@ Roguelike games

Well, almost made it. Got about halfway through zot 5 when I got mobbed. Was one step away from the stairs to save myself when I was smited by a balrog and a guardian finished me off.

Been playing a bunch of Brogue lately. Has a really nice display (originally for Mac OS X terminal output widget, now using libtcod for portability) using 24-bit color foreground/background character cells to illustrate lighting conditions beautifully. The overall mechanics and UI are well streamlined. It has a nice small set of verbs (unifying "eat", "quaff", "zap", etc. all as "apply / activate") which reduces the need for remember strange edge-cases.

Even with the reduced variety of each kind of thing (less stats, less verbs, less item types, less monster types), it does a great job of having each play-through feel unique.

Overall, a very impressive distillation of the core ideas of roguelike play. I've been thinking of working on doing some roguelike development myself, and this stripped-down core feels like a great foundation for new development on a "modern" roguelike model.

Worth checking out if you haven't seen it yet.

IMAGE(http://sites.google.com/site/broguegame/home/screenshots/Fungal-forest.png)

So I've been thinking about building something vaguely roguelike, or perhaps more dwarffort-like. It all started when I was thinking "I bet there's a better way to do liquid simulation than toady's doing. I wish I could try", and then railing against fate because the DF code is closed. (Although I can totally understand why that is.) A bunch of my thoughts from that time then mixed with thoughts I had when playing with Minecraft. And now I've been doing a lot of reading of roguelike development stuff, procedural content generation stuff, etc.

Anyway, I have some ideas and questions I'd like to get opinions on. Right now, particularly about the nature of the world map.

I'm committed to a world which is defined in terms of 3d cells on a 2d surface. That is: You can dig and build up and down directly, and spaces connect to each other in obvious ways (i.e. like DF's level model, not like a more traditional roguelike level model).

I'd very much like a world that is continuous (i.e. not having to choose a "play area" like in DF dwarf mode) and very large.

One option would be to do dynamic generation on the cartesian plane, like Minecraft does (but perhaps more efficient in terms of storage). This would allow infinite (or near-infinite) exploration and growth in all directions, with regularly spaced grid cells. Such a world can be described as "infinite and unbounded". Infinite: there is no limit to the area defined by the map. Unbounded: If you start going in a direction, you can keep going forever--you'll never run into an impassable boundary.

Another option for such a design would be something like the spherical grids described in June on Dungeon League. Having a spherical world is pretty amazingly awesome, and there are a number of techniques that become interesting if the world is finite but unbounded. You still can keep going forever in any direction, although you'll eventually loop back to cover territory you've seen before. That's because there is in fact a finite area defined by the map--the area of the surface of a sphere of some radius r. The finiteness is appealing on the UI side because it makes it easier to take in the features of the world in one go: a mechanism could allow zooming around and looking at everything, looking for interesting features. Or, you could discard the world and make a new one if you didn't see anything you like. On the coding side, it would allow for certain map generation techniques that require global knowledge of the map. (With an infinite area, every technique has to be applied to just part of the map, although the parts can be quite large.)

--

Okay.... So. I talked about all of that because of the question I have:

I could either go with a finite cartesian plane, or with a sphere. If I were to go with a sphere, the best shape for grid cells would be *hexagons* (or mostly hexagons, some pentagons would appear with a tiling as described on that site I mentioned above). Of course, hexagons would also work on a cartesian plane. Squares might be *made* to work on a sphere, although it would either result in massive distortions at certain points, or it could be confusing as the game would have to adjust the frame of reference of the grid periodically as you move (although at large intervals.) (i.e. it's not possible to use the same square grid projection at the equator as at the poles, so at some point or points in between it would be necessary to "shift points of view"--suddenly that last grid cell you left would be divided among multiple different grid cells.)

Note that a spherical hex (with a few pentagon) grid would have points of distortion as well, they'd just be much less painful.

Opinion time: Is the "awesome factor" worth having to go to a totally graphical view of the grid, even if the cells are drawn as text characters on filled backgrounds? (i.e. the map would be of slightly deformed hexagons and pentagons with "@", "#", and friends in their centers). Arbitrary rotation of the map would be allowed (i.e. "choose which way is up"), although the drawn characters would always be oriented in the normal manner in their cells.

Second line of thought: Even with a flat geometry (i.e. the infinite cartesian plane, or the flat toroid, or the flat bounded rectangle like DF), it would be possible to do hexagons. Here, the problem is simpler, since the hexagons would be regular instead of being deformed and mixed with very rare pentagons. The display for this sort of map could be done using, say, double-width character cells (as is sometimes done for Kanji characters and "full width" characters for Japanese text) which are offset by an extra character cell on every other row. Like:

_AABB
CC**FF
_DDEE

I would probably want to render the characters in the center of their cells in this case, because "full width roman" characters in Japanese fonts look like ass. But, it would still allow for the possibility of large ideographic characters to be used for some features on the map, which could be cool.

So, second question: If I wanted to punt on the globe thing and go with a more traditional character cell based system, do people think that this approach to using a hex grid would be interesting? Can you imagine it being attractive?

Or, do people think it would be better to just stick with a standard square grid-cell map rather than dealing with either hex grids or with the extreme possibility of deformed grids on a sphere?

--

Final notes: My working concept for gameplay is something along the lines of: "You are one of the ancient dwarven explorer-kings, who has sat slumbering alone on his throne for centuries, dreaming strange dreams. Now a group of your descendants has discovered your vault and awoken you to lead them to greatness." Think in terms of a building and exploration game like DF, but one in which you are playing an actual individual character, albeit one with a special relationship to your people. The idea here is that your citizens might be a bit more independently-minded and hard to control than in Dwarf Fortress, but at the same time you would have the ability to travel around and perform tasks yourself. Most likely you would be much better doing any individual task than one of the lesser dwarfs of this later more corrupt age, but if you can get all of them working together for you then the overall ability of the group to get things done would be better than what you can do alone (and having the civilization working smoothly without your frequent direct intervention would free you up to pursue other goals: exploring like a good explorer-king should.)

/HEAD ASPLODE

Seriously though, a finite, unbounded sphere with hexagonal (and occasional pentagonal) tiles would be awesome.

Okay. I'll keep thinking in that direction, then. Thanks.

If you need testers at some point even just to help test basic functionality (i.e. the "unfun alpha stage"), I'm sure you'll find several in this thread (myself included) willing to help out.

Heh. It'll have to get out of "concept and... hey, that's a shiny algorithm!" stage first.

You could also code in a higher/rarer resource value for the Pentagonal tiles to encourage exploration or encourage competition for them.

Farscry wrote:

If you need testers at some point even just to help test basic functionality (i.e. the "unfun alpha stage"), I'm sure you'll find several in this thread (myself included) willing to help out. :)

Hear, hear!

Adjacency Computation wrote:

Depth 23.
Surface area: 703,687,441,776,642 cells.
Equatorial circumference: 41,943,040 cells.

Sample cell adjacency:
A35: Northwest=E140737479966755, Northeast=A36, East=A16777252, Southeast=A16777251, Southwest=A34, West=E140737479966754.

Computed adjacency lists for 10,000,000,000 cells in 375.87s: 37.59ns per cell.

This cell names are basically an integer index within a "strip" of triangular icosahedron faces running from the south pole to the north pole. So the letter goes from A-E, and in this case the number goes from 0 to ludicrous. The letter-number pair is just... er... "UI" for a simple number, though. Only, yeah. Doesn't quite work at this scale. When it's like A0-A128, it's a little nicer.

Because I can work with this numeric IDs for each individual cell and compute which cells are adjacent on the fly, I don't have to do anything silly like compute a list of which cells are adjacent to which other cells for the whole world. (Which is good, because with the configuration described above--a planet approximately the size of Earth with each cell being approximately one meter in radius--the amount of memory required just to remember the names of the cells would be a bit much.

Next step: Computing the position of any cell on the surface of the sphere given its cell index.

koshnika: Er. No matter what the resolution at which I generate the grid, there are exactly 12 pentagons. In the Earth-like 1m resolution example above, that means there are precisely 58,640,620,148,052.5 hexagons for every pentagon. Even with a much more reasonable resolution, the pentagons would have to be kind of ridiculously valuable to be anything special. ;>

Implemented my first version of the "given a cell index, what are its coordinates in 3-space?" function, so that I can actually start thinking about how to display things. The implementation is way too slow—at depth 23, it's taking 0.6ms to calculate the location of a cell. (Contrast with the aforementioned 40ns to calculate the adjacency list.) That may not sound like much, but it means that it would take almost 10 seconds to calculate the locations of every cell in a 128x128 grid. :X

So—why so slow? I think I may have a bug in my implementation which is contributing one big chunk. In addition, I'm creating a lot of objects as intermediate results, and this is in Java, so creating objects is about the slowest possible thing I could be doing. I think I have a way to avoid the temporary object creation, and hopefully if there's another bug in the implementation that's making it do a lot of extra work, I'll find that when I'm implementing the new method of computation.

Excellent. I have the coordinate calculation stuff working well now, and I've produced some images:

IMAGE(http://lh6.ggpht.com/_eoPgF1wybxs/TJUq-f0f4OI/AAAAAAAAAEQ/bw6AczZFV3Q/s128/x-grid-3.png)

Of course, those are all global views. I figure a global view is what you'll use just after world creation to look around for a spot you want to start playing. When in actual play, you're always going to be looking at a very small sub-part of that sphere, so the world will be effectively flat from your POV, and tiled with some slightly warped hexagons.

I've also started a blog thingy at j-prevost.blogspot.com. That'll be moving to hypatian.org once I get some name server stuff worked out. I'm working on documenting the stuff I did to make the hex grid work right now. Then I'll probably move into terrain generation next.

Current question for folks:

One thing I've considered is that an Earth-sized world is way way way too big for a player to be able actually find anything interesting in it. What do folks think about that? Would it better to play on an Earth-sized world with far more area than you will ever see, even if you played it for your whole life, or to play on a much smaller world which is a lot less realistic (as a world), but which would be a lot easier to explore?

I'd say MMO-sized world, one that you can go around in a matter of a few hours. I have no idea on how this all works, but that way you can perhaps pre-generate some interesting stuff, instead of infinite random sameness a la minecraft?

Very interesting project

Hypatian wrote:

Current question for folks:

One thing I've considered is that an Earth-sized world is way way way too big for a player to be able actually find anything interesting in it. What do folks think about that? Would it better to play on an Earth-sized world with far more area than you will ever see, even if you played it for your whole life, or to play on a much smaller world which is a lot less realistic (as a world), but which would be a lot easier to explore?

I guess that depends.
Look at the actual earth for reference. How many truly boring places are there? Even Antarctica has seals and penguins and flying Great White sharks. The Sahara has a lot of wildlife if you know where to look.

If you can generate a world and save the seed, then you could be allowed to start in *that* world from a different locale. Possibly even have options like when starting a Civ4 game so the player could pick a climate type in which to start.

You're talking about exploration, so I think bigger is better. Then again, I'm not a completionist. I'd rather be able to explore "endlessly" than have mind-bogglingly abrupt transitions between "zones" or environments. What's the word? Biomes? I forget.

Anyway, it really depends on the type of game and how much it is targeted towards the Explorer type.

duckilama: Well, my worry for the "explorer" type is that if the world is truly Earth-sized and you're playing at human-scale, places like Antarctica and the Sahara might be distinct and interesting... BUT, if you started play in the middle of the Sahara it could easily take you a long long time before you saw anything *except* the Sahara. That might be mitigated with things that allow you to travel more quickly, or an ability to move to a "larger" scale and do travel there with time moving faster... but both of those kind of break down my desire for the world to be truly seamless at the human level.

(On the scale of the Sahara: in the short direction (N/S) it is 1.8 million meters across. Unless there's some other kind of features that appear with an unrealistically high density for such a harsh environment, I'd say that something like that is effectively an impassable barrier at the personal scale of play.)

django: I definitely want to avoid the "infinite sameness" that is present in the current Minecraft. I know that biomes are coming to Minecraft soon, which should decrease the level of "sameness", but I want to go a step beyond that and have biomes and geographic features that go together in reasonably logical ways. As an example: an area of low rainfall should appear downwind from a high mountainous region. I'd also like the map to not be unpopulated (although that's quite a distance out), and population centers should appear where they make sense (i.e. places with natural resources, where water and transportation are available.)

Oh, and:

duckilama wrote:

If you can generate a world and save the seed, then you could be allowed to start in *that* world from a different locale. Possibly even have options like when starting a Civ4 game so the player could pick a climate type in which to start.

Yeah, that kind of thing is definitely on the plate. I'm just worried about the difficulty of locating a place in the world that is both playable and interesting, whether it's the player selecting a location or the player choosing some constraints and asking the game to find a location.

Hrm. I guess my first order of business is producing a basic terrain map and putting a "guy" on it that you can move around, and seeing how it feels. Things like world size and feature density are clearly parameters that can be tuned depending on how it works out.

Hypatian wrote:

duckilama: Well, my worry for the "explorer" type is that if the world is truly Earth-sized and you're playing at human-scale, places like Antarctica and the Sahara might be distinct and interesting... BUT, if you started play in the middle of the Sahara it could easily take you a long long time before you saw anything *except* the Sahara. That might be mitigated with things that allow you to travel more quickly, or an ability to move to a "larger" scale and do travel there with time moving faster... but both of those kind of break down my desire for the world to be truly seamless at the human level.

Faster travel is equivalent to smaller world. Pick one, they are the same thing.

(On the scale of the Sahara: in the short direction (N/S) it is 1.8 million meters across. Unless there's some other kind of features that appear with an unrealistically high density for such a harsh environment, I'd say that something like that is effectively an impassable barrier at the personal scale of play.)

Oh, and:

duckilama wrote:

If you can generate a world and save the seed, then you could be allowed to start in *that* world from a different locale. Possibly even have options like when starting a Civ4 game so the player could pick a climate type in which to start.

Yeah, that kind of thing is definitely on the plate. I'm just worried about the difficulty of locating a place in the world that is both playable and interesting, whether it's the player selecting a location or the player choosing some constraints and asking the game to find a location.

I would suggest ensuring that the engine not start a player "in the middle" of any (harsh) environment. Maybe have a threshold setting and ensure that the player always starts within, say, 2 minutes of an environmental boundary. Or within sight of a secondary environment. Antarctica is the only one that would break the rule, so just don't allow it in its Earth form.

Just some thoughts. Take what you like, leave what you don't. This message will not self-destruct in 5 seconds.

Could go with the Dwarven Fortress method: generate the world, let the player have access to a list of biomes and general danger ratings as they scan for a starting location, then off they go. I would LOVE to play a full roguelike using the DF world generation format (I don't really dig the DF roguelike mode that much; haven't tried it in a while though).

duckilama wrote:

Faster travel is equivalent to smaller world. Pick one, they are the same thing.

This is only true up to a point, and depends very strongly on what you mean by "smaller world". If you scale linear distances down by a factor of 10, that means it takes 1/10th as long to get anywhere. But it also means that all of the features are 1/10th as large linearly, and 1/100th as large in area. So suddenly your cities only have room for 1/100th as many people as they did before, your mineral deposits only contain 1/100th as many resources as they did before (or 1/1000th, if you've scaled down the z-axis as well.)

If you speed up travel by 10x but leave distances the same, you can now zip over the ground at ludicrous speed (or perhaps set a waypoint for where you want to go and then fade to black to get there in zero *play* time, although game-world time may have advanced), but the features are just as large as they used to be. Your city still has room for as many people. Your rivers and mountains still look impressive, etc.

There's also a third alternative, which is to scale down the world while leaving the number of features the same. The idea here is to "squeeze" the less interesting places until they're just big enough to make you think "Oh, big desert!" without actually taking days of play-time to cross. The distance between cities or mineral deposits has gotten smaller, but they're just as large once you get there.

I would suggest ensuring that the engine not start a player "in the middle" of any (harsh) environment. Maybe have a threshold setting and ensure that the player always starts within, say, 2 minutes of an environmental boundary. Or within sight of a secondary environment. Antarctica is the only one that would break the rule, so just don't allow it in its Earth form.

Okay, let's take a different tack here. Imagine a world the size of the Earth, with bronze age population. The player starts by looking at the globe from space, at a scale where you can see a whole hemisphere on your screen at a time. They have the ability to zoom to any scale between that and a 100x100 meter area, and they can zip around and rotate as much as they want, freely.

How long will it take them to find a place they want to set up a new settlement? How long will it take them to find a place that they want to set up and which provides the resources they'll need to get started?

The answer is unclear to me. That's one reason I wonder about the "feature density" idea—making the world smaller without reducing the number of things in it. By doing so, I think the task of finding an interesting and playable location gets easier, because there are many more interesting and playable locations in the world, and many fewer locations that are either not interesting or not playable or both.

Anyway, feedback acquired. Now let's see if I can make something that produces some maps, and then I can figure out how tunable it can be. It would be cool if things can be set up at world creation time to choose exactly which of these varieties of worlds you want to play with—and then you can give the new player a default of "quite small globe, rich in features", and allow the crazy man or experienced player the ability to tune out to "Earth-sized globe, realistic feature frequency".

Hypatian, you are blowing my mind.

I thought I'd mention in this thread that Nippon Ichi Software has a roguelike inspired (a la Shiren or Mystery Dungeon) game coming out in October for the PSP called Z.H.P.: Unlosing Ranger vs. Darkdeath Evilman. That's Zettai Hero Project, in case you were wondering.

There's not a whole lot of info on the game as of yet, but I'm interested purely because it's by some of the guys who did Disgaea and I rather enjoy NIS's humor.

The game is probably not going to appeal to many people in this thread, but since there's not many commercial roguelikes out there I figured it worth a mention. Tom Chick did a writeup up at Fidgit. He also has the first trailer there.

Question of the day: Alongside getting my first map-generation stuff going, I've got to get my first walk-around-on-the-world stuff going. That should be a lot easier. Of course, if this was on the Cartesian plane, I'd be done already. It's going to be tricky.

The question is: So I'm here looking at a hex grid. With a Cartesian hex grid, movement keys are easy—you can either go only nw-w-sw-ne-e-se or you can only go nw-n-ne-se-s-sw. But here I'm looking at a map where each hex is slightly warped from the previous one.

The effect should be much less apparent on the meter scale (I'm working on generating an image of that now), but this picture will give an idea of the problem:

IMAGE(http://lh3.ggpht.com/_eoPgF1wybxs/TJUrAnCfZyI/AAAAAAAAAEU/40EFlPpQSJA/s128/x-grid-4.png)

Now, the gross effects you see here will spread thin on the local scale—the same variation in orientation that crosses a fifth of the globe will still take a fifth of the globe to happen at the 1m level, even if that means it's going to take 8 million steps to get there. However, that curve is eventually going to have an impact. More importantly, there can be fast-ish changes on a smaller scale—specifically around each of the 12 "red" pentagonal cells, the distortion is maximized.

That's easy to display on the screen effectively, and I think the user will be able to follow things okay. One way I've considered keeping things moderately sane is to keep "north" always at the top of the screen (or center if the NP is actually in view, or center the SP if the SP is actually in view.) Probably with some stuff in place so that it doesn't rotate crazily with every step (think of the common roguelike move of re-centering when you approach the side of the screen rather than moving with every step).

I have a number of things I can try there, but they all raise the question of "how should movement keys interact with that?"

For example: Let's say I somehow manage to keep a static viewpoint, never doing any crazy rotations. That means that as the player travels long distances, north ("up" on screen) might change from being a direction you can't travel (the point of a hexagon) to being a direction you can travel (the flat face of a hexagon), while straight east and west become impossible.

If you look at hexes around the top middle red pentagon in the image, you'll see what I mean at its most extreme—however the same shift in hexagon directions will occur periodically for any latitude outside of 60ºN–60ºS.

So: Any ideas on the best way to work this? One possibility is to try to keep rotation to a minimum, use the standard 8 movement keys with two being locked off at any time (although the two might shift). Another possibility is to always shift rotation to keep a set of six keys (say UL-L-DL,UR-R-DR) meaningful, even if that means not always having "north" be the same direction on-screen.

Probably I'll just have to implement things and try them out and see what feels right, but I thought I'd solicit input.

Hypatian wrote:

One possibility is to try to keep rotation to a minimum, use the standard 8 movement keys with two being locked off at any time (although the two might shift).

My gut instinct is that this would be the most intuitive choice.

I've been playing Dungeon Crawl Stone Soup again and must admit that the game can be quite addicting. If you can get past the graphics, it is the next best thing to Diablo2 hardcore. The depth and complexity of the game is surprising.

In any case, I've been working on playing a wizard and finally got myself a respectable build. Level 14 wizard with a concentration on Fire and Poison. All kinds of resists. Several artifacts. I'm actually feeling confident for the first time ever. I've cleared the orc mines, the hive and am standing at the stairs to descend to level 8 of the lair. I am envisioning myself actually holding aloft my first rune. What a moment of triumph it will be!

Suddenly I am informed that I have been banished to the Abyss. WTF?!? Why? After doing some research, I figure out that basically its just bad luck and that I've got maybe a 1 in a 1,000 chance of actually finding a way out before I get my ass chewed by all the demons there. Soon I am dead.

Look, I like a game that makes things difficult. Makes you work and learn ... but don't banish my ass to the Abyss with no chance of survival. The game is difficult enough without jacking around with me. Suddenly I'm not so sure I really want to play this game anymore. What a cruel mistress.

Huh.. I didn't think any Lair dwellers were capable of Banishment. I could see maybe a Zot trap pulling it off though. I haaaateee those things. But then if it wasn't for those and teleport traps, I'd never bother raising Traps and Doors.

Abyss banishment does suck, horribly though. I've only succeeded in escaping once out of maybe a dozen times I've ended up there.

Clearly the game knew you were feeling confident, which is not how one should ever feel in that game. Ever.

If this thread is revived from it's "winter hibernation". I'll just mention I started playing URW (Unreal world) again (after playing Minecraft) . It's a rogue-like survival game with Finn/Scandinavian theme. It's shareware but I got a lifetime registration . (maybe I should get a version as a gift to someone like the author asked)

The new version(3.12 if i remember correctly) improved a whole lot which may cause the game to be a bit harder:

- The map looks like the Scandinavian peninsula
- The map is populated in what seems to be historical location of where the different tribes lived in.
- There is a lot less "magic" in the crafting process. The item crafting requirements are a bit more logical. For exampleif you create a pit trap you'll also need the,slender trunks and branches to hold the cover which are made out of spruce twigs.
- Deconstruction of buildings - real nice feature that was missing in the past.
- The animals leave tracks and you can follow them zoomed in. They also tend to run really fast which makes a bow almost critical for hunting unless you are trapping.
- There are also many types of axes that are best tools for certain tasks (not a key feature if you ask me)
- You have to use move command to drag logs and trees around . This is more time consuming than put a tree trunk in your inventory then dropping it but it's more logical than walking around with a tree trunks in your pocket ;).

So far I started a new game in Spring . I'm building a 5X4~ house(40-50% done) for the winter( I worked my companions 24/7 cutting trees) . I already got tons of food in my cellar ( I hope it won't spoil) after a very successful hunting with a companion NPC . I killed an elk, a stag, and 2 forest reign deers in about 4 game days. It took me about a game month or 2 to find that many animals on my own. It could be just luck or the season. I hope I'll finish my house before winter. I also need to stock up on some firewood .

Winter in this game can be deadly because other than starve and get sick you can die from exposure. Ice fishing might sound like fun but I noticed that the yield of fishing have been significantly reduced . I think the Escaped slave in winter scenario is probably impossible to survive (harder to escape , no equipment ,lower fishing yield which is critical for initial food). In other seasons (provided you can get away) you can get food from foraging and some fishing ( you don't have to break the ice and keep building a fire so you won't freeze).

The game I'm playing now isn't my first game in this version. In the first one I traded my pants (more useful) for an axe so I only had a pair of shoes (well half a pair because I used it for bandages). Then after I built a cellar to preserve my food I made a spiked pit trap right by it. You can guess what happens when you accidentally skinny dip into that pit .

garion333 wrote:

Clearly the game knew you were feeling confident, which is not how one should ever feel in that game. Ever.

Yes, thank you for that slap in the face. You of course are correct.

I've moved on and am trying my hand at a spriggan enchanter, which my research tells me is the 'easiest' combo to win with. We'll see what happens.

Well, the Spriggan Enchanter is the next best thing to hitting the 'I Win' button.

I'm up to Level 25 and have secured my three runes. I have cleared The Lair, The Swamp, The Shoals, The Orc Mines, the Elven Halls and the Vault. I even stumbled upon a Zot trap and got banished to the Abyss again. This time around my speed and cunning saved me and after about 30 minutes of dodging and wandering I found the portal out.

I may pass up The Crypt, Hell, Pandemonium, Etc. I am on dungeon level 22 and, from what I understand, need to make my way down to L27 where the runes will gain me entrance into Zot. I am doing my best not to be over-confident but the trinity of destruction (invisibility, stealth, stabbing) has allowed me to open up a lot of baddies as if they were pillowcases. Throw in some extra tools like might, haste, confusion, repel missile and ice armor and I"m a 3-foot tall version of the grim reaper.

Good luck! Are you playing on an online server or local? You could use a demo Camtasia install to record your last few levels.

Look, I like a game that makes things difficult. Makes you work and learn ... but don't banish my ass to the Abyss with no chance of survival.

Magic resistance, as much as possible, will help keep you out of the Abyss... as well as awake, unconfused, etc.

i downloaded the tile/asci installer of stone soup and click on it, windows asks if i trust this file and i say yes but nothing happens. i'm running vista64, does it just not work on that?