649 réponses à ce sujet

[Tarot Redhand]

How are you going to deal with scree? As placeables, built into tiles or crossers? Or no scree at all?

 

TR

[s e n]

the good thing about using crossers for streams and such is, among other nice things, that you can click/drag the mouse to paint the crosser over a 3 tile combo and being able to place even if the two tiles needed for each of the two tiles crosser variations used to paint without the click/drag trick are missing. of course this has a con when you start to use the eraser and you get a lot of your crosser chain deleted or some wrong tile crosser variation appears because the needed tiles are missing but overall its a great feature for crossers not everyone knows it (i learned from lord of worms tilesets, not sure if any of the bioware ones has such "hidden" crosser variations)

 

on the other hand i am curious about the results md will accomplish with a terrain setup but it looks undoable to me, at least from a builder point of view going after the green light hovering on a very complex height built terrain. the strategy there should be shaping the rocky terrain along with the stream. and once you're done do not touch it anymore because it will too much difficult to adjust. crossers on the other hand require not many variations and are easy to paint

[MerricksDad]

Love your plans for the above.  You're even doing beetle-killed pines?  Those damn things went through our trees like a buzz saw.  Now, a horde of giant beetles to kill among the pines may just be therapeutic for their little cousins killing more than a hundred of our trees. 

 

When you do your final release, are you planning an optional "no trees" set?  It would work great for an above tree line set as well.

Definitely going to have a naked set and a shrubby set. That way the naked set can be used for whatever, including trees other than mine. The shrubby set will have all trees and plants parented by a dummy node you can then make script for which to swap model parts, basically letting you wipe or replace aspects of the tileset. I'll probably provide that script as well, and I'll probably do it both in MaxScript and as a command line tool, which you can specify a file containing model code for the replacement values. Shouldn't be too hard for command line. Super easy for maxscript.

 

This is what I'm doing for dead pines.

 

https://www.google.c...t/data=!3m1!1e3

 

And this is what I'm doing for burnt pines.

 

https://www.google.c...!7i13312!8i6656

[MerricksDad]

How are you going to deal with scree? As placeables, built into tiles or crossers? Or no scree at all?

 

TR

Scree in the igneous areas of the black hills comes in three forms.

 

The first form is granite with the feldspar eroded away, leaving chunks of quartz from small pebbles up to a few inches across, which may include mica disks, so it should have a certain sparkle. That will be represented by a combination of floor decals for flat areas, as well as 3d placeables for use on slopes. Coincidentally, this similar shape, but different color, is the same as crushed granite from mines.

 

In the area of Sunday Gulch, where water constantly removes and decays finer grains from the ground level of pegmatites, it creates a concave region at ground level, while covering the nearby ground with granola-like jagged pebbles. The remaining wall section is equally granola looking, but is larger grained. It often shows 3-4 inch chunks of pure minerals, with great crystal faces. The higher walls away from the water movement, and away from ground-covering snow and ice, are only weathered chemically by rainfall, and those take on a very smooth and greasy look. They erode millions of times slower, and are often glass-smooth, allowing you to look through quartz and clear feldspar crystals to what is inside the rock.

 

https://scontent-dfw...b1c&oe=56C524AD

 

 

IN the picture above, the black line represents the side profile of the sunday gulch granite. The red line indicates where the top of granite is cracked and pried apart by ice and tree roots. The ice-blue region is the part that is often glass-smooth. The outside blue-lavender line represents the usually lichen covered region. Purple is an area where under-erosion is about 50/50, and where mosses and dangly mineral growths can form. Brown is where the pegmatite is coming part in large jagged chunks. Gold is where the chunks fall and continue to decay. The green area is a safe place for gulch plants to grow, and is usually composed of mica flake and sand. Basically granite scree doesn't fall very far at all from the host rock, unless the undercut is so deep that the entire boulder the size of a house just simply falls off and cracks into multiple giant shards.

 

Check out this area of Sylvan Lake. It looks green in this google maps photo, but since it was wiped out by bugs over the last 10 years, there are almost no trees, and therefore all the pine needles have washed away. The ground is now gray, brown, and peach colored and looks more like an old mine. Both sides of this hill are now almost entirely bare, and the scree ranges from large rounded granite, down to less than a foot diameter jagged feldspar chunks, all packed tightly together with weathered quartz, and mica sand. When we were there, we found a hole in a granite wall, much like a house wall with a window cut out. I had never seen this before because it was hidden by trees. I suspect most people had not seen this rock in 50 years. Fascinating how things change so quickly.

https://www.google.c...t/data=!3m1!1e3

 

This was taken on the same spot when it was green. Now it looks like a sun-bleached wasteland, and all the trees are totally gone.

 

https://scontent-dfw...858157386_o.jpg

 

The second form is separated platelet schist. This comes in various sizes, and the majority of schist in the black hills is locally flat shale-looking stuff, but as hard as quartzite. These will be represented by placeables which lay flat, and jagged "boulders" of multiple layers of different sizes, pointing in different directions. Much of the schist in the black hills has layers of blood red and rust colored quartz which fills between certain layers, and is known to commonly be from the height of a pebble up to a few inches. Much of the schist in the keystone are is high in mica in one direction, high in rust in another direction, and dark blue gray in another direction. Whatever face is up is often covered in blue-green lichen. Faces pointing north are often covered in pumpkin colored lichen. I've constructed a texture which accounts for all six colors required to properly portray the schist over Harney Granite (lichen, pumpkin lichen, rusty dirt, bloody quartz, silver and gold mica, and slate blue). Because nearly all of the smaller plates lay flat, those can be displayed as decals on more flat surfaces, and since they often embed themselves in the topmost layer of the soil, and become partially covered with pine needles, that makes a perfect opportunity for those decals. The larger boulders fall off the uplifted sections and strike the ground at different angles, or otherwise roll as the soil changes around them. That means I need lots of big schist boulders which point in random angles. Since the schist chunks will land into soil, and be partially submerged depending on where you place them, I can just randomly rotate the boulders and you can choose to use any cardinal facing on the x/y plane to make one boulder look like 20+ depending on your placement. I figure I need 5 boulders of each of a few different sizes to round out the schist placeable selection, as well as about 10 decal textures. I was thinking to do the decal positions in two ways. You can use the flat decal placeable on flat parts of the map. Because there are so few flat places, and you won't see most of them properly if you place them on the sloped ground, I was thinking to also include pre-existing decal overlays on the tile. If you want an area which is high in schist, you include the texture in a hak or place it in your override. If you don't want the decal to show, you supply a transparent texture. Alternately, I can supply a texture which can be mirrored over Y to provide a yes or no for on-tile schist, where one half the texture is colored, and the other is transparent. For an idea of what schist scree might look like, check out google maps at this very special location:

https://www.google.c...t/data=!3m1!1e3

 

I've never been to that specific location, and didn't know about it until I was house shopping on that road.

 

Here is another common appearance of schist in the Keystone area. Note how the rock goes up hill toward the center of the granite dome. When portions are removed by weather in the direction of the dome, the rocks fall down into the hole and land at random angles, and stay in chunks as large as a house.

 

https://www.google.c...!7i13312!8i6656

 

https://www.google.c...h/data=!3m1!1e3

 

Here is an image showing a schist outcrop, which I intend to use to model some wall-like schist:

 

https://scontent-dfw...787&oe=56C82DBF

 

I've got hundreds of pictures of the area near that outcrop. That is just a tiny one.

 

The third type of scree is mine discard. This is almost entirely high quartz pegmatite, and comes in pink rose, stark white, 50% transparent white, high-lithium black, mica-rich silver, and high-potassium pink-orange. A single texture could easily cover all the bases here. Mine discard comes in size from pebble to chunks the size of a car, or larger if you count collapsed entrance chunks. I need to keep the size of chunks to a game-reasonable size, which is counter to what you'd see in real life. This means I need to make chunks small enough that the user will not constantly put them over a tile boundary, because that will lead to pathing false-positives, allowing the PC to walk through solid rock. I figure I can get away with 50 boulders in various sizes and colors. Half of those can be mine tailing piles and stacked stone walls, where the base is pebble sized texture, topped with larger rocks, and finished with medium rocks. For an example of the large-but-uniform scree in the black hills, check out the base of Mt. Rushmore.

 

https://www.google.c...t/data=!3m1!1e3

 

For a more realistic pile, check out various mines around the Keystone area.

 

https://www.google.c...t/data=!3m1!1e3

 

Other than schist, scree does not naturally form in the igneous sections of the black hills, but since the base rock of this set is igneous, and not the metamorphic schist, I need to not have it as a tile-based option, except in special conditions.

 

I won't be using this set for sediment-based mountains because the height of rises is wrong. If or when I get the limestone canyon set finished, the rocks there will be more square and jagged. Scree will be few inch thick plates of limestone, sandstone, and shale, rather than round boulders. Since the set will be based on the same rock which will make up the scree, the scree will be on the tile. Like the black hills reminiscent set in Granitelands, the limestone canyon set will not be a high tileset, as much as a low tileset with high edges, so you won't see places which represent high himalayan mountains completely covered in scree, like some stone carpet. You'll see very sharp, very high, not at all walkable, walls of 3-4 colors of limestone coming directly out of a floor tile, or a slightly raised tile. Scree will form in the concave corners of those tiles, while house-sized dropped sections will form on the convex corners of those tiles (or technically be on adjacent tiles). What I want to do in that set is keep a height transition of 150 and 300 and mix those, then supply a surround terrain type which is sheer cliffs. Since you won't be able to reach the surround terrain, I can make it very high. I can also supply a +1 on surround terrain without having to worry about walkmesh inconsistencies, or visual perfection/shadows/etc. With this I can portray both the Spearfish Canyon, with it's shallow rapid river and square shaped valley, or recolor it to portray the much deeper Black Canyon of the Gunnison, with it's more U-shaped valley, and much deeper river system.

 

https://www.google.c...t/data=!3m1!1e3

 

Here's pic of the pumpkin lichen on some granite chunks way up high

 

https://scontent-dfw...607&oe=56854F94

[MerricksDad]

I've been puzzling out the stream setups, and I get these three options:

 

One is of course common crosser practice, but with a twist. The streams would:

1. go from 0-0, as well as 1-0, as normal
2. between any two edges which are +0 or +1 in relation to each other
3. which will total 44 tiles, not including similar tiles where one or two corners are 2-3,2-4,or 3-4
4. plus another 32 for half active tiles
5. and would require extra 3-way stream forks on many tiles

This is fairly self explanatory, and only slightly increases the need for stream tiles over normal stream tile implementations. This would work as any normal crosser, going from midpoint of any edge, to the midpoint of another edge. It has the drawback of being square looking, even if implementation tries to be rounded.

 

The second setup works at a terrain type and would:

1. go from 0-0, 1-0, or 2-0
2. along edges or diagonals
3. where we'll call the lowest corner the base corner A, the diagonal from it point R, and the left and right corners from the base point we will call X and Y (it's a trig thing)
4. and will have an option for A-X when X<=R
5. and will have an option for A-Y when Y<=R
6. and will have an option for A-R when R<=X and R<=Y
7. and will have an option for X-Y when X,Y,and A are zero
8. which will total 31 tiles, not including similar tiles where one or two corners are 2-3,2-4,or 3-4
9. plus another 38 for half active tiles
10. and would not have stream forks

This should be fairly simple to implement, but I doubt I can pull it off via script without more work than is needed to just make the tiles. Since water would flow down only the lowest points of the tile, and would be at the edges, this is the most accurate option visually.

 

Instead of drawing from edge midpoint to another edge midpoint, this method allows you to plop down a stream connector terrain at any supported corner. From that corner, the stream will just kind of pool at the placement corner on all adjacent tiles. You then draw another stream connector at another corner of the same tile, which has to be supported by the system, and it will draw a stream from point to point. Streams flowing downhill will have slightly different and more active textures than flat terrain streams. Placeable decals will exist for flat terrain streams so you can specify a direction and speed of the water. In games where water texture does not translate, displaying a flow direction via texture animation, this is how newer games tell you which direction water is moving, by placing decals over the water mesh which animate in a direction, but only a little here and there, rather than the entire mesh. They let your brain imagine the rest.

 

This option is no more difficult than crosser-style stream placement. It simply uses corners instead. Compared to edge-to-edge connectivity, this corner-to-corner connectivity has the same possible number of options, but are just calculated in a different way. From any edge, you can traverse to 3 other edges. From any corner, you can traverse to 3 other corners. The only difference is the elevation decisions, making it so water never flows uphill, or choose a path which is more resistant over one which is easier.

 

The third option is similar but requires/allows a few more tiles, and therefore would require a lot more half active tiles. I detailed that one yesterday.

 

Of all the options, I prefer the second listed here, primarily because it looks more real. Second to that, I can perform stream forks at any supported corner, and it automatically can split up to 3 ways, as long as the adjacent tiles are supported. If additional tiles are calculated, it can split to 4, same with the crosser method. It naturally changes direction at either 90 or 135, and forks at 90 or 135, whereas normal crossers curve at only 90 and fork at 90. It may seem more rigid, but it actually offers more realistic options.

 

If the number of tiles is doubled, option 2 can portray a meandering stream which snakes right and left over the length of at tile, rather than flowing in a straight line. The same can be said of crosser type streams.

 

Option 2 also offers a different left and right side of the stream at any combination, where crosser methods require multiple varieties to achieve the same level of variation. Doubling the selection of stream models in option 2 quadruples the variety, because you have the options for AA, AB, BA, and BB stream side combination.

 

Below is an image which shows the difference between edge-to-edge (pink) and corner-to-corner (green) where the tiles are flat. Keep in mind that corner to corner can do things edge to edge cannot when used on mountainous areas. It also shows where green can apply variation, and where it cannot, in relation to pink. This shows that green cannot vary much on straight away sections, while pink can. Pink cannot vary on turns as much as green can on diagonals.

 

 

There is a 4th option for the future, but I won't bother calculating it now. That option combines a crosser and a corner option, allowing you to switch between edge-to-edge and corner-to-corner streams, which would allow you to make really sharp, or really smooth direction changes and splits, basically offering up to 5 realistic exit directions in relation to any one entry direction, whereas we now have 3:1.

[MerricksDad]

Here's an album showing some of the texture goals for the finished set

 

http://imgur.com/a/hF5nB

[MerricksDad]

Last night and this morning I spent some time wrapping a script around edge-to-edge stream creation. It has some issues. The first being that there are 419 tiles it picked out that need streams put on them. That does not include half-active tiles, or variants of such. That's a problem for me because I don't think that this many tiles is required for a good stream.

 

I found that I can easily automate streams on edges, but I can't seem to get gmax to play nice on streams on diagonals. Diagonals were going to be a huge part of the system, so that is really a bummer. The problem comes up when I try to carve one object with another. Problem one is that you cannot carve a paper-like object and expect good results. The carving object will basically rip a hole in it. So what you have to do is extrude the object and turn it into a 3d volume, and preferably have it closed to outside regions. This makes carving easy and error free. The second problem occurs when you try to then calculate which of the faces needs to be removed. I tried a short version where I paint the faces I don't want after carving to a material type which is useless on the tile, like sand or snow, expecting I can select by material id after carving and just delete those faces. Not true. The third problem is introduced when carving multi-material objects, because union, intersection, and difference functions on meshes also calculate the same for the material references. In doing this, the material handler in Gmax breaks and destroys the material-to-face setup on the mesh, making it so I can no longer delete those faces.

 

I may try using face smoothing groups instead, or otherwise paint the verts to be deleted with a vert color. Something might work.

 

But I'm still worried about that 419 tiles. This only included fully fleshed out stream tiles. For every one of those, for every one of those I need approximately 1/3 as many half-tile streams, where only one corner is "streamed". We're talking nearly a 50% increase in tiles over what this set has so far, just to include a single stream type. I don't think that is worth it.

 

So to stop this nonsense, and just move forward, I'm contracting the scope of streams. I'll report back as I get the data sorted out.

 

Test 1:

restricted water fall from 2 levels to just 1: 373 tiles

 

Test 2:

water falls from 1 only, restricted water from moving into a lake or stream via normal tile (with future stream-to-water tiles being added as groups): 87 tiles

 

Test 3:

water falls from 2 and 1 as before, but no stream-to-water connection: 121

 

Test 4:

edge-to-edge instead of corner-to-corner, water only falls from level 1, no stream-to-water: 200

 

Test 5:

edge-to-edge, water falls only x1, stream shares no space with hills except 0001, no stream to water: 10 tiles

half-tiles to complete this set: 17

additional tiles to connect water: 20 and 40

total super basic crosser stream: 87

 

Test 6:

streams done as single tile groups in the form of edge-to-edge streams, without using a crosser: 30 tiles, no half-tiles, and includes connecting to water

 

Test 7:

corner-to-corner, water falls only x1, diagonals only on flat tiles, no stream to water: 72

 

Needless to say: I am not happy with the results so far. If I have to do any large quantity of physical work on this, streams won't be getting done before winter. If I can fully automate it, I will go for highest quality, even if that bloats the tileset x2.

[s e n]

can't you just do the 0011 0022 0033 0044 values for the stream terrain and work out the diagonals for the height change with a few groups?

[MerricksDad]

I could, but I'm looking for a stream complexity which matches the tileset complexity, as well as puts the stream in the correct location, which standard crosser types won't do.

 

I started working on the automation script about two hours ago, and have it almost done. With it I should be able to fully automate the stream creation process at the highest quality options by about two days from now (to test bugs).

 

The only bug in the script at the moment is some random face deletion when trying to delete faces around the edges. It is picking up faces in the center along the new diagonal cut. Not sure how that works, especially since the faces in the deletion list are not what is shown. There are no modifiers in the stack, so it should be getting only the faces on the mesh base. Once I figure that out, I can stream cut any tile with the click of a button.

 

Edit:

 

seems to be a bug in meshop.deleteFaces when you use the delIsoVerts:true option after doing a bunch of vert movement. The verts set to faces is apparently stored in two places, and this function uses the wrong group. By deleting the verts left behind by deleted faces, it selects verts recently added instead, even after doing an update call on the mesh. I may try using the other mesh within the model.

 

Edit:

 

fix found. setting delIsoVerts to false, and then immediately calling meshop.deleteIsoVerts on the same object does the trick as required.

[MerricksDad]

[MerricksDad]

I called it a night on this to think of some more fixes. Something is wrong in my math when detecting the tile position in the z axis. I can currently process all level 0 to 1 tiles. Anything higher gets mangled and the stream gets shoved up the center.

 

I need to spend some time not in the house tomorrow. I'm starting to feel stuffed in a tiny box and winter is coming. Before the weekend, I may have this streams package ready to export, and we can see how it works and go over the differences in placement (though I already did above). It will be interesting to see this new idea come to life.

 

Trying to figure this out, I went through a butt load of files and a stack of paper, and two pencils. But I like being challenged.

 

The image you see above is the diagonal stream cut for tile 0111. It has been processed with the streamCut script and then processed with the mountainBuilder script to noise it up. I may change the mountain builder script when used on streams to leave more of the stream center untouched. It is a simple matter of using some new functions I wrote to fetch all faces within a specified distance of the nearest point where x=y or -x=y. On a top-down graph, that makes the diagonal baseline slope top-left to bottom right or bottom-right to top-left. I'm building this library of true/false face checking functions which attach to this other function called getFacesMatchingPattern where you specify the object or specific object mesh, the face, the function you are using to match a pattern, and up to three variables you can specify which get passed to the match function. Here's a list of what I have so far (not complete):

 

getIsFaceHorizontal

getIsFacePointingMostlyNorth/South/East/West/Up/Down

getIsFaceNormalCardinalDirection

getIsFaceYNearX

getIsFaceYNearNegativeX

getIsFaceNormalEqualTo

getIsFaceAtTileBoundary

getIsFaceMatID

getIsFaceUnderWater

getIsPointInFace

getFaceRayIntersection

getIsFacePointingMostlyDirection

getIsFaceVisible/InvisibleRelativeToVector

[MerricksDad]

This morning I got the diagonal builder to function correctly, as long as I use it on a direction the river can actually flow without mangling the tile. I tried it on tile 0022, attempting to make a diagonal from upper left 0 to bottom left 2, and it just crashed GMAX really hard. I had previously run it on 0011, and it worked ok. It made the river edge kind of puffy on level 0.

 

It works perfectly on tiles which should be able to be cut diagonally, like 0110 and 0111, as well as any combination height diagonals.

 

I even set it up for the function to accept river cut depth. At 100cm, the river has a lot of breaks in it, which could be textured as stones in the river, allowing you to assume the water is really shallow and just flows through the stones where you cannot see it. Gravel texture or jagged rocks would look good here.

 

At 150cm, the river is mostly clear of debris. Some debris seems to be really angular, and if it spans the entire river, it can look funny, so I may need to hand-edit 1/10 of the processed tiles on average.

 

At 200cm, the river is entirely clear, but takes on a shape which is so uniform it looks like a person cut it into the stone. For my purposes, that just looks weird.

 

The processes can be run in two directions. I can first cut the river into the tile prototype, and then apply a mountain texture to the whole thing. Or I can run the mountain builder script, and then follow with the stream cutter. If I do it that way, the stream is flat bottomed and forms a half-wave gradual slope by using this formula:

 

y=(-(cot (pi*x)) + 1)/2, where x is equal to distance ratio across the tile on the diagonal. It then applies the y output to the mesh z axis, and scales that number by the given height transition, offsetting by Sen's data for tile offsets.

 

This automatically calculates nearly the same slope you would have with the old Rural Soften crosser, making the slope of the stream really gradual, but not in a way that is straight from corner to corner (which I tried and it looked nasty).

 

I think that same formula may have some other uses for me in the future.

 

For the sake of having a more clear stream with less long angular parts inside, I may run the functions backward, mountain first and then stream cutter, and modify the stream cutter to simply add noise as it positions verts. This will give a more U-shaped cut, like those in OC Rural. I need to find a middle ground.

[MerricksDad]

Not much going on this next week here, or for a few months now really. Monday I'd like to finish the walkmesh to pathnode detector. Right now, the previous version detected pathing only based on the name and crossers, which worked well for me with a lower tile count, and with tiles not created using height changing noise.

 

For this next version, I'd like to detect basic pathnodes by actually walking the walkmesh faces of the mesh. To do  [AHIJLNVWXY] I will just check if a) I can get from any given edge to another, and from that edge I can reach ALL other faces on that edge. Doing [BCDEFGKMOWRSUab] should be almost as easy. I just need to figure out the best way to detect that there are two groups of walkable exits on the same side, and only two. If I can detect two, it should be almost as easy to find three, and then I can do [cde]. The key is in doing [Z] correctly, because it mixes both edge to edge travel of any kind, but needs to know that two edges have two exits.

 

I may have to break each of the pathnodes down into its constituents, counting each separately and calculating a pathnode from the parts. For instance, [c] is calculated from 1) a path from one edge to an adjacent + 2) another separate path from one edge to adjacent on the same edges as the previous + 3.) a path from any corner to any corner which is separate from the other two paths.

 

Another example is [F], where the pathnode is made from 1) two exits on one edge + 2) two exits on the opposite edge + 3.) a path that can go anywhere

 

If I have to do all that work, I can also detect in the process tiles which don't fit any of the pathing, or could somehow fit more than one description. I almost guarantee that a detection script like this could throw out some interesting information, especially on faces up against an edge which are walkable on the side of a cliff, like you can get in this granitelands set.

[MerricksDad]

I've spent some time in Drakensang Online, trying to get some more ideas for stream stuff, or even just water. What I found instead was a really cool way of building the same shaped tileset over and over again, but with different visuals.

 

From what I have studied so far, this game uses the following in every tileset:

 

Terrain A

Terrain B

+1 (both)

+2 (both)

Solid (impassible rock, or a building [universal])

Chasm (universal)

Water (universal)

 

If you combine these parts, making solid/chasm/water all universal height, as if terrain level 0 will always attach to water the same way, and +1 and +2 just happen to get a water corner, then the total number of unique tiles without duplicates for variation is just over 1027.

 

Now, you can't portray a good mountain with that, unless you specifically set the height transition higher, or make some specific groups to transition faster, so that is generally incompatible with my granitelands set.

 

What you can do is something I have wanted to do for quite some time, and that is a dual terrain crypt/cavern with pits, water, and solid space to make that enclosed feel.

 

Another thing they do with the same tile combos is create a swamp/forest which feels enclosed by trees or chasm.

 

Their fog falloff is set really low, and they add additional fogging to the Z-axis, so when you look down pits, you don't see very far at all. In terms of NWN distance, their fog falloff on the Z-axis is about 600cm. That is far too low to use in game unless you have a very foggy place. Instead, I'd keep the global fog falloff higher and just add some fog layers to pits. Another alternative would be to have a blanket fog on ALL tiles with a height transition. That might look pretty cool.

 

I had originally wanted to have really super high places, or seemingly deep, yet transitional, drop-offs in my upcoming underdark sets. What I may do instead is a combo terrain style for each of the three. The karst caves was going to be the first I tinker with, and that was going to be primarily limestone pocket caves, long tunnels, and spherical/dome rooms. That would fit very well with my crypt/cavern combo, and I can keep the required tiles under 2k by omitting the other two height transitions.

 

Another thing I can do to reduce tile count in a set like that is to restrict transition change to only +1, but create groups which create a +2 transition for me in other ways. That would be far less work, but would not look as good.

 

For those of you who have played Torchlight 2 or Dungeon Siege 1, you can get an idea of the cave/crypt combos I am planning. You could also take a look at some of the first crypt areas on Drakensang Online.

 

If you do go view Drakensang, you'll also see that the quantity of tiles for just one crypt using the one tileset is probably about 2-3k, just because of the variants, especially on stairs. They also seem to have a "corridor" style crosser.

 

Another game I have been examining lately is Dungeons 2, which is an upgrade to Dungeon Keeper.

[Verilazic]

I played Dungeon Siege (and loved it, for what it was), and I recall most of the caves there essentially being the same as outdoors bounded by tall "mountains" but with different lighting, with the possible exceptions of the goblins base and the final seck prison. Is that the aspect you're referring to?

[MerricksDad]

DS1 has that multi-level effect, but they also transitioned from crypt to catacomb in a single tile. In the first regions of DS1, multiplayer or campaign mode, there was that skeleton filled crypt, but sometimes to get to another portion of crypt, they led you through a section of gray-brown cavern. There was a similar setup later in the mines, when instead of just being in that brown mined stone, you transitioned to dwarven halls. I don't think I will have the dwarven halls be connected on my second underdark tileset, but you will find some dwarven works out in the "wild" areas, as well as door-tiles to dwarven halls.

 

In Drakensang, they built their crypt the same way, occasionally moving you into a cavern area. It is kind of built in a way that makes you feel like the whole crypt is crumbling into the underworld. They even have animations to rumble the screen, and then play some VFX with debris falling from the ceiling. Other tiles have bridges over chasms with big cave formations growing up around the edges, as if the crypts were constructed in the caverns on purpose. In many places, the floor has just caved in to a pit, sometimes with water below, sometimes just bottomless.

 

There are parts of the crypt tiles which are very much rip offs of Diablo 3, with book cases which break very similar. Those would have to be done with damageable placeables, positioned perfectly with the tile. What they appear to be doing, in terms of NWN mechanics, is having a niche into which one placeable goes. When it takes damage, that transitions to another appearance, but also gives off parts which eventually fade away.

 

Some of these tiles might make a good mini-set for this month's wizard tower CCC.

[Tarot Redhand]

What I loved about DS1 was there were no obvious area transitions. You could walk from one end to the other with no obvious waits, at least once you killed everything. That was the other thing I loved about that game - no re-spawning monsters.

 

As far as the actual areas go, don't forget the cellars that had lifts and the haunted crypt right at the start that was full of spiders.

 

TR

[MerricksDad]

That was definitely a plus: the no transition wait. After playing zone-based MMO's for years, it was very refreshing to have tile-based maps which simply led right into the next region. I loved how in the multiplayer world map, you could go from your start area to ALL the areas by simply following a different path.

[Tarot Redhand]

In fact the only thing not to like is that it doesn't work for any version of windows after XP. Oh, and I couldn't find all the sp easter eggs either.

 

TR

[Verilazic]

I got it to work on Vista. Though I only remember that it was a bit of a pain to do so.

[MerricksDad]

I got it to work on 7 without trying. I haven't tried 10 yet.

[Tarot Redhand]

It refuses to install for me (vista & 7). Says it won't work. That is one of the reasons why I have a dual boot (XP/win 7) PC. FYI, I have it on disk not a digital download (if one exists).

 

TR

[Tonden_Ockay]

Hey I seen this youtube video and thought of your (MerrickDad) project

 

 

AnOldFriend (handle/name on nexusmods) is the one that made these textures and said the following about his work.

 

 

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
IMPORTANT NOTE: I am happy to allow anybody to use my work in your mods. Anything I make, and release, can be considered a modders resource. 

However, this only applies to those who are providing the work freely. You may NOT use my work in any mod that you intend to accept donations for. I do not work so you can profit off my efforts. Please respect my wishes.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

 

I plan on uploading all his work (not just the mountain textures) to the vault soon.

[MerricksDad]

I would certainly use those!

[Tonden_Ockay]

Just uploaded aof_skyrim_textures.rar to the Skyrim Textures project on the vault. It has AOF Detailed Mountain Textures.