It's too early to be actually useful, but you should do a search for roguelikes with multiple characters (very specific, no? :V) and play them.
To see what you think of their way of implementing multiple characters (personally I think it's a hard enough to keep 1 character alive ).
My brain's still asleep, can't be more useful sorry. I watched your dungeon exploration video though, it looked pretty nice. Maybe consider showing the outlines of dungeon areas that you've explored but can't see anymore (as opposed to everything just going black again).
Well, there's going to be one character at first, but my eventual aim is to allow more and to try to build that in to the game from the start.
For example instead of giving each PC/NPC a full turn of movement at once, I'm going to break down the turns in to phases, so all the characters in the players team can move in the movement phase, then next the monsters move. Then the shooting/ equipment use phase, where if you're not in combat you can swig a potion or fire your crossbow or whatever. Next will be the combat phase where you can try to kill the enemy or concentrate on hiding behind your shield, and finally the recovery/resolution phase where characters with minus health are going to die (or go in to a coma), or else you will regain your strength and movement points ready for the next round. If there's no enemy around we can skip the combat phase.
I've played a lot of single character RPGs and lots of team RPGs and I just prefer to have a team, someone to tank, someone to heal, someone for artillery, there's just something for everyone to do each turn, and the interplay of characters give life to the game. Also it gives you another level of tactical play, especially with permadeath. What happens if your tank gets killed? What if your healer falls down a pit trap and you have to do without him for 3 levels of the dungeon? What if your scholar gets permablinded by spider venom and can't cast spells anymore? Could you just relegate him to carrying loot while your thief learns magic? It's going to be important not to over specialize your characters in case they need to step in to someone elses shoes.
The videos so far are more kind of code tests, they show what can be done but they will be changed somewhat as time goes on. I'm thinking of having the sections of the dungeon stay visible for a number of turns before turning black, maybe greyed out, or just as it is. If you find paper and pens in the dungeon maybe a character can use a map making exploit/skill to keep the sections permanently visible.
I'm going to see how those things turn out in play testing though.
I have actually produced a fairly "complete" game in the past using the blender game engine, which puts me one up on most Blender game engine users. ^^ I learned a lot about procedural generation and enemy mob generation. It's just a high score game, based on the Battletech tabletop game, I had hoped to continue development but my old code turned out to be difficult to move forward with.
You can check it out here if you're interested:
http://youtu.be/53pDNwKNQ4Yhttp://youtu.be/vrgxdWua0b0I gues it is actually kind of a roguelike as the missions and enemies are procedurally generated, and death results in the total loss of your high score.
For the game I gave each potential enemy a cost in resource points (I wrote a program to calculate this as the original tabletop costs didn't translate well to the real time combat model) and randomly added enemies to the spawn point while reducing the spawn pool of RP. Lists of enemies were drawn up acording to faction and timeline and then selected by using a random index.
I'll probably do the same with the RPG game, with monsters selected by theme and level and then spawned from a pool of RP for each room, and also for a wandering monster pool to populate the corridors. Loot will be handled in a similar way, either placed in a chest in the room, or added to the monster's corpse.