Anthony Stark

AI in video games is a complex subject, and its implementation depends entirely on what sort of game systems you're using. It's difficult to talk about it in broad strokes. In an IE-style game like PoE, the AI is usually broken down into a series of potential actions. A monster, for example, might have two actions that it can perform: -Move to a target and perform a melee attack. -Cast a healing spell if I am below 50% HP. These actions all have a series of conditions that must be fulfilled before that action has the potential to be chosen. Moving to a target requires that the monster is able to move (they do not have a status effect that restricts their movement), that they are able to find a reasonable path to the target (using A* or something similar), and, in the case of PoE, that the monster is not engaged. You can set these conditions to either be absolute (they must be true for this action to be considered) or variable (e.g., even if I am engaged, there's a 20% chance that I will choose to break engagement if the other conditions are true). Once the AI runs through its potential actions, it compiles a list of valid actions that are then "weighted." What this means is that monsters will generally choose to perform certain actions over others if that action is a valid choice. In the case of the example monster above, casting a healing spell at 50% is given a high weighting, while using a melee attack is given a low weighting. This means that, if the conditions for casting a healing spell are met, the AI will almost always try to cast a healing spell over using a melee attack. Weightings don't have to be absolute, so you can still have a variable chance of the AI using different abilities during combat, assuming all of the conditions for every action are met. Once the action is chosen, it's carried out by the AI. After being carried out, the AI then selects another action using the above process. It's important that the AI is checked semi-regularly to ensure that it actually is able to perform its currently-queued action, or else you may end up with a case where the AI has an action queued, is suddenly prevented from performing that action (such as their previously-accessible path becoming blocked), and therefore has no more actions to run and merely sits there dumbly.

The Temple of Eothas is entirely about utilizing proper positioning and downing them before they can get to your backline. I completed all of the Temple with a 3-person party (Priest, Fighter, Wizard) on Normal, and with a 4-person party (Fighter, Wizard, Priest, Rogue) on Hard. Generally, I'll block the Shadows/Shades at a doorway, keeping my backline far enough away that the Shadows/Shades will not teleport to them and instead teleport onto my tank. They seem to have either a cooldown on their teleport ability, or a scripting hook that prevents them from using it with wild abandon, so this gives me sufficient time to position my backline appropriately. From there, it's simply a matter of wearing them down. Arcane Assault (due to its Raw damage type) helps with this, along with Rogue sneak attacks. Fan of Flames and Sunbeam both work excellently on them, especially with a cast or potion of Eldritch Aim beforehand. For large groups, I may simply send in my tank first to grab the attention of the Shades/Shadows, and then move a Wizard or similar class in position for AoE damage. Arcane Assault was given to Wizards so that, even without using spells, they have a way to contribute in terms of both damage and debuffing during combat. Grimoire Slam is alright, but the fact that it targets Deflection and doesn't have the same accuracy bonus that Wizard spells do (which should be fixed, IMO) makes it less good than it apparently seems. In most cases, the engagement-breaking component is unnecessary with good positioning. With regards to the damage of Wizard spells, the first level AoE spells are actually quite good when deployed against opponents without much DR against that damage type. 45 damage per target is a large amount of damage, and I've seen Fan of Flames climb to upwards of 80 with Eldritch Aim. If there's an issue with them, it's that their main method of deployment (cones and touch-ranged spells) are not consonant with how defensively weak Wizards are. Druids, by comparison, have a higher base Deflection, a higher base Endurance/Health, and defense/DR-increasing forms, but their main damage spells at level 1 are largely targeted AoEs, and are able to do comparable damage. The single-target spells, meanwhile, are largely not worthwhile, dealing low or frequently-resisted damage.

At level 9, you will have 5 level 1 spells per encounter if you take that talent.

It's not really that it targets Deflection so much, but that it (somewhat bizarrely for a "panic button" ability) lacks the inherent spell accuracy that all of the other Wizard abilities get. It's only using the base accuracy that Wizards get, which makes it miss a hefty amount of the time.

I believe both the Minor Spellbind and scroll versions of Jolting Touch missed a damage downgrade at some point right before release, so they're vastly more powerful than the current basic spell.

I fear this would mostly lead to what can best be called the "fidget strategy": attack them in melee until they back off, then switch to guns and move forwards just enough to shoot them until they decide to come back into melee, at which point you retreat back into the chokepoint.

I would say that the only realistic solutions are: Design future maps to specifically minimize chokepoints and require the player to engage enemies in much more open spaces. Create hard AI leashes that drag enemies back to certain spawn locations if taken too close to doorways. Create more Shadow/caster-type creatures that can make fighting in chokepoints disadvantageous, either because they can make the chokepoint a deathtrap themselves, or because they can bypass the tank line entirely.

This is a bug, and it happens when hard CC effects are reapplied while a previously-applied effect of the same type is still active.

My overall feeling on the difficulty is that it's simply too easy to block chokepoints, apply CC/debuffs, and chew through opponents without much danger. The funneling method is both very simple to do, and very effective, as it drastically cuts down on the number of opponents attacking your tanks at once and prevents them from being flanked. Opponents either need more forced movement (so that they can break through a front line), more dangerous ground-targeted spells (to make staying in one area dangerous), or more ways of circumventing defensive lines (e.g. Shades/Shadows). Part of this is, admittedly, a consequence of level design. There are few areas, especially at higher experience levels, where you cannot find a chokepoint to block off. While wilderness areas tend to be more open, there are still frequently sections that you can force opponents to run through and therefore stop them with your tank line. Fixing this is easier said than done, as it either requires some sort of leashing (enemies run back to their starting positions after being dragged X meters away), or specifically defining and enforcing a combat arena for each encounter to prevent the overuse of chokepoints. The structure of PoE as an IE-style game largely prevents the latter, though a case could be made for the former.

If you're looking for survivability, you have no need for Dexterity, and Constitution is point-for-point much worse than Perception.

From what I've read on the Nexus, the only way to get around this currently is to use a hex editor to change the necessary lines. If you check out the Arcane Veil Modal entry on the Nexus, the creator goes a bit more in-depth with regards to how they changed it. Could you explain the process you used to edit them? I've done modding with NWN1/NWN2 before, but the process here seems a bit more opaque.

It should be noted that scrolls and potions are relatively cheap to produce in this game, and can be used by any class with enough Lore, so it's easy to just make something like a Paladin who uses Scrolls of Fan of Flames to replicate the effect.

There are two main issues with this: The first "sustainable" Deflection-boosting spell that Wizards can get is Llengrath's Displaced Image, available at level 5 (and with only 4 casts at level 7). That's a fairly long time to be waiting for a solid Deflection score. This also precludes you from actually using Tier 3 and below spells for offensive/CC purposes, since you mainly have to spend them on improving your defense or offense instead. Wizard buffs do not stack with the buffs of other party members, meaning that, unlike other classes, you won't get the increased benefits of Circle of Protection (for Deflection, at least)/Scroll of Valor. It's not that you can't get into melee combat as a Wizard, but that there aren't a lot of good reasons to (save for 1/encounter Grimoire Slam).

Currently, the biggest issue is that the Wizard is one of the slowest starters in the game. Tier 1 spells have very niche applications (especially now that Slicken has had its duration reduced significantly), and are especially lacking in damage options for the early game. This is partly a consequence of the damage types used (Freeze damage is commonly resisted) and partly a consequence of the targeting types used. Both Fan of Flames and Jolting Touch require the Wizard, which is a very fragile class, to be within melee range of opponents (or to maneuver themselves to be open to attacks). By comparison, Druids have several ranged circle-targeted AoE damage spells, allowing them to remain in the backline in relative safety, which means that they have more opportunities to contribute to damage. Honestly, the biggest use that I've found for Wizards so far is utility casting. Arduous Delay of Motion and Expose Vulnerabilities are both excellent spells, and Curse of Blackened Sight is now very handy in large group engagements. I'm not really sure that they measure up to Ciphers still, but you can get some use out of them. I would advocate for some changes to their earlier spells, however, to make them more party friendly.

Skaen is about vengeance and just desserts more than trickery. He promotes deceit, but only in the interest of eventually striking back against those who wronged you. In that sense, he's a good ally for Woedica, as she wants vengeance against the upstart gods and is willing to take underhanded means to do so. Wael is the trickster figure of Eora who wants nothing more than to see the well-laid schemes of the gods implode, even if that means that he's also weakened for it.