2 of 35

Basics

Only Same Weapon/Shield type can combine

One handed sword + One handed sword (YES)
One handed sword + One handed mace (NO)
Spine bow + Thicket bow (YES)
Evasion Shield + Armour Shield (YES)
Staff + Warstaff (YES)

3 of 35

Basics

Tree A + Tree B => Tree C

Item of Tree A (upper of forge) destroyed
Item of Tree B retained

Merged Tree C will be on the same item of Tree B

4 of 35

Basics

Nodes can disappear, stay or altered

“Altered” means mutated, tier upgraded, or tier downgraded

But the node positions won’t change

“Attack Speed” in Tree A at 2-1 pos can only be possible to appear at 2-1 in Tree C

Never possible to appear at 2-2

5 of 35

Basics

Allocated nodes in Tree A or B has higher chance to be selected

Its odds will be covered later in slides

6 of 35

Example

Tree A

Tree B

Tree C

7 of 35

In-Depth Combine Mechanism

8 of 35

Node Position Annotation

1-1

2-1

2-2

3-1

3-2

3-3

4-1

4-2

4-3

4-4

5-1

5-2

5-3

5-4

5-5

Layer 1

Layer 2

Layer 3

Layer 4

Layer 5

Full Tree

9 of 35

Important Concept - CPL

1-1

2-1

2-2

3-1

3-2

3-3

Connectables in Previous Layer (CPL) for a certain node

Definition: the adjacent nodes in the previous layer that can be directly connected to this node
Example:

For 3-2, its CPL is 2-1 and 2-2
For 3-3, its CPL is 2-2

10 of 35

Overview

Combine Algorithm has 3 steps below:

Node Selection - In Tree C, for each position, it rolls dice to decide if it select node from Tree A or Tree B or be empty;
Node Modification - mutate, upgrade, or downgrade selected nodes
Add Bridges - to connect the selected nodes in Tree C

11 of 35

Node Selection (Step 1)

In Tree C, from layer 1 to layer 5

For each pos in the current layer (top to bottom)

Check if any connectables in previous layer (CPL) exists in Tree C:

If no, leave this pos empty and skip
If yes, roll the dice (odds covered later) to:

select the node from Tree A
or, select the node from Tree B
or, leave it empty

12 of 35

Node Selection - Odds

For a certain pos in Tree C, assuming has at least one CPL exist in Tree C:

If both Tree A & B have node

Tree C has 93% (150/161) chance to have node here

If only one of Tree A or B has node

Tree C has 70% (92/132) chance to have node

If none of Tree A or B has node

Tree C has 0% chance to have node

13 of 35

Node Selection - Odds from Tree A or B

If Tree C decides to select a node, below is the chance whether from Tree A or Tree B

If only Tree A (or Tree B) has node

100% chance to select from this tree (trivial)

If both Tree A & B has node, both allocated or unallocated

50% chance to select from A or B (trivial)

If both have node, but only Tree A allocated it

70% (18/26) chance to select the allocated node

Small data size for this number

14 of 35

Node Modification (Step 2)

When node is selected in Tree C, it has

6.6% (16/243) chance to be mutated
2.1% (5/243) chance to be upgraded
1.6% (4/243) chance to be downgraded
In total, 10.3% chance to be altered

15 of 35

Add Bridges (Step 3)

Lastly, add bridges to all selected nodes in Tree C.
For each selected node, check how many CPLs exist:

If only one, 100% chance to connect to it
If two connectables both exist:

1/3 chance to only connect to connectable_1
1/3 chance to only connect to connectable_2
1/3 chance to connect to both

16 of 35

Example Step by Step

17 of 35

Step 1, Node Selection

Tree A

Tree B

Tree C

: allocated

CPL exist?: N/A

Odds:

100% has node
30% select from A
70% select from B

Pos: 1-1

Result: selected from B

18 of 35

Step 1, Node Selection

Tree A

Tree B

Tree C

: allocated

CPL exist?: Yes (1-1)

Odds:

93% has node
30% select from A
70% select from B

Pos: 2-1

Result: selected from B

19 of 35

Step 1, Node Selection

Tree A

Tree B

Tree C

: allocated

CPL exist?: Yes (1-1)

Odds:

93% has node
50% select from A
50% select from B

Pos: 2-2

Result: selected from A

20 of 35

Step 1, Node Selection

Tree A

Tree B

Tree C

: allocated

CPL exist?: Yes (2-1)

Odds:

70% has node
100% select from A

Pos: 3-1

Result: empty

21 of 35

Step 1, Node Selection

Tree A

Tree B

Tree C

: allocated

CPL exist?: Yes (2-1, 2-2)

Odds:

70% has node
100% select from B

Pos: 3-2

Result: selected from B

22 of 35

Step 1, Node Selection

Tree A

Tree B

Tree C

: allocated

CPL exist?: Yes (2-2)

Odds:

93% has node
50% select from A
50% select from B

Pos: 3-3

Result: selected from A

23 of 35

Step 1, Node Selection

Tree A

Tree B

Tree C

: allocated

CPL exist?: No

Odds:

Skip

Pos: 4-1

Result: empty, skip

24 of 35

Step 1, Node Selection

Tree A

Tree B

Tree C

: allocated

CPL exist?: Yes (3-2)

Odds:

70% has node
100% select from B

Pos: 4-2

Result: selected from B

25 of 35

Step 1, Node Selection

Tree A

Tree B

Tree C

: allocated

CPL exist?: Yes (3-2, 3-3)

Odds:

93% has node
50% select from A
50% select from B

Pos: 4-3

Result: empty (bad luck)

26 of 35

Step 1, Node Selection

Tree A

Tree B

Tree C

: allocated

CPL exist?: Yes (3-3)

Odds:

0% has node

Pos: 4-4

Result: empty

27 of 35

Step 1, Node Selection

Tree A

Tree B

Tree C

: allocated

CPL exist?: Yes (4-2)

Odds:

0% has node

Pos: 5-2, 5-3

Result: empty

28 of 35

Step 1, Node Selection

Tree A

Tree B

Tree C

: allocated

CPL exist?: No

Odds:

Skip

Pos: 5-1, 5-4, 5-5

Result: empty, skip

29 of 35

Step 2, Node Modification

Tree A

Tree B

Tree C

: allocated

Pos: 3-3

Result: mutated

30 of 35

Step 3, Add Bridges

Tree A

Tree B

Tree C

: allocated

CPLs: 1-1

Odds:

100% conn 1-1

Pos: 2-1

Result: connect 1-1

31 of 35

Step 3, Add Bridges

Tree A

Tree B

Tree C

: allocated

CPLs: 1-1

Odds:

100% conn 1-1

Pos: 2-2

Result: connect 1-1

32 of 35

Step 3, Add Bridges

Tree A

Tree B

Tree C

: allocated

CPLs: 2-1, 2-2

Odds:

33.3% only conn 2-1
33.3% only conn 2-2
33.3% conn both

Pos: 3-2

Result: connect 2-1

33 of 35

Step 3, Add Bridges

Tree A

Tree B

Tree C

: allocated

CPLs: 2-2

Odds:

100% conn 2-2

Pos: 3-3

Result: connect 2-2

34 of 35

Step 3, Add Bridges

Tree A

Tree B

Tree C

: allocated

CPLs: 3-2

Odds:

100% conn 3-2

Pos: 4-2

Result: connect 3-2

35 of 35

Disclaimer

Hypothesis of the combine mechanism, based on data
Data is from my 30-40 combines, not 100% reliable
Will provide raw data and statistics for your personal analysis
Not take responsibility for any gamble loss