Forums → Games → Gemcraft CS: Rudimentary 128-combine.

Aaaah....at G42e, my 128c is 1.54% worse. Back to the drawing board(yay!).

Since the base stats for both at G8e(G7s) are:

64c - 80.58
128c - 82.18

and then 1.54% worse means that 64c is 1.58% better.

So I need to find a combo that is 1.58% or more better than 82.18 @G7. Fun, huh?

It depends on how psorek weights his algorithm. His keeps or improves on damage as well, but specials are pretty easy. I think a (n+n)+n+n+n does ok for example, but really hurts damage, hit rate and especially range (I think I had a grade 30 with some stupid small range I didn't care about because it's in a trap).

Also hitting u is a terrible comparison, if it weren't, you wouldn't have to worry about complicated gem building

Well how else to compare them without lengthy testing?(comparing at G42 e took me 1.5 hours to do!). It's only one U-grade, take 1.5% off the non-U'd one.

Mine has a slower growth apparently than the 64c. So, use my 128c(haven't been able to improve on it yet) only once then 64c for best result(0.5%? better and also ableto be 1 grade higher compared to a 64spec)

My main focus is managem so damage does not matter.

Hi all, sorry to hijack this thread, but all the more appropriate others were months old and I preferred not to necro them (if there is a better topic to post in please tell me).
I may also have some useful info for you.

I just got into non standard gem combining and found quite an amount of literature on these forums but I still have some questions/obsevations:

Managems (OBR):
1) What should I try to improve ?
Total leech at 0 hits, BB increment per tier (that is BB power -1) or another, more complicated, quantity?
2) Which is the minimum chain hit power I should keep?
3) All this consideration are done without looking at speed, as after a while we'll reach the cap, am I right?

Amplier gems:
4) What should I try to improve ? Leech I suppose.

All gems:
5) Is there a particular reason for using always 2^n combining? (beside being easily confronted between each other)
6) Which are the best known 2^n combines? I only found the 16 combine (psorek method), the rest seem scattered all over countless threads.

Then, after a lot of questions, something I found:

I built a progrm with the same objective psorek had, currently it can only combine orange gems (I used data from http://www.reddit.com/r/Gemcraft/comments/2385eg/sc_about_gem_combining_mechanics_and_why_we/)
and if my algorithm is correct I believe I found something interesting:

You can see all the data gathered at http://goo[ dot ]gl/aM40Os (sorry for using a proprietary file fomat, I may change it if needed), where the growth is calculated from the best combining method I could find (my algorithm is dumber than a rock, some points could be way higher).

Just to check, the method I get for 16c is exactly psorek's.

I discovered that there are much better (at least for pure orange) combining schemes that does not use 2^n gems (I apologize if all this was already found, but AG forum is rather cumbersome to navigate searching for GC2 threads).

The best fast method seems to be 13c (the difference with psorek's is rediculosly small) while for longer combines 35c seems the optimal amount.

Again, my algorithm is terribly slow, so if someone else has already run those tests and has more precise data I'd like to see them.

I posted all this mainly because I'd like to see if all this works in game: I tried to compare 13c to 16c, but I didn't gather enough mana to get them to a similar cost (1e+14 mana cost should be a good spot).
After I type this I'll try 35c and see if I can het it near 2^n gems.

Sorry for poor english.

12345ieee

@12345ieee
I'm happy that somebody else than me started to write programs If you'd like to speak with me just go to irc channel #gemcraft at freenode by eg. kiwiirc.com/client - I'm here often.

1) first of all try to combine orange gems efficiently, then managems (orange/black), then killgems (orange/black/damage).
2) don't bother with red. I just compute w/o it and place it in place of roughly random gem in formula.
3) you are absolutely right
4) exactly, when talking about mana gems. In case of killgem it's more complicated.
5) nope, I use it only becouse it's easy to compare. As you noticed, 13-combine is only tiny fraction better than 16-combine (I've computed 13-combine too)
6) I can give you 16-combine, 64-combine, 256-combine... the first one is good enough, second one is also in my thread. It's very slow to perform ingame further combines :P

If you want, I can give you the python script of mine I use to compute these formulas. Just ask me

I'm going to request again for a more defined presentation for how these work. None of the variables in the formulae are defined and technically the order is not really defined either.

Just like the other guides, the formulae could be (and have been) interpreted in different ways.

Please define your variables and present the exact order of what is done more clearly, thank you.

I was just working on a parser for the terrible order format that my progam internally uses.
I just finished, I'll update the table.

As for the better presentation, I really wouldn't know, I can define better what is what, but I'd say that my table can be considered all but a guide, just a presentation of my results for who already knows the theory.

AG forum will not let me edit my posts, so I'm forced to double-post.

After an enlightening chat with psorek I rebuilt all from scratch and did immense progress.

The interested can find an updated spreadsheet and the source code at:

https://github.com/12345ieee/gemforce

Its not necessarily a matter of presentation 12345ieee.

The forumlae presented above are mathematically and logically flawed.

The variables are undefined, and the number of parentheses are totally unbalanced in many steps. This means that there are actually many different ways you could follow these instructions because they are currently mathematically invalid.

I'm sure that there is a way they are supposed to be, but they need to be edited so that they actually work as intended.

I read psroek's introduction and as a math person, I'm sure if you check the formulae you will rapidly see the parenthetic typos that invalidate the formulae.

Hmm.... Erile .... strikes again!

To say that the instructions given here are "logically" flawed, is showing a basic misunderstanding of what logic is from a scientific point of view. The instructions are in no way meant to be proofs of or arguments for anything.

From a mathemtical point of view the intructions may initally contain errors. But:
1. You alway combine from left to right.
2. The order in which 2 gems are combined is irrelevant as the result will be the same.
This leaves no room at all for ambiguities.

The terms used in the intructions are so basic for anyone who has played a bit of gemcraft, that defining them the way you would do it for a mathematical proof is a waste of time.

So, You claim to have some knowledge of higher mathematics and show a lot of interest in gemcraft, but can't find the exact errors and provide clarification and disambiguation of the overwhelmingly simple instructions yourself?

You're a troll. Go, get a life and do something useful. Like solving the Navier-Stokes Millenium Problem.

To say that the instructions given here are "logically" flawed, is showing a basic misunderstanding of what logic is from a scientific point of view. The instructions are in no way meant to be proofs of or arguments for anything.

From a mathemtical point of view the intructions may initally contain errors. But:
1. You alway combine from left to right.
2. The order in which 2 gems are combined is irrelevant as the result will be the same.
This leaves no room at all for ambiguities.

I may have been a little rude, but I wasn't insulting, or - at least - I was not trying to be.

Those initial errors you're pointing out just happen. After a while someone notices and posts a correction. Furthermore, the instructions given here are for hardcore players. I easily got to level 3000 (MP, obviously) without using any spec- or combining-methods.
The people posting these "guides" are not getting payed for the work they invest, so they don't have to face any responsibility for the mistakes they somtimes make.

Have a nice Day, Erile

Erile, I only add parentheses where its absolutely required and sometimes miss closing them, okay?

(b+b+b) obviously is G2 B + G1B

But like (o+o)+o+b+(b+b) needs parentheses to state order of operations.

Where there's less parentheses I expect you to deduce how to add them from previous experiences...there's really nothing else that say (o+b+b) means besides (o+b)+b) anyway.