Our DM works fairly improvisationally, which results in a playstyle that is fluid and responsive to our actions. We have relatively short sessions, so we like to keep the game moving. There is no criticism of the DM's style here at all, it's what we all enjoy.

We are currently 8th level. The rogue in our party was recently replaced by an wizard, who can cast Arcane Eye. Now, rather than the rogue being able to scout at will, when we encounter dungeons we use Arcane Eye. The problem with this is that the 4th-level slot, 1 hour duration and concentration mean that we want to get the most use out of a single casting by scouting as much as possible in one go, rather than just scouting ahead by one room as we crawl.

So far, so completely within RAW and objectively the most tactical choice.

However, this can be a bit tedious as players to plod through a load of rooms at once, and forces the DM to come up with all the rooms of the dungeon immediately, rather than giving him sufficient time for his improvisational creativity to flourish. Obviously there's the barrier caused by the 1 inch diameter access for Arcane Eye, and what's most likely to happen if nothing changes is that dungeons will become full of suspiciously airtight rooms!

We've thought of some options:

  • The DM could plan ahead more - but we don't want him to have to do this, it becomes much more work for him!
  • We could modify the spell - for example, we could have a homebrew variant that increases the duration and doesn't require concentration, but reduces the distance it can travel from the caster to "next room" distance like 30-60ft.
  • We could just live with it and the DM can obviously decide to modify the world as he sees fit to cope with our abilities - but this seems a little frustrating to us as players (and to my character with a spell that becomes less useful).

How can we reconcile an improvisational DM style with a RAW mechanic that allows rapid access to information about the world? This is just one example, at higher tiers there are a lot of abilities that would allow similar rapid scouting, and there are certainly parallels in other areas, particularly with divination.

In particular:

  • Is the homebrew variant suggested above balanced and a suitable compromise?
  • Are there existing rules and/or advice in the core books or 3rd party literature that cover this situation?
  • Do any other DMs or players have experience with a similar issue, and how was this dealt with?

(Thanks to the commenters who suggested useful rephrasing advice)


3 Answers 3


In a nutshell : Tell the players the main points. And for the lesser details, take a page out of Dungeon World's hold mechanic and let them ask questions later as they play.

As a GM how uses a mostly-improvisation and grid-less (sometime even mapless) style while expecting the players to scout ahead and prepare, here's how I do it. Disclaimer: I don't use D&D for my games, so I won't give exact numbers for skill checks, but I'll try to give an idea of the reasonning behind those numbers. I also describe my dungeons as "many interconnected rooms with a few of importance" instead of a strict map of densely packed rooms like many maps show, this can help the players deal with not having all the information upfront.

In a few steps :

  1. Give the players a bird eye view of the map. The ambiance, the mood, the common features of most of it and the huge hotspot that the GM probably has planned even if he improv everything else (If not, now's the time to make them up).
  2. Let the players ask questions or make any maps they see fit to do, the limit on questions and actions at this point is the 1hour duration of the spell. Scanning a room is short, but precisely mapping part of the dungeon is a sizeable portion of the budget. Possibly the entire spells.
  3. Let the player make a roll, give them tickets they can exchange for answers as they explore later. Things they "would have investigated while scouting if they had all the details". Similar to the Hold mechanic from Dungeon World or to the questions allowed by many divination spells in D&D.
  4. Finally, if, while exploring, the GM wants to add some detail they feel the players would have noticed automatically when scouting, he just tells the player as they explore.

A demonstration

  1. The GM describes the overall atmosphere of the dungeon:

    An old cathedral, taken over by vegetation, sturdy supporting walls with a labyrinth of thinner stone walls and remains of broken wooden separations between individual rooms and private spaces. Maybe close to a steep cliff from which boulders changed the layout.

    GM also describes the main Points of Interest and dungeon layout :

    The heavy outer door is shut (PoI#1), but the eye can squeeze throught. Then the main hall (there isn't much to see here at first sight), then the smaller back rooms meant for the visitors. A river fed from the cliff cuts the back room area clean in half (PoI#2), again the eye can fly over. Finally, in the private back rooms is a huge stone coffin, seemingly airtight (PoI#3) which the party labels the Vampire Coffin.

    If the GM is 100% improvising, these are probably the only things the GM actually prepared before-hand. Everything else will be on the fly.

  2. During the description, players ask questions as they come to their minds. Whatever the GM answers, he notes the answer because those details can't change anymore. In my experience, the players ask question about what is most obvious in the narration : the overall structure and layout and the Points of Interest. If they start exploring every room and layout, see the next point about making a map.

    • If the players want to make a map or note the exact location of something like a monster camp. The GM can show it approximately on the party's map. But more importantly, he notes what the party knows and will notify the players when they approach one such points. If they make a detailed map, it takes time and the eye probably can't go all the way throught the dungeon. Or if it can the players can't ask many questions because the eye spends all it's time making the map.

    The players enquire about the outer gate: it's quickly obvious that the mecanism will require in-person operation to understand.

    They then look for a way for their characters to cross the river. There is no easy way, so the GM tells them that if they want to find another path throught the cathedral's structure, they'll have to roll Investigation to see how long it takes to find a way with the eye. A failed roll will limit how much they can explore further ahead.

    Whatever they do, they enquire about enemies they might encounter. The GM describes some of the patrolling monsters (monsters are not always in the same place or at least they are in otherwise innocuous rooms) and the general area where they are to be expected.

  3. Before finishing, the GM gives the players a number of questions they can ask later to recall information they learned while scouting. Such question could be :

    When they arrive to a suspicious door (for whatever reason, that's up to the GM imagination at this point),

    Players: What is behind that door?

    GM: dash 1 question That is a meditation room, it was probably used as a dormitory. You saw no monsters with the eye, but there were some arounds.

    Then later after clearing the room :

    Players: What is the best way to get out of this area with the least number of fight?

    GM : Is that a question?

    Players: Yes! We're too hurt after the vampire fight.

    GM: dash a question Highlight a path where there were less patrols and suggest a strategy to better navigate this

    This is the part where I'm not going to give much details because the systems are different, but I usually do it so that a bad roll will still net enough questions to navigate throught the dungeon if they use them wisely. Usually that would be something like DC5 (assuming the wizard has a high Int and/or proficiency) for 1 question per section of the dungeon. This way, I can assume they always have enough questions to go throught and if they get more they feel smart for being able to play with the dungeon or use them to get additionnal informations.

  4. If the GM wishes to add something on the fly that he would have told the players when scouting. The idea is that the GM assumes the characters are competent and relays to the player information that the character would know as they explore. Even if event the GM didn't knew it at the time of scouting. He can do so by just telling the players:

    Hey you remember that there is a troll nest in that section of the dungeon. Beyond this door you saw chewed bodies and a few rooms later you'll find the nest itself. The troll was sleeping when you scouted it.

    I personally suggest to keep those additions to things that would not be a big deal for the party. This should be used to warn the players of an upcoming danger but not one they'd wish to avoid. Otherwise the players are likely to say "Well we would have gone a completely different way if we knew there was a troll!". And they would be right and now you have to turn back time or deal with players who knwo you can lure them into an ambush. So be careful with those.


Let your DM declare the dungeon scouted and then improvise scouting information on demand

Just because your characters now know the entire layout of the dungeon doesn't mean that you need to. If your DM is willing to work with you on this, he can just declare "ok, you spent an hour scouting the dungeon", and then going forward, he can give you opportunities to use that knowledge. For example, if the DM would originally have said something like "you come to a fork in the hall, with 2 identical-looking hallways stretching ahead to the right and left", he can instead tell you about the rooms that each hallway leads to. At a higher level, the DM could ask you what kind of route you are looking for: the shortest route, the least-patrolled route, the route with the most hiding spots, etc. Then they can "ok, you follow that route" and adjust the number and nature of the encounters along the way accordingly. In fact, your DM might be thrilled with this idea, because it effectively gives him a way to ask you directly what kind of dungeon experience you're looking for and then give it to you, all within the framework of the rules. Or, to look at it another way, it gives you a rules-supported way for the DM to let you collaborate with him on the (improvised) dungeon design.

If you want to hang some more concrete mechanics off of this, you could borrow the idea of "holds" from the Apocalypse system. For example, your DM could decide that 1 hour of Arcane Eye is enough to scout out 10 rooms. Your DM might determine the exact number of rooms based on an investigation check, or he might ask you how thoroughly you are scouting, e.g. just getting the layout vs. detailed observation of enemy numbers and behaviors. In any case, let's assume for example purposes that he decides on 10 rooms. That's 6 minutes per room, including travel time between rooms. So, you get 10 holds, each of which you can spend at any time during your dungeon delve to know the contents of the next room ahead of you along your current path (or along the path of your choice if you're at a fork). Each time you spend a hold, you are essentially ret-conning that this room is one of the rooms you scouted back when you cast the spell. Your DM then tells you everything that your character has already learned about this room from their Arcane Eye spell. Narratively, you could think of it like a flashback, of the kind that is common in heist movies (e.g. Ocean's Eleven).

Essentially, this is having your cake and eating it too: you get to implement one-room-ahead scouting, thus allowing your DM to maintain his improvisational DM style, but you don't have to actually change how the spell works. You are just taking the knowledge gained from the spell and making it abstract, with the concrete details to be filled in later on demand. The main obstacle to making this solution work is that you need the DM's buy-in, and you need to trust your DM to "play fair" on his end by giving you opportunities to make a meaningful choice based the knowledge your character has gained.

However, it's also worth noting that this solution is a compromise: you aren't going to get exactly the same information from the spell that you would get if the DM had the dungeon already mapped out. The only way to do that is for the DM to actually map out the dungeon. But you are already familiar with this compromise, because it's ultimately the same compromise you're already making when you play with a DM who improvises dungeons on the fly.* If you're OK with that compromise, you can still get the kind of value you want (scouting information) in return for the resource you spent (a spell slot) without forcing the DM to drastically change his style of play.

*Not a compromise in the sense of necessarily being a worse experience, but in the sense of being an imperfect emulation.

  • \$\begingroup\$ I like where you're going with this, but have a question. How would this system work when it comes to surprises like "the enemy circled around you through an unexplored part of the dungeon that the Eye totally must have visited" ? \$\endgroup\$
    – Erik
    Aug 9, 2020 at 17:41
  • 4
    \$\begingroup\$ @Erik Ultimately, this solution relies on abstraction, and it isn't going to work exactly the same in the case of details like this that require full knowledge of the dungeon to resolve accurately. For this to work, the DM has to "play fair". If the DM knows that they're planning to have some enemies circle around like this, they could ask the Arcane Eyewitness whether they are scouting narrowly for a single path to their goal, or exploring all the branching paths, and adjust accordingly. \$\endgroup\$ Aug 9, 2020 at 17:49
  • 1
    \$\begingroup\$ @Erik to add to Ryan's comment, the amount that the DM has to "play fair" doesn't really change. Every spot that hasn't been carefully investigated can potentially hide a secret passage that allows enemies to circle around. This is true whether the dungeon has been scouted by a rouge, an arcane eye, or the party as whole. Afterall the eye doesn't automatically notice everything, no more than the wizard or rouge would in person. \$\endgroup\$
    – Ruse
    Aug 9, 2020 at 19:42

Be grateful! In 3e, the Arcane Eye spell only lasted for 1/min per level. Having an entire hour is a lot of extra time, unless you are playing at 60th level or so.

It sounds like keeping careful track of rounds while you explore would run counter to your freestyle play.

I have a minor homebrew suggestion that shouldn't be too excessive:

  • Allow the Eye to exist in an "inactive" state for a certain amount of time (6 hours, or 8, or even one day). Alternatively, allow the Eye to be switched to an "inactive" mode a specific number of times per casting, say 5 or 10 or something.
  • Only the "active" time counts against the Eye's 1 hour duration.
  • When first cast the Eye is always in an "active" state.
  • Minimum "active" time is 1 minute and usage is always counted in minutes even if it is only used for part of a given minute.
  • While "inactive", it is visible, immobile, and has to be carried around like a marble. Bonus points if it looks like an eye.
  • The Eye can only be switched active or inactive by the caster's touch.
  • If necessary, have the caster pay for the added functionality with an extra spell slot expended, of a level the DM feels appropriate, or raise the level of the spell by one.

Adjust these suggestions as would best fit your group's ideals and preferences.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .