First of all, this guide is beneficial only for 500+ level players with all skills unlocked and at least +20 to all skills talisman fragments. Note that talisman runes are not important at this level. Talisman fragments with good properties are to be prioritized, with +1 to all skills being the most important property.

If you are lower level, or don’t have a good talisman, take a look at <LOW LEVEL GUIDE THAT DOES NOT YET EXIST> or <TALIFARM GUIDE THAT DOES NOT YET EXIST>.

The guide targets version 1.2.1+ of the game (it was made during v1.1.2b), the bulk of it would be valid for 1.0.x as well, but some things changed.

To know more about the game before following this guide there are lots of other information sources spread on the internet.
The guide will mention many resources where they are relevant, here are some more:

• Stash rewards and other stuff[docs.google.com]
  
• Steam Guide on how talisman runes work
  
• GemCraft Wiki[gemcraft.wikia.com]
  
• Compendium of game formulae[gist.github.com]

We also mention the GC Discord server[discord.gg], a place for general discussion of the GC games, in particular the pins in the #1e308-iq-gc-talk channel, which hosts the latest theorycrafting developments.

In this guide, several abbreviations are used:

• g: stands for gem grade, typically used as g1, g28, g45 and so on.
  
• y yellow, 2y refers to that same gem upgraded once (1x U)
  
• o orange, 3o refers to that same gem upgraded twice (2x U)
  
• r red, you can guess what 4r would mean...
  
• b blue
  
• So g45rb stands for a dual red/blue gem of grade 45
  
• Spec, speccing, specced or just 16s, 32s and so on stands for a gem built by a special recipe from a certain number of g1.
  
• Combining, combined or just 16c, 32c and so on stands for a gem combined by a special recipe, where the resulting gem has a lower grade than just upgrading the gem with ‘U’, but better stats.

Playing GC:FW efficiently means efficiently earning XP. To get the most out of our playtime, there are some things to consider.

XP sources
Once you are properly enraging, the only XP source that matters is the XP given from monsters, since enraging only affects monsters and thus only increases the amount of XP which monsters give.
Flying ones, spell bonuses, killchain, beacons and all the other tiny contributions all pale in comparison.
So our aim is to enrage monsters as much as possible and to kill the most amount of them in the least time.

The enraging socket
This game features a socket for enraging above the wavestones, in contrast with gem bomb enraging in all earlier games. To enrage a wave, place a gem in this socket and leave it there until you no longer wish to enrage monsters. Removing or replacing the gem in the enraging socket will remove any rage on the next wave, but upgrading it with ‘U’ will leave the rage modifiers intact.

The effect of an enraging gem depends only on the gem’s grade and increases the number of monsters (additive +5% per grade, up to +100% at grade 20 and beyond), monster health (multiplicative +80% per grade), monster armor (multiplicative +50% per grade) and monster xp (multiplicative +20% per grade).

You will need at least a +20 to all skills bonus from the talisman to really follow along.

The correct skills distribution is a complex topic, we refer to the skills guide:
https://gtm.steamproxy.vip/sharedfiles/filedetails/?id=1999407961

At this level, talisman runes are not important. It is obvious that the vertical bonus of extra mana per wave is useless since it’s a flat value and is worthless when you’re able to leech so much more, but the horizontal bonus of extra initial mana isn’t important either since starting mana has very little effect on your xp gain. Talisman fragments with good properties should be prioritised over rune bonuses. Talisman fragment rarity is also far less important than whether a talisman fragment has good properties. For example, a sub-rarity 100 fragment with good properties is far more useful than using a rarity 100 fragment with bad properties just because it's rarity 100.

Useful talisman properties in order of importance

If a talisman property isn’t in this list, it’s either because it’s useless (most of the other properties) or because it doesn’t do enough to care about getting as much of it as possible (such as damage bonuses). A full list of talisman properties can be found at https://gemcraft.fandom.com/wiki/Talisman_(GCFW).

By far the most important talisman property is +to all skills. If a talisman fragment doesn’t have +to all skills, it is a waste of shadow cores to upgrade and should be removed as soon as a replacement which has +to all skills has been found. There is absolutely no reason to use a talisman fragment which doesn’t have +to all skills over one which does.

There are some in-game mods that help a bit as well:

Wealthy monsters
Gives monsters a chance to drop shadow cores when they die, in addition to any other drop which they may have been given when the wave was created. Can’t really hurt to use but might marginally increase lag when lots of monsters are killed at once.

Spell preparation
All spells start charged at 200% at the beginning of the battle. Very useful to place barrage immediately on the first two mana gems.

Wasp filled shells
Exploded barrage shells generate gem wasps at the impact point. Makes tower barrage slightly stronger.

Deep learning gems
Decreases the number of hits gems need for the first few hit levels. After the first hit level which requires more than 1000 total hits, the hit count requirement starts increasing at the same rate regardless of whether this is enabled though. Overall, for a given number of hits (which is at least 1000), using this mod will result in around 4 more hit levels.

For more details on the topic, see the Steam Guide on in-game mods: https://gtm.steamproxy.vip/sharedfiles/filedetails/?id=1984376311

To ensure the game runs smoothly during endurance, it’s necessary to put all the graphics setting to the minimum possible, you need to:

• Toggle “Force minimum visual settings” on
  
• Toggle "Render monster bodies only", in the “Hardcore wizard options” (needs to have completed the story for it to appear)
  
• Set the “Snow/rain opacity” slider to less than 42% (basically just turn it down to the minimum), this will disable weather altogether
  
• Turn "Spill blood on the ground" off

You can check that the game is running smoothly by turning on either the game or Steam’s FPS counter and checking it stays at 30.

Unlike in Chasing Shadows, all fields have the same hit points increment per wave in endurance (10%) and beacon farming is no longer a worthwhile source of xp (so how much space there is for beacons no longer matters) so there isn’t as much difference between the fields for xp gain.
The only differences between fields are:

• layouts: some are better for mana farming than others, some lack space for amplifiers (such as most of the fields in the P map tile) and some don’t have the monsters converge to a single path before reaching the orb.
  
• first wave hp: the more health the monsters start with at wave 1, the earlier the wave where the monsters become too strong. Higher first wave hp reduces xp a small amount due to having a bit less time to mana farm.
  
• wave composition: swarmling waves give the most total xp, followed by giant waves and then by reaver waves.
  
• armor increments per wave: higher is better since having more armor will make monsters give more xp, the best fields give around 15% more exp than the worst before taking other factors into account.

Here is a map of the armor increments of the game fields (green = more armor = more xp from armor):


Green - 1.105 Yellow - 1.1 Orange - 1.095 Red - 1.09

Overcrowd: YES
More monsters means more xp and makes it easier to have monsters on the mana traps, should always be used when you want xp.

Adaptive Carapace: YES
Worth using at all levels since it does next to nothing if you intend on one-shotting all the monsters and even at worst, a 60% damage reduction is only a 150% increase to health and the increase to traits xp multiplier more than makes up for this.

Strength in Numbers: YES
The amount of armor this trait gives is negligible at high levels and it means nothing at low levels too if you use traps.

Dark Masonry: YES
Doesn’t really do much since beacons are much weaker than the monsters you’re killing if you’re enraging properly and you’re going to have to deal with beacons (due to beacon summoning giants) regardless of whether this is used. You can use walls to block off all spaces where beacons could spawn to stop them spawning entirely too.

Ritual: YES
Doesn’t really do much since flying ones are much weaker than the monsters you’re killing if you’re enraging properly and you’re going to come across flying ones regardless of whether this is used. Note that this trait can spawn wizard hunters when it is set to 8 or more so don't set it above 7 if you don't want that.

Giant Domination: YES
If you are able to spam ice shards, you will be able to hit giants many times with ice shards before they get to your kill gems so they become easy to kill and how much health they start with is irrelevant. Giant waves give more xp than reaver waves so when this trait replaces reaver waves with giant waves, it increases your base xp gained too. If there are no reaver waves left to replace, it ends up replacing swarmling waves though which reduces base xp gained since swarmling waves give the most xp but overall it ends up always being better for xp after taking the increase in traits xp multiplier into account. If you're not able to spam ice shards then using this trait will cause giant waves to become too strong for you to kill earlier but this won't reduce xp gain unless you downgrade the enrage as soon as the giants become too strong.

Insulation: YES AT WL > 1K
Markedly improves the instakill phase, as it gives 12 extra mana hits to every monster. Mana traps work through shields anyway. Lanterns are great at removing shields, so even once you’ve stopped mana farming, they won’t cause problems unless you call extreme numbers of waves early.

Corrupted Banishment: YES AT WL > 1K
Once you’re strong enough to avoid leaks, this will have no impact. If struggling, turn this off.

Thick Air: START AT 0, INCREASE AS YOU IMPROVE, USE 12 AT WL > 20K
You'll be hitting monsters more than 24 times each, so this is negligible after you are able to reach the fire rate cap for traps (180 shots per second) by an early point in the battle. You should set the trait below max level if you struggle early in a run. Don’t call too many waves at once when using this trait or your kill traps will get overloaded.

Hatred: YES AT WL > 10K
This trait will decrease the raw xp you gain but increase your battle traits xp multiplier. Hatred decreases raw xp gain for two reasons: monster armor and mana farming time.
Because hatred increases monster health but doesn’t increase monster armor, monsters with a given amount of health will have less armor with hatred than without and will thus give less base xp with hatred than without. Hatred also causes you to have fewer waves before needing to downgrade the enrage, lowering potential mana farming time. Hatred will reduce your xp per hour slightly until around WL 30k but will make battles take less time so it may make levelling up a bit faster overall due. For example, doing 2 battles with half xp potential in half the time it would take you to do 1 battle with full xp potential, you will gain more xp and levels since you're stronger in the second battle so you get more xp than you'd get from a single full xp potential battle.

Swarmling Domination: YES AT WL < 1K, ONLY AS HIGH AS YOU CAN STILL EFFECTIVELY KILL SWARMLINGS IN WHITEOUT AT WL > 1K
If you don’t care about killing things in whiteout (such as if you’re too low of a level to be able to spam it), this is great since swarmling waves give the most total xp so you’ll get a lot more xp. But if you actually want to kill as many monsters as possible in whiteout as you can, putting this too high will lose you xp due to the reduced debuff duration making whiteout not last long enough. If you are going to use this and you’re not a low level, you should have all strike spells capped and it is recommended to not put it higher than 6 or so.

Haste: NO
Haste leads to a lower amount of xp per battle due to making waves come faster, thus reducing the time before a given wave comes and thus leading to less mana farmed by a given wave and a lower enrage grade, causing you to get less xp. While haste could theoretically increase your xp/time at the cost of a lower amount of xp per battle by making battles shorter, it has several problems which make that unlikely:

• Waves coming faster leads to monsters entering at a faster rate, increasing the number of monsters on the field and increasing lag, reducing the rate at which mana is leeched in real time.
  
• Haste won't increase the speed of the battle when you're calling waves early (such as after finishing mana farming) or when paused.
  
• The more significant the previous xp record is on a field compared to what you will get next time is, the more worthwhile it is to maximise xp per run.

So in order for haste to increase xp/time, you need a combination of having the increase in lag from using haste to be low enough, the amount of time spent mana farming to be significant enough compared to the time spent calling waves early or paused and the previous xp record to be low enough so its effect on xp gained is sufficiently low.

Swarmling Parasites: NO, NEVER
The xp from them does not scale so all you really get from this trait is an increased traits xp multiplier. Extremely annoying when combined with thick air since you’ll need to deal even more hits per monster. High health giants will cause big problems since you need to reduce their health with ice shards in order to kill them and then once you kill them, 2 full health spawnlings come out which you’d then need to whittle down with ice shards again.

Vital Link: NO, NEVER
Increases monster health significantly when bringing enraged crowds of monsters (swarmling waves can have over 200 monsters in them when enraged with g20+ and overcrowd on 12) and doesn’t even increase their base xp. Furthermore, vital link makes estimating how much health monsters can have before you need to downgrade the enrage gem much more difficult and it stops you from calling waves early so much which slows down the end of endurance.

Awakening: NO, NEVER
This trait will decrease the raw xp you gain by so much that the increased traits xp multiplier doesn’t even come close to making up for it. Awakening decreases raw xp gain for three reasons: health scaling rate, monster armor and mana farming time. By making the monster health increase at a faster rate, you will get less total xp from killing monsters up to a given amount of health than you would when not using awakening. Because awakening increases monster health but doesn’t increase monster armor, monsters with a given amount of health will have less armor with awakening than without and will thus give less base xp with awakening than without. Awakening also causes you to have fewer waves before needing to downgrade the enrage, lowering potential mana farming time.

In this game, pure and dual gems are used exclusively, simplifying the complex gem-building techniques seen in previous games.

Upgrading these gems using ‘U’ is nice at the beginning of the game, but as soon as you want a little more, more mana gain, more gem grade, more waves beaten, more xp-outcome and therefore more wizard levels to get even more of all this, it is not the best strategy.

To get the best out of your gems, you have to combine multiple copies of them together (not just 2 like U does) with precise recipes, called combine recipes.
For dual gems it also matters how the initial dual is built from the base g1s, recipes to improve the initial gem are called specs. The only dual we use in this game is slow/bleed, so we’ll concentrate mainly on combines.

A combine recipe is a series of duplication & combination steps to perform on a given gem, that result in a gem that is N times costlier and P times more of the stat of interest.
The simplest “recipe” is U, also called (g+g) or 2g, that simply duplicates a gem, then combines the original with the duplicate. The resulting gem is 2x as costly as the original and has 1.38x the leech (and 1.58x the max dmg, 1.38x the crit mult).

Let’s now consider a slightly more complex one, the leech 4c: (((g+g)+g)+g):
The recipe says to dupe the base gem 4 times and then combine 2 together, then add the 3rd, then the 4th.
The resulting gem has 1.92x the leech, 2.47x the max damage and 1.89x the crit mult.
This is in contrast to using U twice, which is ((g+g)+(g+g)), still costs 4x, but has 1.90x the leech, 2.50x the max damage and 1.90x the crit mult.

This means that the leech 4c is a superior combine than U for leech, because it results in a gem of the same cost, but better leeching power.
If we were to apply it over and over on the resulting gem (like you press U over and over a gem to increase the grade) it’d become relatively stronger and stronger.

Conversely, the leech 4c is a worse combine than U for crit. This comes as no surprise, it is a combine optimized for leech. There are crit combines already at length 5 that are better than U.

There are countless ways to make a spec/combine recipe of length N, and while for small sizes it’s possible to enumerate and test them manually, we have enlisted the help of computers, to find the best ones by computing quadrillions (not an exaggeration) of recipes and comparing them.
The program that does the calculations is gemforce[github.com] and the results can be found at the gem-recipes repo[github.com].
This link goes to the readme, read it, then scroll up for the recipes.
Generally, longer recipes are more powerful than short ones, and non-power of 2 lengths are better than power of 2. The readme has a detailed explanation of the metrics.

Because performing these combinations manually is extremely tedious, a mod called gemsmith has been developed to do them automatically (see the Modding section). Below, we will show the two basic leech and crit recipes that are short enough to be doable by hand, showing what gemsmith does behind the scenes.

Process from left to right. The “o” represents the gem you’re using this combine on (and its duplicates), so if you were to only use 13c on a gem for the whole run, you’d have:

• 1st round: “o” is a regular orange g1, result is a gem of grade 4 that costs 13 g1 and 12 combinations, call it A
  
• 2nd round: “o” represents what we called A, the result is a gem of grade 7 that costs ~13x the previous, call it B
  
• 3rd round: “o” represents what we called B, the result is a gem of grade 10 that costs ~13x the previous, call it C
  
• 4th round: “o” represents what we called C, you get it…

For simplicity in the following I’ll talk as if the “o” represented a g1, “2o” a g2, and “3o” a g3.
Remember that “o” may actually represent whatever base gem we’re running this on and “2o” and “3o” its upgraded counterparts.

We have 2 g1 in the 1st subgem, then 5 single ones, then 2, then 4, for a total of 13, as expected.
Here is a simple construction sequence:

• Duplicate the base gem 8 times (1 per subgem)
  
• Use U on the first gem, becomes a g2
  
• Use U on the 7th gem, becomes a g2
  
• Use U twice on the 8th gem, becomes a g3
  
• Now all the subgems are correct
  
• Combine one by one the g1s into the first g2
  
• Now you have a fat g2, a normal g2 and a g3
  
• Combine both g2 together
  
• Combine both remaining g3 together

Like before, the “y” represents the base yellow gem we’re running this on, and “2y” and “3y” its upgraded counterparts:

• Duplicate the base gem 6 times (1 per subgem)
  
• Use U on the first gem, becomes a g2
  
• Use U on the 5th gem, becomes a g2
  
• Use U twice on the 6th gem, becomes a g3
  
• Now all the subgems are correct
  
• Combine one by one the g1s into the first g2
  
• Now you have a fat g2, a normal g2 and a g3
  
• Combine both g2 together
  
• Combine both remaining g3 together

Non-standard combines sacrifice fire rate for specials/damage, because fire rate is capped anyway.
Therefore they aren’t worth it before you can easily reach the fire rate cap with amped gems in traps (around WL 5k) and even then aren’t really needed until levelling progress starts to get too slow for your liking.

The most important combines to use are the ones on the managems, as they have a much bigger effect on xp compared to the crit/bleed ones.
The best strategy is to start with the biggest and most efficient combines (the 6M mana combine at the time of writing) and then gradually use smaller combines as it gets more difficult to accumulate mana, all the way till U is the only reasonable one.
Once enough mana has been accumulated, it’s possible to restart a gem using only the best combine and straight up replace the current gems that have accumulated weak combines. While selling and rebuilding a gem uses extra mana, it is well worth it since it increases your mana gain substantially.
Whatever combine you run on a trap/lantern gem, you should also run on all its amps, to keep max efficiency and preventing amps desync.

The essential set of combines therefore is:

• the best one available for the color: 6M leech, 180k crit, 1.4M bleed at the time of writing
  
• a bridge combine with length around sqrt(length of best combine): 2027 leech, 230 crit, U for bleed
  
• optional: 188c leech, 13c leech and 11c crit to bridge even smaller gaps, might be useful depending on playstyle
  
• good old U

Now for the actual gameplay. First, you have to set up your map. You need a mana farm and a kill area.

Here you have an example of a battlefield-setup:



In this picture, you see the mana farm (“Z” shape), the kill area (“U” shape), and the maximum range of lanterns and spells. Although you don’t need amplifiers in the early stages of an endurance run, you still want to place your structures with them in mind.

The mana farm consists of a path of several (3 or more) pure orange traps with amplifiers. The gems in the traps should be enhanced with barrage to increase the amount of mana they leech, beam is useless in the endgame due to traps being limited to 180 shots per second. It is not worth using more mana traps than you can effectively whiteout, since traps outside will give less mana and monsters are less likely to get hit by whiteout and still be in whiteout when they die, giving less xp.
Because of this, it is a good idea to limit the size of your mana farm based on your strike spell range, though you will want to use a smaller farm than the largest you can for your strike spell range if you are a low level. The best shape for a mana farm is one with a bend (or two) since it means that the amplifiers will support more traps on average (increasing mana leeching potential) and monsters will take a longer path through your strike spell range than if they just went in a straight line through it (making fast monsters more likely to get hit by whiteout before they leave).

The final kill area consists of a path of several (3 or more) pure yellow traps with amplifiers. The gems in the traps should be enhanced with bolt to increase the damage they deal, bolt is better than barrage as bolt gives +100% damage while barrage gives +30% specials which results in only +30% damage. This should be placed just past the mana farm so that monsters which are whited out in the mana farm get killed before the whiteout expires. Ideally, there should be a small gap between the mana farm and first kill trap so that re-socketing giants being killed by the front kill trap won’t cause any part of the mana farm to be re-socketed. Because of this, there should not be any structure with anything other than a crit gem in it within 2 tiles of the front kill trap. Also because of re-socketing giants, you absolutely need more than 1 kill trap. Kill areas should be a bit larger for players using thick air (5 traps in a straight line or 7 traps going around a corner) than otherwise (3 or 4 traps).

Slow will be used once the monsters stop dying to the mana traps in order to make the monsters take longer to get through the mana farm. Use a lantern set to Least Affected by Specials.
Ideally, this lantern would have the entirety of the mana farm in range, but if your slow duration is long enough, it is enough that monsters will pass by it before entering the mana farm. This lantern should not share amplifiers with the mana farm, but it should use at least one amplifier of its own. The maximum range of an amplified lantern is double that of one with no amplifiers.

Bleed will be used later in the battle in order to make monsters take more damage and thus allow stronger monsters to be killed, allowing for more xp gain, and will be inflicted by using a lantern set to Least Affected by Specials.
Ideally, this lantern would have the entirety of the kill area in range, but if your bleed duration is long enough, it is enough that monsters pass by it before entering the kill area (unless you’re using swarmling parasites for some horrible reason). This lantern should not share amplifiers with the kill traps; it should have as many dedicated bleed amplifiers as possible since only bleed amplifiers can make the bleed effect stronger. If not using bleed/slow dual gems as detailed in their respective section below, you will need a second place for your lantern/amplifiers.

Lastly, a yellow gem in a tower which is weaker than the gems used in the kill area can be used to kill flying ones (and beacons if you don’t block them). This should not share amplifiers with the kill area, as that would give it so much damage that it risks killing monsters early. Amplifiers should be used to give it enough range to cover the map, though excessive amplification will increase firing speed such that it lags the game. This can be built later into the battle as needed. Position matters little since tower range can span the whole map.

Traps prioritize monsters by sorting a list of monsters in range of each trap, and then firing at monsters one at a time in order. This means that if the trap fires more shots in a frame than there are monsters on the trap, all monsters will be hit and priority is not important. If you avoid crowding the map with hundreds of monsters on each trap, setting random priority is best because it skips the sorting of the list (so it’s not truly random), reducing lag.

• you get more mana when you have at least 2 monsters on every mana gem at once
  
• you get more xp when you have whited out your monsters before they reach the kill area. Talisman bonuses can increase the xp for enemies kill while whited out substantially.

At the start of the battle, the monsters will die very quickly to mana traps (in 1 hit if you don’t have insulation or thick air on).



Since mana gain at the start of the battle causes you to need to upgrade gems very frequently, it can be easiest to invest most mana into a single gem which is amplifying as many traps as possible and using bolt and beam to re-socket it when you upgrade it (barrage should ideally be reserved for the traps).

At this point in the battle, slow should not be used since the slow gem will hurt monsters and thus reduce the number of hits your mana traps can get on them. Bleed also kills monsters faster, so save that for later.

As you get more mana, you should upgrade your mana/enrage gems.



Once the rate at which gems need upgrading slows down, the first traps and amplifiers should be prioritized since they get the most hits on the monsters.

Eventually, the monsters will stop dying so much to the mana traps and you can set up a proper mana farm by filling any empty mana traps and amplifiers.



At this point, you should start using a slow lantern set to least affected by specials to make the monsters take longer to get through the mana farm, making the traps have monsters on them more often and making monsters easier to hit with whiteout. If affordable, use a grade 23 slowing gem because that is enough to reach the slow cap of 70%. The slow lantern should not share amplifiers with the mana traps because it will give it too much damage and the slow lantern doesn’t necessarily need 8 amplifiers since slow power, lantern range and lantern fire rate have limits. Bleed should still not be used or at least shouldn’t be used in a way which would lead to bleeding monsters dying in the mana farm.

The mana farm should have whiteout cast on it whenever possible to leech more mana from the monsters per hit, to further slow them down, and to increase the xp gain from killing the monsters.
Freeze should be used to keep monsters on the mana traps, but only when strictly needed (there are no extra monsters approaching a mana trap, so not freezing would leave a trap unoccupied).
Ice shards are not required while monsters are weak. However, if you have thick air on, it can be worth using ice shards as soon as monsters are strong enough to survive the mana farm so that the kill traps need fewer hits to kill the monsters.

It should be noted that the number of monsters on a given trap doesn’t make much difference since there is no chain hit, so don’t constantly cast freeze on the mana farm and end up with a 4 digit monster count. A trap can only hit a given monster 3 * game speed times per frame (that is, 3 times on 1x speed and 9 times on 3x speed) whereas a trap can fire up to 6 * game speed times per frame (that is, 6 times on 1x speed and 18 times on 3x speed) and this is reached when a trap has a fire rate of at least 180 shots per second. Because of this, a trap only needs 2 monsters on it in order to reach its full potential.

It is ideal to not spend too much mana on the enrage gem early on since the monsters don’t have much base xp and focusing more mana on the mana farm will allow you to use a better enrage later on. It is worth upgrading the enrage gem if monsters start dying to the mana farm again.

In this stage, the kill gems don’t need to have significant amounts of mana spent on them and just need to be strong enough to kill the monsters.

After some time of mana farming, it will be difficult to further upgrade you mana gems. Sell the mana farm and use the mana to upgrade the enrage gem and kill gems.

There is no point in mana farming for too long, it seems that it is best to stop mana farming after you are at a wave which is just after half of the transition wave, that is the wave at which you’d be forced to downgrade the enrage because even maximally upgraded killgems can’t keep up anymore.
If you remember when this happened in the previous endurance run, you can use that as an estimate for the transition wave. Otherwise, the skills table in the skills guide has an estimate for your level.
Mana farming for longer than this doesn’t give more total XP since the enrage gem gets upgraded too late for it to be any better than just selling the mana farm. Furthermore, it uses more time than is needed.



Before selling the mana farm, the enrage and kill gems are not at their maximum grade.



After selling the farm, most of the mana has been spent on the enrage gem. The remaining mana went into the bleed/kill zone.

Ideally, you will want to have the monsters be slowed and pass through a bend before they reach the kill traps so that you can hit them with whiteout more reliably before they get killed (the longer the path they take through your spell range, the more likely they will get hit by it before leaving). If your mana farm was a suitable shape, you can use the old mana farm for this purpose. Otherwise, you can consider moving the kill traps to a more suitable location and creating a maze with walls before the traps.

If you haven’t already, it is time to start using bleed to improve your killing power. Like slow, the best way to inflict bleed is to use a lantern set to least affected by specials. Unlike your slowing lantern, it is important to use as many amplifiers for the bleed lantern as possible since they’ll make the bleeds inflicted stronger and thus allow you to kill stronger monsters. You should spend about 20% of the mana you spend on crit gems on bleed gems.

Once you have spent the mana gained from salvaging the mana farm on upgrading the enrage, crit and bleed gems (with most mana being spent on the enrage gem), it is a good idea to estimate the amount of health which you can deal with so you have an idea of when the enrage will start needing to be downgraded.

Like in the main mana farming phase, you should cast whiteout on the area before the kill traps whenever possible to further slow the monsters down and to get more xp.
Freeze should be used on the kill traps whenever monsters are getting too far through the kill area. Note that freeze’s increased critical damage against frozen monsters does effectively nothing when you have a good critical damage multiplier since it only adds onto it.
Like whiteout, ice shards should be cast whenever possible to weaken monsters. Casting in the exact same spot as whiteout is convenient and effective.

At this point, it is time to start calling waves early to get the battle over with faster. Don’t call too many at once or your kill traps will get overloaded when using thick air (the main problem is swarmlings and to a lesser extent, reavers), your lanterns will not be able to remove shields fast enough to effectively inflict everything with slow and bleed when using insulation, and you’ll start lagging from having too many monsters on the field.

Eventually, the monsters will get too strong to kill, and you will need to downgrade the enrage gem in order to continue killing monsters. In this phase, you just continue doing what you were doing before, but you have to downgrade the enrage gem as you continue.
The first time you replace the enrage gem you’ll gain a sizable amount of mana, use it to further improve bleed/kill gems, granting some extra power.



If you have ice shards at a high level and a good amount of hit points taken by ice shards (multiple talisman fragments with ice shards takes +x% hit points, the default amount of 10% is absolutely terrible), you can ignore the amount of health giants have since you can hit them with ice shards so many times before they get to being killed that they cause no problems. Instead, pay attention to the health of reavers and swarmlings, particularly in waves with health or speed buffs.



Because the xp from monsters decreases exponentially as you downgrade the enrage gem, it is not worth it to continue all the way down to killing monsters with no enrage gem.

The plot below shows the exp gain per wave and the enrage grade around WL 30k:



The blue curve shows the relative amount of experience gained each wave, so total experience gained can be visualized as the area under the blue curve. Note that most of the experience is gained before the XP peak (wave 320 in this example), and like 99% of it is gained by the time you downgrade the enrage gem 20 times.
At that point, continuing to degrade is just a waste of time, it’s better to remove the enraging gem and just finish the map by cleaning up remaining enraged monsters and then ending the endurance.

As you get stronger the XP fraction gained before the transition wave raises even more, making long degrading periods even more useless.

If you have thick air on, it can be a good idea to use a low grade enrage while waiting for the last monsters which were properly enraged to get killed instead of completely removing the enrage gem, due to monsters which heal nearby monsters on death.
Because you can only deal damage up to a given percentage of a monster’s max health each hit, monsters which heal nearby monsters on death can make killing even weak monsters take a long time due to them healing faster than you can damage them. However, how much they heal on death isn’t affected by enraging, so by enraging, you make the health percentage which they heal decrease thus making them not heal monsters faster than you can damage them.

The community has developed a modding framework for GC:FW, this is the first game in the series to have one.
You can find it here: https://github.com/gemforce-team/BezelModLoader , its authors update it for every new game version.
By modding the game, a lot of things are possible, we mention below some pure QoL mods under the Gemforce Team group:

• Gemsmith[github.com]: Mod to perform the non-standard combines we talked about earlier as easily as you click U, with a full in-game UI
  
• ManaMason[github.com]: Building blueprinting mod, useful especially for talisman farming
  
• Autocast[github.com]: Automatically casts strike spells on a given spot when ready, useful to cast whiteout and ice shard on all monsters

Since least affected by specials just targets monsters with the least number of gem special effects (poison, bleed and slow) on them, it has a weakness when dealing with multiple specials. If you have some monsters which are slowed and some monsters which are bleeding, a bleed lantern set to least affected by specials will view monsters which are only bleeding and monsters which are only slowed as being the same since they have a total of 1 special effect on them and thus it might just end up hitting monsters which are already bleeding. The best way to solve this is to use a dual bleed/slow lantern instead of a pure bleed lantern and to still use pure bleed amplifiers (that way the bleed power loss is reduced and the loss gets smaller the higher your amplifiers skill gets).

If you use a dual bleed/slow gem, you can get away with using only 1 lantern (and thus reducing lag a bit). To do so, the lantern needs to cover:

• some space before the area you’re casting whiteout (and ice shards) on (so the monsters are slowed in that area and easier to hit with whiteout before they leave)
  
• the whole kill area (technically not needed but if it doesn’t and a monster gets onto the kill area without getting hit, you’re going to have problems)

You’d start by using a slow gem in the lantern after the monsters get strong enough to not die to the mana farm and only change it to a bleed/slow lantern with pure bleed amplifiers once either the monsters are strong enough to need it or you’ve finished mana farming.

You could create a dual bleed/slow gem by simply combining a g1r with a g1b and upgrading from there but by making a dual gem you not only lose 30% bleed power due to impurity but you also lose some more bleed power due to having combined the bleed gem with a slow gem, which reduces the raw component power and this can be mostly prevented by using a better spec to increase the bleed component power at the cost of some of the slow component power. The recommended spec for a dual bleed/slow gem is the 32 spec.

32 spec for dual bleed/slow gems

How much the gems in amplifiers should cost compared to the gems they support depends on your amplifiers skill level, how many gems are being supported, the way in which the amplifier gems differ to the gems which they’re supporting (only matters when using dual bleed/slow gems with pure bleed amplifiers) and the manner in which the gems are being upgraded.
This section assumes that simple u upgrades are used to get the amplifiers into the correct ratio.

In order to find your spread values, you can hover over a gem in an amplifier.
Amplifiers have non-zero base spread values therefore the values on the amplifiers skill aren't your spreads.




What the grade of a mana leech gem in an amplifier should be relative to the gems it supports depends on the specials spread your amplifiers have and how many things it is supporting. In order to find what grade the amplifier gem should be relative to the gems it supports, you add the delta grade for your specials spread from the first table to the delta grade for the number of things supported from the second table. If you are able to remember the second table then all you need to do in order to put your mana amplifier gems at the right grades is to look up the delta grade for your specials spread.




What the grade of a critical hit gem in an amplifier should be relative to the gems it supports depends on the damage spread and specials spread your amplifiers have and how many things it is supporting. Note that while the first table only has damage spread in it, specials spread has also been taken into account for the delta grades (this is possible because the specials spread is always exactly 5% lower than the damage spread).
In order to find what grade the amplifier gem should be relative to the gems it supports, you add the delta grade for your damage spread from the first table to the delta grade for the number of things supported from the second table. If you are able to remember the second table then all you need to do in order to put your crit amplifier gems at the right grades is to look up the delta grade for your damage spread.



What the grade of a bleed gem in an amplifier should be relative to the gem it supports depends on the specials spread your amplifiers have and what spec the gem being supported was formed from. Find your specials spread in the column for the spec your bleed lantern gem uses (left is pure bleed, middle is a 32 specced bleed/slow and right is a 2 specced bleed/slow) and use the delta grade from that row.

• The lanterns skill is capped, whereas the traps skill isn’t. This means that the higher your level is, the better mana traps get compared to mana lanterns.
  
• Traps leech much more mana per damage dealt than lanterns since traps have a specials bonus and a bigger damage penalty than lanterns. This means that traps can mana farm better in the early waves.
  
• Lanterns cause much more lag than traps, so much lag that it isn’t even worth using cheap mana gems in lanterns adjacent to the amplifiers supporting the mana traps since the rate at which mana is leeched in real time goes down because the extra mana leeched isn’t even near enough to offset the lag.

• Towers ignore bleed when estimating damage against monsters and will thus massively overshoot bleeding monsters which would take many hits to kill without bleed but after taking bleed into account will die in a few hits whereas traps deal damage instantly and do not have this problem because they don’t need to estimate damage at all.
  
• The traps skill is uncapped, so the higher your level, the bigger the specials bonus gems get from being in a trap gets.
  
• Towers cause much more lag than traps.
  
• At high levels, your amplifiers skill gets so high that amplified towers end up with uncontrollable range, with being reduced to even 5% range still possibly being enough to cover the map, causing problems with not killing monsters in the mana farm.

• Traps have a rather short range, so short that if you put two traps next to each other, there will be gaps between them where monsters won’t get hit by either trap. Because of this, it is possible for fast monsters to run right over a trap when on 3x speed and thus fail to get hit.
  
• Being in a trap only boosts bleed and slow durations but if you’re a high level then these durations will already be rather long even without the boost from being in a trap, so you don’t lose any power from using a lantern.
  
• Lanterns can have 8 amplifiers around them whereas traps can only realistically have 6 amplifiers adjacent to them. This makes bleed gems in lanterns be able to inflict stronger bleeds.
  
• Because lanterns have a decent range and a faster rate at which they can hit targets (60 shots per second * 21 targets per shot = 1260 hits per second, 7 times faster than a trap), it is very difficult for a monster which enters the range of a lantern to not get inflicted with the lantern gem’s effects (provided that the monster doesn’t have any gem effects on it which the lantern doesn’t inflict), even if they have shields (since the lantern can shoot fast enough to rip them off).

LAG. The gain of the extra hitlevels is negligible, especially since the damage bonus from hitlevels is only additive with the damage bonus from the mana pool level. It is better to do the run without the lag and be able to start the next run much sooner than to be able to kill a couple of extra waves of monsters before downgrading the enrage gem.

The mana amp delta grades are nice but they only work when u upgrades are used to get the amplifier gem grades into ratio. There exist formulae for the true desired cost of a gem in an amplifier compared to the gems it supports for a given growth (in fact, the delta grades are derived from these formulae).

The cost of a mana leech gem in an amplifier relative to the traps it supports should be
Where thingsSupported is the number of traps the amplifier is supporting, specialsSpread is the specials spread of your amplifiers and gm is the growth of your mana combine (gm = log2(1.38) for u upgrades).

After taking the base 2 logarithm of the cost equation with gm = log2(1.38) and rearranging it a bit, you get
The real delta grade to be used would be this rounded to the nearest integer, but that makes the effect of thingsSupported change as specialsSpread changes. In order to make the effect of thingsSupported be invariable as specialsSpread changes, the two components can be rounded and then added together, which makes it slightly less accurate in exchange for being much easier to use.

The cost of a critical hit gem in an amplifier relative to the traps it supports is well approximated by
Where thingsSupported is the number of traps the amplifier is supporting, damageSpread is the damage spread of your amplifiers, specialsSpread is the specials spread of your amplifiers and gk is the growth of your crit combine (gk = log2(1.58 * 1.38) for u upgrades).

After taking the base 2 logarithm of the cost equation with gk = log2(1.58 * 1.38), using the fact that damageSpread = specialsSpread + 0.05 and rearranging it a bit, you get
The real delta grade to be used would be this rounded to the nearest integer, but that makes the effect of thingsSupported change as damageSpread changes. In order to make the effect of thingsSupported be invariable as damageSpread changes, the two components can be rounded and then added together, which makes it slightly less accurate in exchange for being much easier to use.

The cost of a bleeding gem in an amplifier relative to the lantern it supports should be
Where specialsSpread is the specials spread of your amplifiers, supportedGemStrength is how strong the bleed component of the supported gem is (which is 1 for pure bleed gems, 0.81/1.16*0.7=567/1160=0.4888 for rb2s gems and around 0.7 for well specced rb gems) and gb is the growth of your bleed combine (gb = log2(1.16) for u upgrades).

The total cost of bleeding gems relative to the total cost of yellow gems should be:
Where gb is the growth of bleed (0.214125 for U) and gk = gd + gc is the killgem growth at speedcap (1.145648 for U). This yields the ~20% figure used above in the guide.

As discussed in a section above, it's not worth it to put managems outside whiteout range, because they leech much less but lag the same, so they hurt XP/time.

Nevertheless, they do improve the efficiency of the manafarm, because duplicating a managem is more cost-effective than upgrading it (even with combines), so if one is going for maximum XP at the massive expense of time spent it makes sense to cover the track in them.

Managems outside of whiteout range should be lower cost than the ones inside, the optimal ratio is:
With max talismans (+74% mana) this ratio is ~3, which is between 1 and 2 grades. This also applies to amps that amplify these gems.

This guide was developed by

• 12345ieee
  
• Bill Wilson
  
• Tim

Note: Names in alphabetical order

We also thank the authors of the mods mentioned above and many of the players in the GC Discord for their input and their help in testing.