slightly under half of ZBLL known and committed to muscle memory) so blockbuilding in this manner was difficult and EO+DFDB was confusing. I generated MU based EO+DFDB algs that were as move efficient as they could be whilst still maintaining ergonomics, and i'm very glad i took an algorithmic approach.

ZBLL recognition is pretty much dismissed with this approach as you instantly know your corner recognition after F2B and you know the edge placement from as early as just ending EO+DFDB. The EO+DFDB algs are not even "algs" and seem as easy as commutators to learn (seasoned Roux uses will find this step extremely simple). Most of the cases are reducible into another case with one move and very quickly become second nature due to the intuitiveness of the step.

To summarize, This method harnesses the low move count and awesome flexibility of F2B, has an extremely easy to learn and fast to execute EO+DFDB stage and cuts the recognition time of ZBLL drastically. It's a great hybrid of two well proven methods and has potential to be very fast. The only problems i have with this method are the changing of hand placements between steps and that fact you will need full ZBLL to really harness this method.

It's worth trying.

 

[@Ratas]

I felt the need to add an opinion to this.

I first heard of this approach from a video by Tao Yu and since then i have been messing around with this method on and off, my first approach was to mix it with TSLE+TTLL but i would not recommend that as it is extremely inefficient.
I then took the proposed approach of EO+DFDB and then ZBLL and i can say that i love it. My previous main methods were CFOP and ZZ (Both sub-12 but CFOP was faster on average with slightly under half of ZBLL known and committed to muscle memory) so blockbuilding in this manner was difficult and EO+DFDB was confusing. I generated MU based EO+DFDB algs that were as move efficient as they could be whilst still maintaining ergonomics, and i'm very glad i took an algorithmic approach.

ZBLL recognition is pretty much dismissed with this approach as you instantly know your corner recognition after F2B and you know the edge placement from as early as just ending EO+DFDB. The EO+DFDB algs are not even "algs" and seem as easy as commutators to learn (seasoned Roux uses will find this step extremely simple). Most of the cases are reducible into another case with one move and very quickly become second nature due to the intuitiveness of the step.

To summarize, This method harnesses the low move count and awesome flexibility of F2B, has an extremely easy to learn and fast to execute EO+DFDB stage and cuts the recognition time of ZBLL drastically. It's a great hybrid of two well proven methods and has potential to be very fast. The only problems i have with this method are the changing of hand placements between steps and that fact you will need full ZBLL to really harness this method.

It's worth trying.

Can you share your EO-DFDB algs?

 

[Twocubersyt]

It's basically the EOLR algs but track the df db during second block then do some nice zblls

 

[jdh3000]

I'm somewhat a late comer to considering this:

I just started looking at Roux because I really don't care for cross in cfop(yes because I'm not the best at it)

I have so much time and effort in cfop that I doubt I want to completely switch, I was just wondering if there's any merit to learning partial Roux and build the f2l from that, then finishing up with my cfop oll/pll.

I never really messed with Roux much until last night, but I think with some practice I could accomplish an f2l quicker in some cases than cross+f2l

I understand that Roux is more complex than just solving 2 blocks and m slicing to make an f2l, and I know what I'm talking about is a hybrid probably viewed by many as an abomination.

I watched cyoubx's video on rotationless, and he briefly touches on this, saying it's just a stupid hybrid or something to that effect, but I also recall Feliks saying in a video that he uses a combination of methods depending on the situation.

I've also read some of the comments in this thread. just wondering if you all have any other thoughts on this.

 

[Silky]

I'm somewhat a late comer to considering this:

I just started looking at Roux because I really don't care for cross in cfop(yes because I'm not the best at it)

I have so much time and effort in cfop that I doubt I want to completely switch, I was just wondering if there's any merit to learning partial Roux and build the f2l from that, then finishing up with my cfop oll/pll.

I never really messed with Roux much until last night, but I think with some practice I could accomplish an f2l quicker in some cases than cross+f2l

I understand that Roux is more complex than just solving 2 blocks and m slicing to make an f2l, and I know what I'm talking about is a hybrid probably viewed by many as an abomination.

I watched cyoubx's video on rotationless, and he briefly touches on this, saying it's just a stupid hybrid or something to that effect, but I also recall Felix saying in a video that he uses a combination of methods depending on the situation.

I've also read some of the comments in this thread. just wondering if you all have any other thoughts on this.

It would just be better to learn LLOB with OCLL/PLL or COLL/EPLL. The main advantage is that you can incorporate ZBLL but if just don't do EOFDBD you can't really make use of that.

 

[efattah]

As most people refer to LLOB as ZBRoux, I have been a strong believer in this method, but only if it used with full ZBLL. Thus far to my knowledge no expert Roux solver knows full ZBLL, so we don't have real data on the method potential. Many Roux solvers have argued *against* ZBRoux, but I have rejected their arguments as unfounded.

 

[kubnintadni]

As most people refer to LLOB as ZBRoux, I have been a strong believer in this method, but only if it used with full ZBLL. Thus far to my knowledge no expert Roux solver knows full ZBLL, so we don't have real data on the method potential. Many Roux solvers have argued *against* ZBRoux, but I have rejected their arguments as unfounded.

How would you compare LLOB to LEOR? They're very similar: they just swap the order of the middle two steps.

 

[Jam88]

...but I also recall Felix saying in a video...

Felix?????????????? Feliks

 

[jdh3000]

Felix?????????????? Feliks

Yes, Feliks, sorry... I just wasn't paying attention to what I was writing..

 

[Jam88]

Yes, Feliks, sorry... I just wasn't paying attention to what I was writing..

its fine-just a bit insulting to the GOAT

 

[Silky]

How would you compare LLOB to LEOR? They're very similar: they just swap the order of the middle two steps.

I think this is actually a very interesting comparison similar to that of ZZ and Petrus.
For LEOR the main advantages are that there is less emphasis on slice moves ( higher TPS ), more can be planned in inspection ( this comes at the cost of having to solve EO in the middle of the solve if you can't plan FB + EO in inspection ), and flexibility of the first step. This however comes at the cost of efficiency. On the other hand, LLOB is more efficient and has much easier ZBLL recognition ( since you can identify corners during EODFDB and prediction of edge cycles is possible during the placement of DFDB ). I think that LLOB is a more Rouxesque approach to the method while LEOR is a more ZZesque aproach. Seems that LEOR would have a better used in OH while LLOB would be better used in 2H.