Ok, so combat. This is the real meat of things where WH3 is concerned.
First, I should note that it does a perfectly fine job for out-of-combat stuff. The skill list is long enough to give some sense of variety, and the stat-emphasis more or less mandates a certain amount of competence. There are good guidelines for reading the dice in useful ways out of combat, excellent advice on how to say yes with qualifiers (in the form of misfortune dice). For non-combat resolution, the game is not going to make me stand up and shout, but I have no complaints. To break out the inevitable 4e comparison, I feel like WH3 managed to provide a much more robust non-combat engine for play.
So that said, onto the fighty bits.
In a fight, you determine initiative in a classic sort of way and keep that order for the duration of the fight. Straightforward enough. Each turn you also set your stance, which will either be reckless or conservative. There's a physical track to represent it, and each step on the track is also a numeric one, so if you're three steps down the conservative track, that has a specific mechanical impact when the dice roll.
When you go, you get to take and action and a maneuver, which D&D players will roughly equate with a Standard/Full action and a Move/Minor action. There's no idea of a free action as its own thing, but things feel fast and loose enough that it doesn't seem entirely necessary.
To take an action, you play a card. Every character gets a set of core cards by default with things that everyone can do, like make an attack or go full defense, as well a some that have stat dependencies (you need to have a decent agility to get the "dodge" action). You also have a few more cards that you purchased during character creation that are hopefully full of cool and interesting things you can't wait to do.
The action cards are double sided, and which side you use depends upon your stance: use the green side if you're conservative, use the read side if you're reckless. The mechanical difference between the sides can be non-existent or profound, and it quickly becomes clear that you want to pick action cards that complement your preferred stance. The card tells you what dice to roll, against what, and you scoop up those dice.
This step is a little convoluted, but I suspect it gets faster with time. Constructing your die pool basically follows this process:
1. Pick up a number of blue stat dice for the stat you're using, and a number of purple difficulty dice based on how hard the action is (usually 1, plus 1 or two more if there are modifiers - you only get 4 purple dice, so they are far form numerous).
2. Based on your stance, swap out some blue dice for red (reckless) or green (conservative) dice. The number of dice swapped out is based on the number of steps you are into the stance, so Reckless 3 replaces 3 blue dice.
3. If you're using a skill you're trained in, add a yellow expertise die. If you have any talents or other bonuses going into them, add some number of white fortune dice to represent them
4. If there are any things working against you, like enemy skill or situational penalties, the GM adds them as black misfortune dice.
So, that all sounds a little complicated, and it is. The player should be able to build most of his pool of stat, stance and expertise dice on his own, and some fortune dice may come from known sources, like talents or training. The GM contributes the number of difficulty dice, and fortune and misfortune dice as appropriate. I think it takes time and familiarity for the cadence of this to get as comfortable as calling out (or knowing) difficulty numbers. Initially it feels a bit like a game of its own, and that's distracting, but it gets much easier. I wonder if it would be easier still if there was enough transparency for the player to build a pool on their own, but the ability to sprinkle fortune and misfortune dice is one of the GM's big tools, so probably not.
The actual dice are simple: all the good dice can generate successes and boons, and some can generate critical successes, and the bad dice (purple difficulty and black misfortune) generate failures, banes and critical failures. This is only slightly muddled by stance dice, which are statistically better than the stat dice they replace, but also carry risks: red reckless dice have big rewards, but can also cause failure or penalties on a bad roll. Green conservative dice produce reliably good results, but can slow you down, which can suck.
Reading them is a bit more complicated. At it's core, there are six outcomes a die can throw up: a success, a failure, a boon, a bane, a critical success or a critical failure. There are certain other special outcomes, like multiples (double success, double failure), modified successes (a success but you're delayed, a success but you gain stress) or the symbol that indicates a critical wound, and while they're interesting, they're unimportant.
Success/Failures and Banes/Boons cancel out, so at the end of the roll you will only have the net value represented. In the abstract, success/failure speaks to what you're actually doing, and bane/boon speaks to the stuff around it, so you might end a roll with some successes and some banes, and narratively that might mean "You hit, but end up in a disadvantageous position", while a failure with boons might be "You miss, but you get a chance to catch your breath".
The critical success and failure (Sigmar's Comet and The Star of Chaos) are like super banes and super boons, though they don't cancel out.
Mechanically, this is handled with result lines on the action cards. They basically list the various outcomes based on the number of results (in many cases, for example, a single bane or boon triggers nothing). The effects of the comet or star are usually listed here as well, often as something more potent than a simple bane or boon outcome.
At the table, this all resolved pretty easily. There were some issues of dice visibility, especially with the stance dice. They're d10s, and some of the faces have 2 symbols on them, which lead to a few cases of "is that an axe (success) or an eagle (boon)?" The difference is pretty clear on the 6-sided fortune, misfortune and expertise dice, but it's not as obvious writ small. Thankfully, this was not a huge problem, but it was a bit of a speedbump.
Anyway, once you take your action, you resolve it based on the outcome of the dice and the effects listed on the card. No problem. This probably results in some damage to the target. While the system is incredibly familiar (weapon establishes base damage, armor reduces it), the damage itself is handled quite cleverly. "Wounds" are cards handed to the person hit. In the case of a critical (or in certain situations where your wounds can be exacerbated), you simply flip over a random wound card, and the rules for whatever critical it is are on the card. For example, our Wizard took a "Blow to the Head", so he had to roll an extra misfortune die when rolling intelligence.
The other thing you can do, your maneuver, is where some of the interesting stuff lives. Most of the uses of a maneuver are predictable - draw a weapon, pick something up and so on - but it also covers movement, and that is very curious indeed.
Combat in WH3 is gridless, but it does use miniatures. Specifically it uses cardboard standups, but you could just as easily use D&D minis to the same effect. A lot of the ideas are going to be familiar, but they're tied together nicely. The basics are pretty simple.
If Bob is fighting a skaven, then we establish how far apart they are. They might be at extreme range (just within sight), Long range (close enough to shout), medium range (a couple dozen paces), Close (just a few steps away) or Engaged (Close enough to fight or otherwise interact. Once we establish that, they can use maneuvers to change that distance, though it takes more maneuvers to traverse longer distance. You place a token between the two people or groups for each increment of distance.
If you add more people, then entire knots of people can be engaged. If Sarah and a Goblin join the fight, then Bob, Sarah and the Skaven might all be engaged, while the Goblin is at Medium distance from the engagement, peppering things with his bow. The distance is represented by 2 tokens between the goblin and the engagement - if he was further out, it'd be 3 or 4 token.
Now, there's an obvious problem with this, which will probably jump out at anyone whose done engagement-based design (it's not an uncommon idea). If Sarah disengages (takes one maneuver, puts her close to the engagement) and then moves further away to medium distance, how far is she from the goblin? The game doesn't really address this.
This is not a dealbreaker, but it would be nice to have more guidance form the text. Still, one joy of approximating things with minis is that its not hard to adjudicate. If Sarah's mini was slid away from the goblin, then she moved away from him, putting him at relative long range to her. If she moves towards him, then maybe she's close. It's just something to keep track of, and it's likely to be problematic in more complicated melees.
Whew. Ok, lots more tidbits and impressions, but they can wait until tomorrow, where this hopefully wraps up.
1 - You are expected to track this with one of the jigsaw puzzle tracks and some tokens. This works poorly, and is one of the areas where the philosophy of "components for everything!" shows some holes.
2 - It could technically be neutral, but that seems to be a terrible, terrible idea. I'm not even use you can use actions in a neutral stance.
3 - The stance system is pretty cool, and it's an interesting class differentiator, since certain classes are more reckless or conservative than others, represented by how many steps they're allowed. A balanced class has 2 of each, but some are 3 and 1 or 1 and 3. This is also greatly helped by the fact that advancement allows you to purchase more "steps" over the course of your career.
4 - One of the brilliant things of the game is hidden in here, and it's something 4E badly needs to mimic (and I'll testify that I do something similar, and it works). One of the cards is "Perform a Stunt", which is to say, to do something not covered by the cards. Obviously, characters should always be able to do such things, but by explicitly calling it out as an option, it makes it much more likely that its an option players will take. All 4E needs is a power block for "Do Something Awesome" on the character sheet or among the cards, and it could gain the same benefit.
5 - In a bit of genius, fortune and misfortune dice don't cancel out. So if you've got superior position (and gain a fortune die) but bad visibility (gain a misfortune die) the GM gives you both dice. This is wonderful, at leas to me, because more dice == more fun. It would be far more dull if the dice canceled out and it was a wash.
6 - Powers take a certain number of turns to recharge, and you track this by keeping tokens on the cards. Conservative dice sometimes cause you a delay, which means the GM can put two more recharge token on one of your powers. I suspect the intent may be that the powers have to be ones with recharge tokens already on them, but since it doesn't say so, then the GM can put them on anything, including things like basic attack. This is moderately rough, but it's hugely bad for wizards, because the GM can put it on the action that allows them to generate power for spells, and pretty much jam the wizard up entirely. Yes, they can still do basic actions, but it's kind of anti-fun
7 - You get one free maneuver per turn, but you can take more maneuvers by accruing fatigue. The fatigue and stress (mental fatigue) rules are in the game, and they're fine, but they're not really interesting enough to bear much mention.
8 - And everyone in that engagement would get hit by an AoE attack. As a "dark" game, they can get away with this, since dangerous explosions are more in theme than the pinpoint blast radius of 4e.
9 - This is an area where the limit on the number of players in the core set (3) helps the rules. 3 players means the fights can only be so complicated.