649 réponses à ce sujet

[MerricksDad]

A bit unrelated to the black hills tileset, but I was trying to get back interest in water features. While playing some Drakensang Online, I was sketching out all it takes to make their wave-filled ocean scenes. I think I got it down, and it comes out pretty good. Just needs a little tweaking, and then pick out an enmap, and I should be able to make a water that waves, sparkles, slides, and animates all at one time, using only two medium poly planes, and about 10 frames.

[KlatchainCoffee]

...  wave-filled ocean scenes. ...

 

Yes please.

 

There is the moving sea water from TNO and that sea travel placeable stuff done by W0rm and some others...   But anything that would make the ocean or the beach surf (or even waves smashing against rocks) more real....   

 

 

By the way, I finally downloaded the granite hills tileset and played around with it. As any WIP, it's rough around the edges, but I really think it has the makings of one of the best tilesets out there. It's inspiring to see you work in general, but I hope you will complete this one and make into something truly amazing.

[MerricksDad]

I really really want to. Starting possibly tonight, I will mostly be snowed in, until I start school in January, so it has that going for it.

 

The water that has already been done is similar to the one I'm working on. From what I remember of that one, I did like it, but found it fairly plain. I also have a rocking sea tileset planned. What It will be is an ocean that actually moves around the few tiles you can actually stand on. Your perspective from the ship will stay fixed as you walk around on deck, but the entire area around you will rock and roll around, churning as 50 ft waves crash around. Because we can create an entire area on a single tile, this should be fairly easy, but because any single tile can only export so much going-on per frame using nwmax animesh, it can be broken into a tile group instead.

 

So this waves thing right now is a placeable I am testing, and you might see it in the wizard's tower package I put out this month for CCC. it has a 5x5 wave grid. When half go up, the other half go down. On top of that, anything that goes up also has the texture shifted toward the point, so it looks like the foam on top churns to the peaks of the waves, or stretches as it rolls through the trough. Below that, a second animated mesh slides by at 45 degrees on the x/y plane, causing texture noise and more watery look. That texture will eventually have an environment map set, and will also use TXI to animate just the alpha channel. Hopefully that will fully duplicate the waves from Drakensang.

 

The model I have now has a simple up and down motion, but does not crest in a sine wave, and looks jerky. I just need to triple the frames and create the sine effect in both the vertical motion, and the texture bunching. Even now, it looks pretty close to what they did. I just wish I had a texture as good as theirs.

[MerricksDad]

They've got a second type of wave which is just a 2D wave directed at the shoreline. It hits straight on, so looks a little gimpy. It comes out about like the one in Dungeon Siege I around the black-sand lakes. I much prefer the ones around the ocean beaches that come in at an angle. I duplicated the black-sand beaches from DS1 years ago and still have the models around somewhere. It used animated bones instead of animesh, so I'd like to convert it to animesh and reuse it.

 

The straight-to-beach waves in drakensang have about a 20 cm wave height maximum, and are very transparent, which leads me to believe by the colors shown, that there are about three planes of water stacked on top of each other which complete the deep water look. It comes across looking more like there is a fog below the topmost water plane, and that might actually be what they did somehow. Not sure.

 

Anyway, I much prefer to work on the 3d wave noise than the 2d wave at the moment. I even smoothed it off by applying two wave modifiers to it at right angles in gmax. Wave mods alone were too plain, so I stacked them with the animated manipulation structures.

 

I need to fine a setup that lets me create GIF videos really easily.

[MerricksDad]

So last night I ran into this issue I should have remembered when playing with OTR's envmapped animesh.  Basically on mine, I can't see it. I'm sure it's working, because OTR's stuff works in the video. But I can't see it to test it on mine. So no shiny animated water for me at all ever it seems. Not on this kind of machine anyway. I guess I will offer the animated mesh stack, and just let somebody else play with the rest. It won't do me any good as I had intended it.

[MerricksDad]

Due to the visual impossibilities of my animated water attempt, at least on my computer, I went back to working on the mountain streams, just so I don't get lost in stacks of other crap which will eventually make me not finish this project.

 

I specifically went back to working on the corner to corner stream maker scripts becuase I really really really want to see what that would look like for real on that mountain setting, and then judge with a proper example if that is the way I want to go. Running around the deep gullies on the testing module I definitely feel that is the only logical way to go, because edge to edge streams simply cannot go in that place, so it will not look like a real mountain stream cascading down a slope. As I mentioned a few pages back they have their own issues, but I think I can work past those and ignore the extra load of required tiles.

 

After weeks of not working on this trig, I'm having a problem wrapping my brain (double caffeinated brain) around a negative formula for the diagonal stream. Just so you can see what I'm up against, here is the diagonal stream height formula for top left to bottom right:

 

y=((-cos (radiansToDegrees (pi*x)))+1.0)/2.0 ..... haha. No seriously.

 

My double caffeinated brain is easily distracted when christmas music is playing, and a child is ignoring that in addition to constantly begging his mama for help with a game he is too young for. "Mama mama mama mama mama" at a very low yet high pitched tone. It is no wonder that I cannot simply figure out exactly what to replace -cos with, or if that is even going to work for a topleft to bottomright formula.

 

I suppose I will put this off a little longer and hope for the best in the two days I have next week to myself.

 

I've also been examining the recently uploaded Skyrim textures Tonden_Ockay put up, and I'm seeing a different way of splitting my meshes which does not have so many visible seams at the cliff edges. While I purposely want some of those angles between smoothing groups, I don't want most of them. As OTR pointed out a few pages back, they probably look better to most players smoothed off. I've noticed that smoothing them off invites another host of issues into the visuals, so I've gone to a midground on that issue, rounding many, but leaving important ones unrounded so that more flow can be had from tile to tile, while using the noise of the rest to hide where tiles actually start and end.

 

The Skyrim method uses a majority of partially transparent meshes with a single solid mesh under all that. I had been trying to figure out exactly how NWN2 accomplished similar, but had never taken the time to really get into the guts of the NWN2 maps. Now I can also see in Drakensang exactly how they duplicate the methods used in Skyrim. If I were to mimic that method, it would greatly increase the poly count of my models, perhaps even doubling it, and that is only if I take it easy and stay simply with two textures. Either way, if walkmesh grass is disabled, I think it would look far more spectacular than what I have now. It would also give me the opportunity to do the grass-rim meshes differently, as well as other rim-type meshes, including water edges. So I'd really like to give that a shot.

 

What I'll do is modify the in-progress-script which chooses a mesh family for faces at border terrain types, like those you can see make sharp triangles at the edge of cliffs in this set. What I had intended to so is have the script force those faces to choose one texture or another, without doing the random mixing along the edge. This would benefit the meshes by creating a hard texture seam along the edge. Originally I thought the opposite was much better, but now I have more options, and so don't actually need to hide the edge in low-tech ways.

 

What I'll then do is clone off anything to be textured with grass, instead of splitting it at the seam. I'll then extrude the outermost edge(s) and apply a texture with a transparent boundary. I'll then keep anything which was supposed to be rock texture, plus one more face over the rock-grass border. That extra face will stay tucked under the grass with the transparent edge. The underlying rock texture can remain a solid nontransparent texture. All other faces from the original mesh will be deleted to reduce duplication. That should only increase the face count by about 10-20% if I am guessing well at all, but should remove almost all hard texture transitions from the model. Really looking forward to trying that now that I understand it better from the Skyrim textures, and can see it in action in Drakensang.

[Tonden_Ockay]

That sounds great MerricksDad I can't wait to see it when your done.

[Tonden_Ockay]

Sorry I just seen that when I uploaded AOF Skyrim Textures I uploaded them as .dds Skyrim files. So I just did an update and uploaded them again as .tga files so they are now fully usable with NWN1.

[MerricksDad]

No problem. I used GIMP to convert.

[MerricksDad]

I've been trying to come to terms with the idea that I might not finish this, as much I want to, until maybe next summer. That kind of bums me out...well not just kind of bums me out, it really kills me. In the meantime, I've been working on a few smaller projects in what time I have had. I started asking for help in maxscript forums worldwide so I could figure out the quat math issues I've been having with the spellfire arm sheath. In the process, I figured out a way to use the wing node to possibly do other interesting wearable weapons. I don't really like using nodes with other specific purposes for such things though, as it requires that the wings of certain characters be placed on the main model, or the robe, and just creates more work.

 

In the process, I've come to the realization that if I had simply worked on the animation 100% manually, it would have been done by now, especially since I realized I was only going to be able to do combat cycles on the OC models. Maybe I should have a team of people as advisors who can calculate the potential hours for a project before I spend my life trying to train a robot to do that for me.

 

While I wait for the needed help on the arm sheath, I'm examining the stream idea again, especially the corner to corner one. I'm really liking how it turns out when I make them by hand. The hard part is getting enough time to myself so I can teach the robot how to do the other 1000 tiles in the set. The holiday season seems to be the worst for such tasks.

 

The edge seam fixer looks to be working as well. Instead of fixing the edge as the tile is made, I went with an alternative where I can retroactively fix any of the premade tiles. This lets me also fix the separate stream tiles later without having to reconstruct the whole subset script again. It simply works hard to create a continuous line of texture from one edge to another depending on the tile. As planned, it leaves some of the grass patches mid-tile away from the primary seam.

 

In addition, I'm working on a script which does all the multi-texture blending I need to really make these tiles flow like I wanted. I've been spending more time in games like Drakensang, really looking hard at those tiles they use while my slow internet connection struggles to visualize all the mesh parts. I guess having a slow ass computer is somewhat beneficial at times.

 

I've also been really studying water features in a few games. If any of you played with the CCC package for wizard's tower last month, you will have found a sea water prototype I jammed in the corner of the room. Ultimately, that single mesh was supposed to blend animesh with envmapping, which I know some people can see. Since I cannot, it makes it really hard on my mind to continue working on that. Instead I'm taking an entirely more basic approach. Some future sea water tiles will include a minimum of 3 animated meshes.

 

The first mesh, the one you see in the wizards' tower, will have a sea foam appearance, but be otherwise mostly transparent. The underlying mesh will have the same physical animation, but the texture will slide instead of bunch and part. Together with a sea floor mesh, these two will portray clear water, like from the carribean. A fourth mesh placed at a specific depth can represent deepening darkness as you leave the shoreline, and will make use of two textures. The first is solid color and shares an opacity across the entire plane. The other is for use against the beach where it fades from that opacity to 100% transparency as it climbs the beach up and out of the water.

 

The third animated mesh is for objects like rocks which are under water. Drakensang uses a different method, which is actually a visual mistake in the calculation. Basically when a solid texture is behind a transparent one, it tends to modify the opacity of the above texture somehow. They left it because it looks good. NWN tends to paint underwater rock a different texture, but with an unmoving edge. With moving water, you need a moving texture edge to portray what is under water and what is above.

 

I sat yesterday for about 10 minutes at the edge of a river so I could get a feel for what is actually happening, and I've joined that with the math mistake in Drakensang to create a darkening texture which animates in the same wave pattern as the water meshes. Because it relies on the specific tile it is associated with, I could never use this on placeables. Basically it moves the darkening mesh in a way which matches the water level, changing the appearance of underwater rocks to look very realistically wet, and darkened by the water volume blocking the ambient light.

 

I don't yet know what kind of load a large area of this will put on the system, so I don't yet know if this will only be useful for small areas, or if it can be used for fantastic playable settings. That is something I should find out soon really.

 

I'm also collecting a set of placeable-like animated decals for use on river surfaces. Instead of making a very complex texture, I can simply modify a few patches here and there, making the river look turbulent or foamy in various places to denote it's direction of flow. This will come in handy when I install the 8m wide river tiles. These are specifically different from the corner to corner stream I'm still working on. In all, there will be 4 water features in the mountain set. You already have the prototype for shallow and deep water packed within the set. You'll also have the two non-crosser stream and river tiles. The 8m river will only naturally fall 1 height change without additional group features. That will keep the tile requirements for that specific tile type down to about 30 or less. I'll be able to cut them in the exact same way I am cutting the corner to corner streams, with only slight modification to the script.

[3RavensMore]

I've been trying to come to terms with the idea that I might not finish this, as much I want to, until maybe next summer. That kind of bums me out...well not just kind of bums me out, it really kills me.

 

Don't sweat it MD.  Life happens, and we all do that with big projects.  If the process becomes a death march, it's not going to be fun anymore.  Take you time; I think everyone here will just be happy to have it, even if it takes more time--maybe even a lot more time--for it to be delivered into our eager paws.

 

Take care, and happy holidays. 

[MerricksDad]

Nifty little video of the StreamCutter2 script in action, followed by the DoBuildMountains script. I cut out the delay time while mountain building processes. It seems to be taking longer today to do a single tile than ever before. Maybe I have too many other things running.

 

If you can't tell form the picture, this is tile 0111, not a flat tile. The stream flows down to the upper left corner from the bottom right corner. The script also does diagonal the other direction, as well as steeper elevation changes. The stream follows a stretched sin wave slope, instead of falling sharply, or at a perfect incline. This allows for better transition at the bottom and top tiles.

 

 

The hard part is fully out of the way. I just need to work out a few bugs. One of those is that the down-right diagonal cut tinkers with the wrong corners and moves a few verts it should not. It took me most of the morning to detect why the slope was wrong, and that is the only part left I need to fix.

 

The edge cuts will be fairly simple to write. They require only a single cut, and I can leave out over half of the auto-fix steps because there is no extrusion, therefore no vertical faces, and no fudging numbers at corners.

 

You can see that the mountain builder fills the streams with walkable rock levels. This is random and can sometimes put a big ass rock right in the stream, fully blocking it off. If it happens at the very top or very bottom of the stream, there is a big vertical wall which looks really gimpy.

 

In the last few frames, you can see me playing with the water level. The script initially sets it to a level which represents what most streams in the black hills look like about 2 hours after a storm. If you ever see the black hills putting out as much water as in the last frame, you should leave the area, because you might not be living much longer. When the battle creek fills up, it rarely crests the bank. It did this spring though, just due to snow melt, and flooded many of the yards in the main path down through Keystone. One woman I talked to had said in her entire life of living there, she had never seen a year like that. She wasn't around during the reservoir break, so this would have been nothing in comparison.

[MerricksDad]

I don't have anything to show for it yet, but I have also started putting together a small script kit based on recent mountain building functions I made this year. It basically grows a cave system in the same way as MineCraft, using voxels. It then smashes some of the voxels into smaller ones, removing a random quantity of the divisions which are not fully enclosed. Then it removes all the inward pointing faces, and shared faces, leaving only the outermost mesh walls, like a perfect hull system (just like I always wanted for my tree builder).

 

After that I plan to have it smooth and optimize and all that good stuff. So far, it accidentally makes tiny arches, and some weirdly stacked rock piles.

 

What I really want is a texture preserving optimize script. I just don't understand how to best play with tverts yet. My study of using normal vectors in texture application hasn't really given me what I need to fully automate the manual effects of ChiliSkinner, and I'm not sure that is really what I want anyway.

 

While playing with the tree builder again, I determined that I really need a way to just map all the textures based on cylindrical centers and vectors from that center. But I need to be able to do that on every branch section separately. The same would work for the voxels. The only alternative I see is to pre-texture all the cubes and branches, weld them together with a custom optimizing script, and hope for the best.

 

This is about all it can do right now.

http://i28.tinypic.com/vrykqu.png

 

What I want it to do is this.

http://i.imgur.com/0XjOs.png

 

But the requires a different approach to get the multi-level openings.

[MerricksDad]

If I work on this further this year, I will probably try to include a bubble volume method, instead of systematic stacking and smashing. If I use volume, I can create a noise pattern in 3D which tells me where to put rock volume. I just need to make sure it doesn't leave parts hanging in mid-air... unless I want those. I might be able to physically drop chunks to the ground if they become separated, by detecting face centers on their hull, and then doing a collision test against floor mesh hull faces. I could even rotate the mass before dropping it.

[MerricksDad]

 

So today, I finally received some advice on the spellfire arm sheath issue. It appears that I need to rotate on the local z-axis after I set the facing, or otherwise merge the 2-dimension vector to rotation with the original local z-axis rotation. Now I just gotta find out exactly at which point to do the combo, or how best to do it. I've found a few issues with my method, and now in trying to fix it and merge these new parts, I have screwed up one, two, or three sections of my bone-reversing code. It isn't back to scratch, or even nearly, but it feels like a job now. At least I understand a bit more about 3d rotation in vector defined coordinate systems. Still waiting to see if anybody has an exact fix for my code out there. That would just be spectacular if somebody else finished my work

 

I should have put more work into finishing the stream tiles instead. I dreamt two nights ago about how to best pick and build the single corner stream varieties. At first I was going to omit them, or otherwise use streamless copies of normal tiles and just keep those as placeholders. But then I thought builders might actually want actual stream ends to just end on that tile. They'll look different than OC streams, so I didn't think they were important. I also never liked that without a group/feature tile. But then I thought about all the useful placeables that could cover the endpoint, and decided to just make them anyway.

Basically speaking, what I'm going to do with stream endpoints is only allow an endpoint on a tile which could also have a stream on the entire edge. So if a tile could have one edge with a stream, I need two varieties for stream ends on that same tile. Two edges means 3 end varieties. I'm not making tiles for three edge streams because this system has no room for parallel streams, or U-shapes on a hill. I will still do three edge streams on some flat tiles though. I don't have a figure for the additional tiles this will require, but it is probably quite a few given the +1 and +2 height transitions I am allowing streams to cover. I'm guessing more than double what the actual stream tiles were going to be. I think I listed that above in my testing phase.

I don't have a pic handy, but the smooth texture edge builder is working very easily. That was the simplest thing to make in a while. It also lets me make those easier transitions to modern texture application, like those in Skyrim and Drakensang, especially using round-texture decals, like OC rock piles, dirt patches, and grass splotches.

Now that the foliage is all down, I've also been out taking angle measurements on evergreens for my tree builder script. I hope to be putting that to use shortly when I decorate the newly textured base tiles.

[MerricksDad]

Two coffees today and I have the basic stream sections, and the code for building them, 100% done. Now to just build the picker for half-edge stream ends and I'm done with streams except for special group tiles.

 

Well almost. I have noise on the z-axis, but no noise in the length of the stream. I think I can create some stream bend tiles mechanically instead of by hand, but really, I could do it manually at this point and do it faster. I guess knowing which job is quicker is part of the job.

[MerricksDad]

If anybody is interested, Sword Coast Legends on steam is 50% off this weekend, and free to play until Sunday. I pulled it down yesterday to try it. It took me all night to install it, and isn't as fun as I had hoped. It looks like Baldur's gate and the plot runs a lot like OC NWN2. The gameplay is about as bad as Pool of Radiance, and it seems to have as many bugs.

 

I ended up turning down all the graphics so I could actually get plot text faster. It also greatly reduced battle-to-conversation loading times. Not sure why they picked that method, but it seriously takes away from the action of the game (not that it has a lot).

 

Neither high end or low end graphics has anything I was interested in using to base new NWN models from. The gameplay videos before it came out actually look a lot better than the game actually is.

 

Still, if you feel like slug-crawlling through the linear plot, or building with it, definitely go check it out.

 

My number one complaint is that the whole thing moves at one time. The trees, the bushes, and sometimes even the camera. It just makes you sea sick to play it. If each tree moved individually to wind, and bushes shook randomly or in location-driven gusts, it would be not only more realistic, but something a bit better at keeping your food down.

[Kendaric Varkellen]

Just wait until you give their editor a try...

 

While it's absolutely user-friendly (anyone can build with it), it's also horribly limited in functionality.

[Jedijax]

Not to kick it while its down, but... reviews were very low on that one. To be honest, I don't think I've seen a good D&D game since NWN and maybe NWN2. Dragon Age should have been it but... well, you know... franchise keeping is costly and we got that and Knights of the Mass Effect.

Close

[DM Frayne]

Well this tileset is about as cool as it gets.

 

My Native American themed mod is set in a fantasy variant of North & South Dakota with a bit of Montana and Wyoming thrown in so this would be fantastic for the highland bits.

 

Best of luck finishing it.

[MerricksDad]

Still going strong, but much slower now. Still very hopeful.

[YeoldeFog]

I think it's time for a nice screenshot MD! *waiting*

[MerricksDad]

So over Christmas, I purchased and played, all the way through, Sword Coast Legends. It had some parts I was really interested in, but mostly for the visuals. The gameplay seemed to get better as it went along, and it only crashed about 7 times without reason, and once with an error log. Overall, I guess that is better than Pool of Radiance, by far.

 

I was really interested in the underdark areas, because their Forlorn Cliffs region had a lot of what I want to incorporate into my middle and deeper dark areas, with the bridges and chasms, as well as the open-seeming spaces away from the main walkable area, or otherwise down inside chasms. That region, as well as some of the other cavern areas, have some false perspective sections that I found very interesting. At one location, you look down into a hole and see a miniature tunnel and bridge over a stream. The distance down is less than a tile width, yet the false perspective gives the feeling that it is at least 10x deeper. A larger region of such would look far more impressive, because as it stands, sometimes you wonder if what you are looking at is simply a cross-tunnel for smaller species.

 

After playing all the way through, I started checking out the DM area builder. As mentioned, it is fairly user friendly, but has little to no customization available. When building "dungeon" type environments, I enjoyed the room-to-room feel, and can see how building 1x1 or 2x1, or even 3x3 room sections would make for a very nice tileset, especially when the only other terrain you might have access to in the editor is the corridor crosser. It left no room for height changes in itself, but if converted over to NWN style maps, it could very easily lend itself to such.

I found the "area" style areas not at all spectacular. As much as I enjoy the Forlorn Cliffs area, as well as some of the other larger areas, the inability to customize the area layout AT ALL was maddening. Combined with the inability to zoom out in such a way you could view the entire area, it makes overland-style maps nearly impossible to like.

 

There is definite potential for much of this to be converted over to NWN, and I'd really like to try my hand at some of the cave systems they use. It reminds me of the stuff I was doing with hexagonal mapping, which didn't really turn out. Together though, everything I've got already, and what I'm seeing in this game, possibly gives me the ability to create something like this:



...just as long as the walk-under parts are filled and unwalkable, or the walk-over parts are broken or somehow unreachable.

[OldTimeRadio]

I love the pic of that diorama.  You know, I don't think there's anything stopping you from one-offing a little setting like that, but maybe where the upper portions of the tile(s) lean away from the lower ones (to accomodate the walkmesh) and then force the camera into a side-scroller view.  Would love to see some old fashioned parallax scrolling in NWN.

[MerricksDad]

For "dungeon" type settings, where there is no skybox, I was trying to figure out a good ratio to trick the viewer into thinking short distance was much longer. Having seen at least one way of doing it in SCL, I was also thinking to do a vertical bowing, based on some artwork I was using for crags, which come from a set of Lovecraftian concept art for something (not sure what). If I can assume that the maximum height in an area is 140 meters, and the maximum depth is 30 m (maybe it is 25 based on my exploration in the badlands tileset), perhaps I can create a set of room-like tiles which bend inward as they go up. Using two types of "surround" tile, I can display a panoramic deeper area, as well as a very hard enclosing surface around, and even into, other walkable tiles.

While trying to flesh this out this morning, I came up with these terrains:

 

Chasm - a base terrain with nothing but stalagmites and strewn boulders in a pit. Instead of EVER reaching the bottom, you'd traverse a series of bridges from tower to tower, with occasional islands ringing a larger -tite-mite column. The column an also serve as stairs to a higher level which connects at this point.

Chasm Bridge Crosser - as mentioned above, this bridge passes from column to column, or touches a few taller stalagmite mounds from below. Tower parts within the bridge system can offer ceiling regions visible within the visible depth of the environment.

 

Cave/Solid rock - a base terrain of enclosed unwalkable region, into which you place corridors or prefab cave rooms. One or two of the cave rooms should open into other terrain types.

Cave Corridor (type A: thin tunnel) - This thin tunnel represents mine shaft quality tunnels between dungeon style rooms

 

Cave Corridor (type B:  wide tunnel) - This wide short tunnel represents an expanse of natural cavern and has a nice flat bottom made of hard-packed debris. Additional placeables could be built to turn any given wide tunnel into a partially collapsed cave run with fallen cave-clouds or other formations which have come loose form the ceiling.

 

Cave Corridor (type C: lower chasm) - This thin tunnel, with no visible ceiling, represents the bottom of a chasm, or deep drainage cleft in the surrounding rock. The walls could come in some really interesting shapes which block 1st person, or top-down view at many points. This corridor could also be used as a height transition change if T-tile are created with a ledge system.

 

Cave Rooms - A great number of cavern rooms, both natural, or man-made grottoes, should be created in the size range from 1x1 for closet spaces, 2x2 for common combat locations or treasure filled places, 2x1 for unique transitional halls, 2x4 for unique transitional rooms, 3x3 for smaller boss rooms and great features, such as camps, or 5x5 for greater boss rooms and fantastic locations. Additional 3x3 rooms can bridge the transition to other areas, such as up and out, down a spiraling tube, great stairwells, or portal locations.

 

Surround Wall - an unwalkable terrain which leans into adjacent tiles using a hyperbolic curve, fully blocking view and access to adjacent areas, or off-map regions. surround wall can also lend a ceiling mass which can be seen, even in the darkness of a cavern.

 

Surround Chasm - a very deep region that trails off into darkness. The expanse is so wide that you cannot make out any features of another side. Down below, there may be mushrooms or other things visible. Everything at a distance should be down-scaled to exaggerate distance. This terrain should only be placed near chasm terrain, so that the player can never look directly at the scale offset in an adjacent tile.

 

Dome Cavern - a base tile which can go inside the solid cave tile, or out inside the chasm base terrain. This region easily lends itself to simple height transitions, or 2 at once, without getting in the way of any other transition types. This is also the best place to create building structures, or open markets.

 

Dark Lake - A water terrain type which you cannot (and would not want to) enter. This would be available to a flat walkable terrain, such as the dome cavern, only.

 

There is no need for water within the dungeon areas, and no need for stream type crossers, as the room-based areas can have them built in already.

Since everything will be corridor-connected, doors, secret doors, and open hallways will need to have standardized door placements.

Given the quantity of tiles available in all SCL tilesets, there is enough variety to give a great base to this kind of underdark dungeon system. All the art I have downloaded over the last two years related to caves, as well as 3d maps and panoramic views of actual cave rooms (from national park service stuff), will give an almost unlimited supply of new tile groups which can be released in packages later, after the base is available.