Gem Transmutation Calculator

Plan the cheapest path to your target gems — everything runs locally in your browser.

Your Gem Stash

How many of each gem do you currently own, per color and size? (not counting equipped gems)

Equipped Gems

Your 4 currently equipped gems. They're melt-able / usable for transmutation, just like stash gems.

Resources On Hand

Aether and Godstones you already own, by rank. The calculator uses these before buying more.

Aether

Godstone

Gem Color Priority

Drag to order your colors by value — most valuable on top. Colors above the border bar are never recolored into another color (they can still be the result of a recolor). Colors below it are marked “to be recolored”: the lower a color sits, the sooner it is sacrificed when a gem needs recoloring.

Target Loadout

Pick the 4 gems (color + size) you want equipped at the end.

How this calculator works & assumptions
  • Upgrading requires 3 gems of the exact same size and the exact same color, plus 1 Aether whose rank matches the source size (3× size 1 + Aether rank 1 → 1× size 2; 3× size 2 + rank 2 → size 3; 3× size 3 + rank 3 → size 4). You cannot mix colors or sizes, and the resulting gem keeps that same color.
  • If a gem's color doesn't match your target, a recolor (1 Aether rank 3) is paid immediately when the gem is picked up.
  • Recoloring is a flat cost (1 Aether rank 3) regardless of gem size. That makes wrong-colored stockpiles valuable: the solver may upgrade a stack in its own color (e.g. 9 spare Diamonds into one size-3 Diamond) and pay a single recolor on the finished gem instead of recoloring every input.
  • Gem color priority — drag the colors into a value order, most valuable on top. Colors above the border bar are never used as recolor material — they can only be the result of a recolor, never the source. Colors below the border may be sacrificed, least valuable (bottom of the list) first; ties between equal positions go to the color you own most of. Dragging the border to the very bottom disables recoloring entirely.
  • Creating fresh size-1 gems from a batch recipe (e.g. rank 3 → 3 gems) produces bonus gems when you only need one — those extras are kept in your pool for reuse on other targets, they are never wasted.
  • Owned Aether/Godstone are not treated as free: the plan is chosen to minimize the total vendor value of everything it consumes — owned stock counts at its replacement price, purchases at their MoF cost — with real MoF spent breaking ties. So the solver won't burn a pile of your rank-3 resources just to avoid buying a couple of cheap rank-1 ones.
  • Exact color+size matches already sitting in your stash or equipped slots are used first, for free, before any crafting is considered.
  • Random gem generation (the collapsible Gem Gambling panel in the results) — the vendor can mint one size-1 gem of a uniformly random color for 5 MoF, half the 10 MoF a deterministic craft costs, but you don't pick the color. Opening the panel adds a best / expected / worst-case estimate for gambling on the fresh size-1 gems the plan would otherwise craft. Everything is priced at vendor value, never as free: a crafted gem is a Godstone + Aether pair worth 10 MoF and a recolor is a rank-3 Aether worth 15 MoF, whether they come from the shop or out of your own stock — owning a material doesn't make it free, it makes it worth its replacement price. (The stock-subsidised cash total is shown only as a small secondary reference.) A pull only helps if it lands on a color the plan is still short of, or on one your color priority allows as recolor material — a color protected by the priority border is dead weight, since you've told the tool not to melt those down. So with k colors short and d recolorable, only (k+d) of the 6 outcomes are worth anything, and gambling costs 5 × 6 ÷ k MoF per gem you need (or (30 + d × 15) ÷ (k+d) if recoloring the donors is cheaper) — which beats a 10 MoF craft only when several colors are short at once. The panel shows the odds colour by colour, then either the smart mixed play (gamble while it's ahead, craft the tail) or, when gambling loses, what gambling anyway would really cost. Worst case is the 95th percentile — bad luck you'd stay under 19 times out of 20, since a truly unbounded run of misses isn't useful to quote. Everything above size 1 is held at its deterministic cost; only the fresh size-1 gems are gambled on.
  • This is a fan-made tool based on the rules as described; it is not affiliated with the developers of Fellowship.

Save or share this setup

Copy your inputs as JSON to move them to another device — or to hand them over when a recommended path looks wrong. Nothing is uploaded; the text goes to your clipboard.

The inputs sit under solverInput in the same shape tests/scenarios.js uses, so a pasted configuration can go straight into the solver test suite.