NEW FEATS
I also added many feats in the
Feats web page and in the pages of the character classes. Now I think I am finally up-to-date with all the feats I wanted to create, at least in terms of their presence in the web pages. Here are some feats I added recently:
(All) Defensive Archery: You no longer provoke attacks of opportunity (AOOs) from enemies when shooting or when reloading a crossbow. Requires: Point Blank Shot.
(All) Improved Potions And Herbs: You do not trigger attacks of opportunity when drinking a potion, and you only need a move action, rather than a standard action. You also get these benefits when using other consumable items, such as medicinal herbs.
(All) Improved Escape From Snares: You can attempt to break free from the Stuck condition obtained from effects like Entangle, Web, Ranger Web, Resilient Web, Glue arrows and Snaring weapons using a move action, rather than a full-round action. As a result, you could perform two attempts per round. In addition, whenever you are given a reflex saving throw due to a Grease effect, if you fail on your saving throw, you can immediately attempt a second one.
(All) Improved Escape From Paralysis: You can attempt to break free from the Held condition obtained from effects like Hold Person, Hold Monster, Chromatic Orb, Colour Spray, Paralysis Domain Power, Hold arrows and Coup de Grace weapons using a move action, rather than a full-round action. As a result, you could perform two attempts per round. In addition, whenever you are given a fortitude saving throw due to a coup-de-grace attempt against you, if you fail on the saving throw, you can immediately attempt a second one.
(All) Death Spell Resistance: Whenever you are given a fortitude saving throw due to a disintegration, death effect or death spell, if you fail on the saving throw, you can immediately attempt a second one. The effect can originate from a Slaying arrow or the spells Slay Living, Destruction, Implosion, Phantasmal Killer, Finger of Death, Circle of Death, Disintegrate and Greater Disintegrate.
(All) Overrun: With this feat, a character can move through squares occupied by enemy creatures of a size category no greater than the character's own size category. A character moves through enemy squares at half speed and this movement must not end on an occupied square. The character still triggers attacks of opportunity as normal. For example, a Half-Giant Fighter with the feat Overrun can move through the squares occupied by Hill Giants, but he cannot move through the squares occupied by creatures of size Huge and Gargantuan. Requires: Str 13.
(All) Fast Whirlwind Attack: You perform whirlwind attacks as a standard action, rather than a full-round action. Requires: Whirlwind Attack, BAB 2, Dex 13.
(All) Devastating Charge: The extra damage you deal on a charge is upgraded from 1d6 points to 1d8 points. Requires: The character must be a Centaur.
(All) Intensified Energy: The energy damage you deal on attacks is upgraded from 1d6 points to 1d8 points. Requires: The character must be a Half-Salamander.
(All) Cone Breath Weapon: Your 30' line breath weapon becomes a 20' cone breath weapon. Requires: The character must be a Drake.
(All) Improved Breath Weapon: Your breath weapon gains one additional damage die and the radius increases by 50%. Requires: The character must be a Drake.
(All) Swift Breath Weapon: Your breath weapon can be used as a move action, rather than a standard action. Requires: Level 6, the character must be a Drake.
(All) Breath Weapon Shaping: The creature's breath weapon or gaze attack no longer affects allies in the area of effect. Requires: The creature must have a breath weapon or a gaze attack.
(Death Knight) Swift Life Drain: Your Life Drain ability now requires a move action, rather than a standard action. In addition, you can use Life Drain once per round as a swift combat manoeuvre. Requires: Level 6 Death Knight.
(Champion) Divine Weapon: This feat allows the Champion to bypass aligned DR, meaning that his weapons will always be considered to be of the right alignment for the purpose of bypassing the enemy's DR. In addition, the Champion gains a bonus of +2 to damage rolls whenever using a weapon enchanted with Holy, Unholy, Axiomatic or Anarchic against enemies of the opposed alignment. Requires: Champion level 1.
(Champion) Greater Divine Weapon: The damage bonus from the feat Divine Weapon increases from +2 to +5. Requires: Champion level 12, Divine Weapon.
(Cleric) Improved Energy Defence: You can cast Protection from Energy, Mass Protection from Energy and Greater Protection from Energy as move-action spells, rather than standard-action spells.
(Cleric) Improved Holy Smite: When you cast the spell Holy Smite, the spell's damage limit of 5d8 is removed and the slow effect lasts 1d6 rounds rather than 1d4 rounds.
(Cleric) Improved Magic Stone: When you cast the spell Magic Stone, each stone deals 1d8 points of damage, rather than 1d6 points of damage.
(Cleric) Improved Sound Burst: When you cast the spell Sound Burst, the damage increases by one point per level.
(Fighter) Wade In: The full-attack action is now considered to be a standard action for you, rather than a full-round action. Consequently, you can launch a full attack after moving up to your speed, or after performing a single move action. Requires: Fighter Level 9.
(Fighter) Precise Swing: Your melee attacks, including attacks with a Reach weapon, ignore both cover and light concealment. Requires: Fighter Level 6.
(Paladin) Improved Holy Shield: The reflex bonus against Breath Weapon attacks obtained when holding a shield increases from +4 to +6.
(Paladin) Mystic Shield: When the Paladin is holding a shield, he gains a bonus of +2 on reflex saves made against spells. Anyone standing behind the Paladin also gains this bonus (a creature is considered to be standing behind the Paladin if that creature would not have been touched by the spell had the Paladin's square been blocked by a wall). This bonus stacks with any bonus obtained from Cover.
(Psionicist) Energy Sculpting: Select one of the following powers: Energy Bolt, Energy Cone, Energy Ball and Energy Wave. When you activate the selected power, you manipulate the area of effect so that allies standing within the normal area of effect remain unaffected by the spell.
(Psychic Healer) Healing Focus: All your Cure spells heal two additional points of damage. In addition, you activate the power Cell Division as a level-2 power, rather than a level-3 power, and the number of points cured by your Cell Division power is increased by 50% automatically.
(Psychic Healer) Psychic Healing: Your psionic powers Cell Division and Purification become Touch-range powers, instead of being strictly Personal-range powers. Requires: Level 6.
(Psychic Healer) Improved Psychic Healing: Your psionic powers Cell Division and Purification become Close-range powers. Requires: Level 9, Psychic Healing.
(Ranger) Water Affinity: When fighting underwater or while swimming, the character gains a +2 circumstance bonus to Attack Rolls and Armour Class.
(Ranger) Forest Affinity: When fighting in a forest or jungle, the character gains a +2 circumstance bonus to Attack Rolls and Armour Class.
(Ranger) Fire Resistance: The character gains resistance 10 to fire damage.
(Ranger) Cold Resistance: The character gains resistance 10 to cold damage.
(Wizard) Improved Magic Missile: You deal an extra point of damage with each missile from the spell Magic Missile. Requires: Level 1.
(Wizard) Greater Magic Missile: You deal an extra point of damage with each missile from the spell Magic Missile. Requires: Level 1, Improved Magic Missile.
(Wizard) Improved Range Magic Missile: Your Magic Missile spells are now long-range spells rather than medium-range spells.
(Wizard) Improved Range Chromatic Orb: Your Chromatic Orb spells are now long-range spells rather than medium-range spells.
(Wizard) Improved Range Grease: Your Grease and Resilient Grease spells are now long-range spells.
A WORD ON COMBAT ACTIONS
I recently implemented the Bull Rush action, but I made it a lot more flexible than in KotC 1. In KotC 1, you could only push forward a creature by one square. In KotC 2, you can do the same, but you can also decide to pull the target creature backward by one square, or slide the target creature by one square into any of the squares adjacent to your character. You can even decide to swap positions with the target creature (if both creatures are small or medium in size).
In the menu for combat actions, there are now three different options: Bull Rush, Pull and Slide. Use the Slide option if you want to swap positions. I envision Pull and Slide as a viable tactic for helping someone out of a web or grapple, or simply to throw someone off a ledge or ladder.
Here are some screenshots of Slide, Bull Rush and combat: (Click on a picture to enlarge)
Here are some screenshots associated with looting:
On the subject of combat actions, I've also been thinking about implementing a Distract action, which would be the equivalent of 'Aid Another' in the original D&D 3.5 rules. Distract would be automatically successful and it would decrease the Armour Class of a single target by two points against the next attack suffered by that target. The action will give the condition 'Distracted' and this condition will disappear after one round (when the turn of the initiator comes up again). The condition can be applied several times if several creatures use it on the same target.
A WORD ON SPELL EFFECTS AND THE EDITOR
Probably you have seen from the recent gameplay video that I have finally implemented the spell 'pipeline' (targeting, then visual effect, then gameplay effect), at least for simple damaging spells like Fireball.
Not much to say here apart from the fact that many different visual effects can be created with particles and colours. In the spell editor, we can choose a type of effect and a colour. I'm thinking to allow pre-viewing visual effects directly from the spell editor, but it might be complicated to do. It would mean switching from the editor to the game, loading an empty module and using a script action to execute the visual effects repeatedly. Anyway, here are some screenshots of a fireball in various colours:
And here are some screenshots associated with spell targeting:
Speaking of the editor, one more thing I'd like to have is a tighter integration between the editor and the game. For example, it would be good to be able to switch from the editor to the quick-save by just pushing a key, then switch from the game to the editor (and automatically save on the quick-save slot) by pushing a key, like Alt+S.
That way you could immediately edit the map which you've been playtesting. Or you could immediately correct or expand a dialogue script which you've been going through in the game. Then you could quickly go back to the game after completing the edit. It's important for the editor to be as good as possible so that many modules will get created. The more easy and fun it is to create, playtest and correct a module, the more modules I expect will get created.
On the subject of spells, I've added some code to display a tooltip explaining why you can't cast a particular spell. Just mouse over the spell's name and it will display the reason. I remember how in Drakensang you can't cast when wearing armour or holding a shield, I find it totally confusing when it happens in-game and the game does not give the reason for not being able to cast.
Here's a screenshot showing the explanatory tooltip, plus a screenshot from character creation:
On the subject of visual effects, I've added some random blood stains on the floor whenever an enemy is defeated. The red stains remain after the end of combat and they are even saved with the rest of the game's data when saving a game. There will be a script action to clear them up, if necessary. You can see the blood stains on some of the combat screenshots above.
Also on the subject of visual effects, threatened areas are now represented with semi-transparent red circles similar to the big red circle used when targeting a spell like Fireball. Here is a screenshot showing this effect, plus a character-creation screenshot showing the various tokens available to player characters:
KOTC 1 ARTIFICIAL INTELLIGENCE AND ITS LIMITS
I have been working on the foundations of the AI too. So far I've just been adapting the existing code from KotC but there is one thing I've started adding: a utility calculation.
The AI in KotC 1 was based on three factors: pre-set data for any given monster (e.g. a pre-set percentage chance to cast a spell, pre-set percentage chance to use bull rush, etc); rules (e.g. better flank a target than not flank it, better cast a high level spell than a low level spell, etc); and luck (e.g. between two high-level spells that are expected to be effective pick one at random, roll a dice and see if it exceeds the percent chance to use bull rush before using bull rush, etc).
It's a pretty good AI. It's fast and does reasonable things most of the time. However, there are limits. The limits are more apparent when a creature has access to a lot of options.
For example, in KotC 1, the Red Dragon, Tyrannixus, could not very well choose between casting a spell, break-free from a web, performing a full attack, or using the breath weapon. The creature would cast a spell when a full attack or a breath weapon attack would have been much more effective. There is no code in the game that compares the expected outcome of a full attack to the expected outcome of casting a spell.
UTILITY-BASED ARTIFICIAL INTELLIGENCE
When we compare the expected outcome, we make what is called a utility-based decision. We give a score to each action that we want to compare and then we choose the action with the highest score (or we choose randomly out of the four or five actions with the highest score).
This approach can be used to decide what spell to cast, for example, between a cone of cold and a fireball. Cone of cold should not be assumed to be better just because it's a spell of higher level. Instead we give a score to fireball based on the expected outcome of casting fireball and we give a score to cone of cold based on the expected outcome of casting cone of cold.
Likewise, we can use the approach to verify whether a full attack would be better than a spell. Say we have determined that our best spell (the one offering the highest score) is cone of cold. Is it really the case that casting cone of cold would be more effective than a full attack? To know this we also need to score the action 'full attack' based on the expected outcome of a full attack.
If the score for a full attack is higher than the score for cone of cold, we use full attack and ignore our ability to cast a spell. The human player often has to make that kind of decision. For example, when controlling a mid-level cleric. A mid-level cleric can fight reasonably well and he also has spells, but none of his spells are earth shattering. So a cleric must constantly decide whether he would be better off attacking or casting a spell.
However, there are also limits to a utility-based AI in a game like KotC.
LIMITS OF UTILITY-BASED INTELLIGENCE
For example, take the action to cast a spell to heal oneself. Firstly, it's a purely defensive action. So it cannot easily be compared to purely offensive actions like casting a damaging spell, performing a full attack, or using a breath weapon. Secondly, a human being would not select this action based on its utility value. Instead, a human player would select that action based on the desire to save an important character (even though this may not be the best move in terms of strategy).
Another example comes from combat manoeuvres like grapple, feint, disarm, bull rush, sunder or trip. A human being would not select these actions based on a deep assessment of the potential outcome versus the outcome of a standard attack. If an assessment was made, it is likely that the full attack action would often be the winner (trip is a possible exception). Instead, a human player would select these actions based on the desire to use a certain specialised ability for which the acting character has received training (feats).
The effectiveness of these actions, and thus their likelyhood to be used, very much depends on character specialisation through feats and ability scores - not so much on a utility value derived from the expected effect of the action on the target. Of course, it would make no sense to use trip on a creature who's immune to trip, but that consideration is more like a rule-based behaviour than a utility-based behaviour.
KOTC 2 ARTIFICIAL INTELLIGENCE
So my plan for the AI of KotC 2 is to use a mix and match approach. I'm not throwing away the bulk of the AI of KotC 1 just because utility-based decisions sound good. Instead I will keep the bulk of the KotC 1 AI, and on top of it, I will sprinkle utility considerations when these seem appropriate to me.
For example, I've already adjusted two functions: one function that picks a target out of all targets in melee-attack range, and one function that picks a target out of all targets in ranged-attack range. These functions now assess the chance to hit, expected amount of damage and expected effects on the target in order to assign a score to each target.
This assessment does not replace old KotC 1 code. Instead, it comes in addition to the other considerations (rule-based behaviours such as: it's better gang up on an enemy than attack a new one, it's better target a pure caster than a non-caster, etc).
HOW TO SCORE TARGETS
Now you may be wondering how we can score a target. I use the following chart:
The first part of the score is the effect we expect from physical and energy damage. For example, if the total expected damage is 100 and the target creature has 95 Hit Points and the chance to hit is high, then the score is equal to 1. The score decreases when the expected damage is lower or when the chance to hit is lower.
The second part of the score is associated with enchantments such as a Blinding weapon or Sleep arrow. Say we have a high chance to apply the Blinded condition, in this case it would yield a second-part score of 0.70.
Then we combine the 1st part and 2nd part. I want a score that reflects the effect on a single target, so I cannot just add up the first part and second part. For example, if the damage is highly likely to kill the target AND the enchantment is highly likely to disable the target fully, this is not much better than an attack that is highly likely to kill the target and has no enchantment. You cannot kill the same target two times.
So what I'm doing now is this: I take the highest value of the first part and second part, and then I add 25% of the lower value. So the score for that target cannot exceed 1.25.
UTILITY OF A FULL ATTACK OR DAMAGING SPELL
You may also wonder how to calculate the utility of a full attack or damaging spell. Well, for a spell like Fireball, we can simply assess the effect on each affected creature and add up the scores for each target. For example, if the fireball is expected to result in moderate injuries in four targets (all with a high probability), then the score would be four times 0.5, or 2.
For a full attack, it's more complicated because you can't really forecast what the sequence of attacks will be. For example, if your character has four attacks, maybe he will kill the first target in the first attack and then use three attacks on a second target; or maybe the character will require two attacks to kill the first target and then use two attacks on a second target. Also, the feat 'Cleave' may or may not be used to produce an additional attack on the second target or on the third target, if any.
In this case, I think it would be a mistake to try and assign a probability to each scenario like a statistician would do. There is an easier solution which I expect will yield good results.
The solution is this: first assess the score associated with the first attack on the first target. If this score is one or higher, assume that the target will be knocked out by the first attack and all subsequent attacks will apply to the next target. So now we calculate the score of further attacks on the second target, then we add up the scores on all targets.
But if the first score was below 1, we must assume that the target would not be knocked out by the first attack, and that at least one more attack would be necessary to kill the target. So now we calculate the score associated with applying the second attack to the same target. Then we add up those two scores. If the total is above 1, that means we're likely to kill the target in two attacks, and all subsequent attacks will apply to the next target.
So the system for knowing how many attacks will apply on each target is based on an addition of the scores associated with each attack. Whenever the total of the score exceeds 1, we assume that we will be switching to the next target. Then we total the scores for each target (making sure that the maximum score for any single target is 1.25) and the result is the score for the full attack action.
LOW-INTELLIGENCE MONSTERS AND REACTIVE BEHAVIOUR
As requested many times by players, low-intelligence monsters will not have the same behaviour as more intelligent ones. A low-intelligence monster is one that has an intelligence score of 5 or less. They will target the nearest enemy, or the enemy that damaged them in the previous round, if any.
For that purpose, every monster will keep track of the damage received in the current round and since the beginning. I might use that information for reactive behaviour.
For example, if a creature has received fire damage in a past round, and it is carrying a potion of fire resistance, then on its round, it might decide to drink that potion as a move action, rather than use the move action to move (there will be a feat to drink a potion as a move action).
Or if the creature can cast Protection from Energy as a move action, and on its round it has a spare move action, then it could just use that move action to cast the appropriate Protection from Energy, depending on the damage received earlier.
AI SPELL PICKING
In KotC 1, spell picking was essentially a non-weighted random choice between all of the spells of the highest spell level available to the caster, after pruning away spells that would not affect any targets. There were a few exceptions for some spells to be cast even though they are not of the highest level available, for example Fireball.
In KotC 2, the AI will be doing the same thing, but there will be a twist. Spells will be given a utility score. Then, instead of a non-weighted random choice out of all possible spells, we can do a weighted random choice out of, say, the five spells with the highest utility. For example, let's say the AI has pre-selected the following five spells:
- Cone of Cold, score 4
- Fireball, score 3
- Stinking Cloud, score 3
- Lightning Bolt, score 2
- Scorching Ray, score 2
In this case, we can do the following weighted random selection:
- Cone of Cold, probability 28% (4/14)
- Fireball, probability 21% (3/14)
- Stinking Cloud, probability 21% (3/14)
- Lightning Bolt, probability 14% (2/14)
- Scorching Ray, probability 14% (2/14).
You may ask, how to deal with spells that allow the caster to choose an energy type (fire, cold, electricity, acid, sonic)? In the AI processing code, those spells need to be expanded into five versions, each version corresponding to a different energy type. Then we calculate the utility score for each version. Finally, we keep only the version offering the highest score. If several versions offer the best score, we pick one of them at random.
COMPARING SPELLCASTING WITH OTHER ACTIONS
Once the AI has done its weighted random selection, we can compare the score of the selected spell to the score of other actions in order to avoid having the AI cast a spell when in fact something better is available. For example we may have the following list:
- Spell: Fireball, score 3
- Melee full attack, score 1
- Ranged full attack, score 0
- Breath Weapon, score 4
And then we just pick the action with the highest score. So it could be that the creature ends up not casting a spell.
EDITOR SCREENSHOTS
I recently upgraded the appearance of all editor screens, so here are some screenshots of the various parts of the editor.
Of special note is the Creature Wizard which I have upgraded to allow specifying things like a Druid's Sphere, a Wizard's Attunement or a Bishop's domains:
Some more screenshots of the various editor screens:
NEW BACKGROUND IMAGES
Roman, the graphic designer, has created a nice environment containing several bridges hanging over a big river of lava.
Here it is in full size, for your eyes only:
http://www.heroicfantasygames.com/FWE/Environments/LavaBridge.jpg
And here it is as it appears in the game and in the editor (the first screenshot below is extra wide):
In addition, I've asked him to do about 7 or 8 more big maps. They will all be typical D&D environments: dungeons, temple, sewers, caves, castle, and a forest clearing.
OKAY, SO WHAT'S NEXT
Next for me is, first, to fix a bunch of bugs. Every time I make an update and collect screenshots for the update, I seem to harvest a number of bugs.
After that, I want to continue implementing the AI for melee attacks and ranged attacks. I can see this taking a couple of months.
Once that's done, I guess I'll continue working on spells. First by implementing the Wizard and Cleric spells of early levels and adding new visual effects for spells.
Then I need to finish implementing combat actions. I've done bull rush / grapple / melee attacks / ranged attacks, now I need to implement the remaining combat manoeuvres, and the box dedicated to automatic manoeuvres conducted as part of AOOs.
Then more work on AI - I will need to implement the AI for spellcasting.
Then more work on combat actions - Ready versus Spell, counterspell, domain powers, turn undead, spell-like abilities, breath weapon, activate item, special class abilities like those of the Ranger, special monster abilities like Death Throes, weapon enchantments and abilities associated with Feats.
Then more work on spells - notably, I need to implement spells that affect the environment like Web and Solid Fog. And spells that are more complicated than others like Dispel. And psionics which allow selecting an energy type.
Then more work on AI. I can easily see myself spending 5 months on AI and 5 months on spells. Or more, easily.
I'm planning to do a web page with screenshots, too. As for future updates, I'm hoping that they won't need to be as big as this one, since I've finished most of the web pages I wanted to do, and the game's interface might not change much anymore. So I expect that future updates are going to be mostly about spells, AI, combat rules implementation, feat implementation, environments and monsters.
Separately, I have plans to move and buy property. Those things may slow down development at some point, hopefully not too much.
Thank you for reading and, as usual, feel free to comment anywhere in the forum, on youtube, RPG Watch, RPG Codex and elsewhere.
