minetest-mm/mods/biome_lib/API.txt
2021-05-16 11:33:49 +02:00

656 lines
29 KiB
Text

This document describes the Plantlife mod API.
Last revision: 2021-04-20
=========
Functions
=========
There are two main functions defined by this mod:
biome_lib.register_active_spawner()
biome_lib.register_on_generate()
There are also several internal, helper functions that can be called if so
desired, but they are not really intended for use by other mods and may change
at any time. They are briefly described below these main functions, but see
init.lua for details.
Most functions in biome_lib are either declared locally or kept within its
own namespace to avoid collisions/conflicts with other mods.
=====
biome_lib.register_active_spawner(biome)
biome_lib.register_active_spawner(sdelay, splant, sradius, schance, ssurface, savoid)
This first function is an ABM-based spawner function originally created as
part of Ironzorg's flowers mod. It has since been largely extended and
expanded. There are two ways to call this function: You can either pass it
several individual string and number parameters to use the legacy interface,
or you can pass a single biome definition as a table, with all of your options
spelled out nicely. This is the preferred method.
When used with the legacy interface, you must specify the parameters exactly
in order, with the first five being mandatory (even if some are set to nil),
and the last one being optional:
sdelay: The value passed to the ABM's interval parameter, in seconds.
splant: The node name of the item to spawn (e.g.
"flowers:flower_rose"). A plant will of course only be
spawned if the node about to be replaced is air.
sradius: Don't spawn within this many nodes of the avoid items
mentioned below. If set to nil, this check is skipped.
schance: The value passed to the ABM's chance parameter, normally in
the 10-100 range (1-in-X chance of operating on a given node)
ssurface: String with the name of the node on which to spawn the plant
in question, such as "default:sand" or
"default:dirt_with_grass". It is not recommended to put air,
stone, or plain dirt here if you can use some other node, as
doing so will cause the engine to process potentially large
numbers of such nodes when deciding when to execute the ABM
and where it should operate.
savoid: Table with a list of groups and/or node names to avoid when
spawning the plant, such as {"group:flowers", "default:tree"}.
When passed a table as the argument, and thus using the modern calling method,
you must pass a number of arguments in the form of an ordinary keyed-value
table. Below is a list of everything supported by this function:
biome = {
spawn_plants = something, -- [*] String or table; see below.
spawn_delay = number, -- same as sdelay, above.
spawn_chance = number, -- same as schance, above.
spawn_surfaces = {table}, -- List of node names on which the plants
-- should be spawned. As with the single-node "ssurface"
-- option in the legacy API, you should not put stone, air,
-- etc. here.
---- From here down are a number of optional parameters. You will
---- most likely want to use at least some of these to limit how and
---- where your objects are spawned.
label = string, -- set this to identify the ABM for Minetest's profiler
avoid_nodes = {table}, -- same meaning as savoid, above
avoid_radius = num, -- same as sradius
seed_diff = num, -- The Perlin seed difference value passed to the
-- minetest.get_perlin() function. Used along with
-- the global Perlin controls below to create the
-- "biome" in which the plants will spawn. Defaults
-- to 0 if not provided.
light_min = num, -- Minimum amount of light necessary to make a plant
-- spawn. Defaults to 0.
light_max = num, -- Maximum amount of light needed to spawn. Defaults
-- to the engine's MAX_LIGHT value of 14.
neighbors = {table}, -- List of neighboring nodes that need to be
-- immediately next to the node the plant is about to
-- spawn on. Can also be a string with a single node
-- name. It is both passed to the ABM as the
-- "neighbors" parameter, and is used to manually
-- check the adjacent nodes. It only takes one of
-- these for the spawn routine to mark the target as
-- spawnable. Defaults to nil (ignored).
ncount = num, -- There must be at least this many of the above
-- neighbors in the eight spaces immediately
-- surrounding the node the plant is about to spawn on
-- for it to happen. If not provided, this check is
-- disabled.
facedir = num, -- The value passed to the param2 variable when adding
-- the node to the map. Defaults to 0. Be sure that
-- the value you use here (and the range thereof) is
-- appropriate for the type of node you're spawning.
random_facedir = {table}, -- If set, the table should contain two values.
-- If they're both provided, the spawned plant will be
-- given a random facedir value in the range specified
-- by these two numbers. Overrides the facedir
-- parameter above, if it exists. Use {0,3} if you
-- want the full range for wallmounted nodes, or {2,5}
-- for most everything else, or any other pair of
-- numbers appropriate for the node you want to spawn.
depth_max = num, -- If the object spawns on top of a water source, the
-- water must be at most this deep. Defaults to 1.
min_elevation = num, -- Surface must be at this altitude or higher to
-- spawn at all. Defaults to -16 meters.
max_elevation = num, -- Surface must be no higher than this altitude.
-- Defaults to +48.
near_nodes = {table}, -- List of nodes that must be somewhere in the
-- vicinity in order for the plant to spawn. Can also
-- be a string with a single node name. If not
-- provided, this check is disabled.
near_nodes_size = num, -- How large of an area to check for the above
-- node. Specifically, this checks a flat, horizontal
-- area centered on the node to be spawned on.
-- Defaults to 0, but is ignored if the above
-- near_nodes value is not set.
near_nodes_vertical = num, -- Used with the size value above, this extends
-- the vertical range of the near nodes search.
-- Basically, this turns the flat region described
-- above into a cuboid region. The area to be checked
-- will extend this high and this low above/below the
-- target node, centered thereon. Defaults to 1 (only
-- check the layer above, the layer at, and the layer
-- below the target node), but is ignored if
-- near_nodes is not set.
near_nodes_count = num, -- How many of the above nodes must be within that
-- radius. Defaults to 1 but is ignored if near_nodes
-- isn't set. Bear in mind that the total area to be
-- checked is equal to:
-- (near_nodes_size^2)*near_nodes_vertical*2
-- For example, if size is 10 and vertical is 4, then
-- the area is (10^2)*8 = 800 nodes in size, so you'll
-- want to make sure you specify a value appropriate
-- for the size of the area being tested.
air_size = num, -- How large of an area to check for air above and
-- around the target. If omitted, only the space
-- above the target is checked. This does not check
-- for air at the sides or below the target.
air_count = num, -- How many of the surrounding nodes need to be air
-- for the above check to return true. If omitted,
-- only the space above the target is checked.
plantlife_limit = num, -- The value compared against the generic "plants
-- can grow here" Perlin noise layer. Smaller numbers
-- result in more abundant plants. Range of -1 to +1,
-- with values in the range of about 0 to 0.5 being
-- most useful. Defaults to 0.1.
temp_min = num, -- Minimum temperature needed for the desired object
-- to spawn. This is a 2d Perlin value, which has an
-- inverted range of +1 to -1. Larger values
-- represent *colder* temperatures, so this value is
-- actually the upper end of the desired Perlin range.
-- See the temperature map section at the bottom of
-- this document for details on how these values work.
-- Defaults to +1 (unlimited coldness).
temp_max = num, -- Maximum temperature/lower end of the Perlin range.
-- Defaults to -1 (unlimited heat).
humidity_min = num, -- Minimum humidity for the plant to spawn in. Like
-- the temperature map, this is a Perlin value where
-- lower numbers mean more humidity in the area.
-- Defaults to +1 (0% humidity).
humidity_max = num, -- Maximum humidity for the plant to spawn at.
-- Defaults to -1 (100% humidity).
verticals_list = {table}, -- List of nodes that should be considered to be
-- natural walls.
alt_wallnode = "string", -- If specified, this node will be substituted in
-- place of the plant(s) defined by spawn_plants
-- above, if the spawn target has one or more adjacent
-- walls. In such a case, the two above facedir
-- parameters will be ignored.
spawn_on_side = bool, -- Set this to true to immediately spawn the node on
-- one side of the target node rather than the top.
-- The code will search for an airspace to the side of
-- the target, then spawn the plant at the first one
-- found. The above facedir and random_facedir
-- parameters are ignored in this case. If the above
-- parameters for selecting generic wall nodes are
-- provided, this option is ignored. Important note:
-- the facedir values assigned by this option only
-- make sense with wallmounted nodes (nodes which
-- don't use facedir won't be affected).
choose_random_wall = bool, -- if set to true, and searching for walls is
-- being done, just pick any random wall if there is
-- one, rather than returning the first one.
spawn_on_bottom = bool, -- If set to true, spawn the object below the
-- target node instead of above it. The above
-- spawn_on_side variable takes precedence over this
-- one if both happen to be true. When using this
-- option with the random facedir function above, the
-- values given to the facedir parameter are for
-- regular nodes, not wallmounted.
spawn_replace_node = bool, -- If set to true, the target node itself is
-- replaced by the spawned object. Overrides the
-- spawn_on_bottom and spawn_on_side settings.
}
[*] spawn_plants must be either a table or a string. If it's a table, the
values therein are treated as a list of nodenames to pick from randomly on
each application of the ABM code. The more nodes you can pack into this
parameter to avoid making too many calls to this function, the lower the CPU
load will likely be.
You can also specify a string containing the name of a function to execute.
In this case, the function will be passed a single position parameter
indicating where the function should place the desired object, and the checks
for spawning on top vs. sides vs. bottom vs. replacing the target node will be
skipped.
By default, if a biome node, size, and count are not defined, the biome
checking is disabled. Same holds true for the nneighbors bit above that.
=====
biome_lib.register_on_generate(biome, nodes_or_function_or_treedef)
To register an object to be spawned at mapgen time rather than via an ABM,
call this function with two parameters: a table with your object's biome
information, and a string, function, or table describing what to do if the
engine finds a suitable surface node (see below).
The biome table contains quite a number of options, though there are fewer
here than are available in the ABM-based spawner, as some stuff doesn't make
sense at map-generation time.
biome = {
surface = something, -- What node(s). May be a string such as
-- "default:dirt_with_grass" or a table with
-- multiple such entries.
---- Everything else is optional, but you'll definitely want to use
---- some of these other fields to limit where and under what
---- conditions the objects are spawned.
below_nodes = {table}, -- List of nodes that must be below the target
-- node. Useful in snow biomes to keep objects from
-- spawning in snow that's on the wrong surface for
-- that object.
avoid_nodes = {table}, -- List of nodes to avoid when spawning. Groups are
-- not supported here.
avoid_radius = num, -- How much distance to leave between the object to be
-- added and the objects to be avoided. If this or
-- the avoid_nodes value is nil/omitted, this check is
-- skipped. Avoid using excessively large radii.
rarity = num, -- How rare should this object be in its biome? Larger
-- values make objects more rare, via:
-- math.random(1,100) > this
max_count = num, -- The absolute maximum number of your object that
-- should be allowed to spawn in a 5x5x5 mapblock area
-- (80x80x80 nodes). Defaults to 5, but be sure you
-- set this to some reasonable value depending on your
-- object and its size if 5 is insufficient.
seed_diff = num, -- Perlin seed-diff value. Defaults to 0, which
-- causes the function to inherit the global value of
-- 329.
neighbors = {table}, -- What ground nodes must be right next to and at the
-- same elevation as the node to be spawned on.
ncount = num, -- At least this many of the above nodes must be next
-- to the node to spawn on. Any value greater than 8
-- will probably cause the code to never spawn
-- anything. Defaults to 0.
depth = num, -- How deep/thick of a layer the spawned-on node must
-- be. Typically used for water.
min_elevation = num, -- Minimum elevation in meters/nodes. Defaults to
-- -16 meters.
max_elevation = num, -- Max elevation. Defaults to +48m.
near_nodes = {table}, -- what nodes must be in the general vicinity of the
-- object being spawned.
near_nodes_size = num, -- how wide of a search area to look for the nodes
-- in that list.
near_nodes_vertical = num, -- How high/low of an area to search from the
-- target node.
near_nodes_count = num, -- at least this many of those nodes must be in
-- the area.
plantlife_limit = num, -- The value compared against the generic "plants
-- can grow here" Perlin noise layer. Smaller numbers
-- result in more abundant plants. Range of -1 to +1,
-- with values in the range of about 0 to 0.5 being
-- most useful. Defaults to 0.1.
temp_min = num, -- Coldest allowable temperature for a plant to spawn
-- (that is, the largest Perlin value).
temp_max = num, -- warmest allowable temperature to spawn a plant
-- (lowest Perlin value).
verticals_list = {table}, -- Same as with the spawn_on_surfaces function.
check_air = bool, -- Flag to tell the mapgen code to check for air above
-- the spawn target. Defaults to true if not
-- explicitly set to false. Set this to false VERY
-- SPARINGLY, as it will slow the map generator down.
delete_above = bool, -- Flag to tell the mapgen code to delete the two
-- nodes directly above the spawn target just before
-- adding the plant or tree. Useful when generating
-- in snow biomes. Defaults to false.
delete_above_surround = bool, -- Flag to tell the mapgen code to also
-- delete the five nodes surrounding the above space,
-- and the five nodes above those, resulting in a two-
-- node-deep cross-shaped empty region above/around
-- the spawn target. Useful when adding trees to snow
-- biomes. Defaults to false.
spawn_replace_node = bool, -- same as with the ABM spawner.
random_facedir = {table}, -- same as with the ABM spawner.
}
Regarding nodes_or_function_or_treedef, this must either be a string naming
a node to spawn, a table with a list of nodes to choose from, a table with an
L-Systems tree definition, or a function.
If you specified a string, the code will attempt to determine whether that
string specifies a valid node name. If it does, that node will be placed on
top of the target position directly (unless one of the other mapgen options
directs the code to do otherwise).
If you specified a table and there is no "axiom" field, the code assumes that
it is a list of nodes. Simply name one node per entry in the list, e.g.
{"default:junglegrass", "default:dry_shrub"} and so on, for as many nodes as
you want to list. A random node from the list will be chosen each time the
code goes to place a node.
If you specified a table, and there *is* an "axiom" field, the code assumes
that this table contains an L-Systems tree definition, which will be passed
directly to the engine's spawn_tree() function along with the position on
which to spawn the tree.
You can also supply a function to be directly executed, which is given the
current node position (the usual "pos" table format) as its sole argument. It
will be called in the form:
somefunction(pos)
=====
biome_lib.update_plant(options)
This third function is used to turn the spawned nodes above into something
else over time. This function has no return value, and accepts a biome
definition table as the only parameter. These are defined like so:
options = {
label = string, -- set this to identify the ABM for Minetest's
-- profiler. If not set, biome_lib will set it to
-- "biome_lib.update_plant(): " appended with the node
-- in grow_plant (or the first item if it's a table)
grow_plant = "string" or {table}, -- Name(s) of the node(s) to be grown
-- into something else. This value is passed to the
-- ABM as the "nodenames" parameter, so the plants
-- themselves are the ABM trigger, rather than
-- the ground they spawned on. A plant will only grow
-- if the node above it is air. If you use a table,
-- note that all nodes referenced therein will be
-- grown into the same final object.
grow_delay = num, -- Passed as the ABM "interval" parameter, as with
-- spawning.
grow_chance = num, -- Passed as the ABM "chance" parameter.
grow_result = "string", -- Name of the node into which the grow_plant
-- node(s) should transform when the ABM executes.
---- Everything from here down is optional.
dry_early_node = "string", -- This value is ignored except for jungle
-- grass (a corner case needed by that mod), where it
-- indicates which node the grass must be on in order
-- for it to turn from the short size to
-- "default:dry_shrub" instead of the medium size.
grow_nodes = {table}, -- One of these nodes must be under the plant in
-- order for it to grow at all. Normally this should
-- be the same as the list of surfaces passed to the
-- spawning ABM as the "nodenames" parameter. This is
-- so that the plant can be manually placed on
-- something like a flower pot or something without it
-- necessarily growing and perhaps dieing. Defaults
-- to "default:dirt_with_grass".
facedir = num, -- Same as with spawning a plant.
need_wall = bool, -- Set this to true if you the plant needs to grow
-- against a wall. Defaults to false.
verticals_list = {table}, -- same as with spawning a plant.
choose_random_wall = bool, -- same as with spawning a plant.
grow_vertically = bool, -- Set this to true if the plant needs to grow
-- vertically, as in climbing poison ivy. Defaults to
-- false.
height_limit = num, -- Set this to limit how tall the desired node can
-- grow. The mod will search straight down from the
-- position being spawned at to find a ground node,
-- set via the field below. Defaults to 5 nodes.
ground_nodes = {table}, -- What nodes should be treated as "the ground"
-- below a vertically-growing plant. Usually this
-- should be the same as the grow_nodes table, but
-- might also include, for example, water or some
-- other surrounding material. Defaults to
-- "default:dirt_with_grass".
grow_function = something, -- [*] see below.
seed_diff = num, -- [*] see below.
}
[*] grow_function can take one of three possible settings: it can be nil (or
not provided), a string, or a table.
If it is not provided or it's set to nil, all of the regular growing code is
executed normally, the value of seed_diff, if any, is ignored, and the node to
be placed is assumed to be specified in the grow_result variable.
If this value is set to a simple string, this is treated as the name of the
function to use to grow the plant. In this case, all of the usual growing
code is executeed, but then instead of a plant being simply added to the
world, grow_result is ignored and the named function is executed and passed a
few parmeters in the following general form:
somefunction(pos, perlin1, perlin2)
These values represent the current position (the usual table), the Perlin
noise value for that spot in the generic "plants can grow here" map for the
seed_diff value above, the Perlin value for that same spot from the
temperature map, and the detected neighboring wall face, if there was one (or
nil if not). If seed_diff is not provided, it defaults to 0.
If this variable is instead set to a table, it is treated an an L-Systems tree
definition. All of the growing code is executed in the usual manner, then the
tree described by that definition is spawned at the current position instead,
and grow_result is ignored.
=====
biome_lib.find_adjacent_wall(pos, verticals, randomflag)
Of the few helper functions, this one expects a position parameter and a table
with the list of nodes that should be considered as walls. The code will
search around the given position for a neighboring wall, returning the first
one it finds as a facedir value, or nil if there are no adjacent walls.
If randomflag is set to true, the function will just return the facedir of any
random wall it finds adjacent to the target position. Defaults to false if
not specified.
=====
biome_lib.is_node_loaded(pos)
This acts as a wrapper for the minetest.get_node_or_nil(node_pos)
function and accepts a single position parameter. Returns true if the node in
question is already loaded, or false if not.
=====
biome_lib.dbg(string, level)
This is a simple debug output function which takes one string parameter. It
just checks if DEBUG is true and outputs the phrase "[Plantlife] " followed by
the supplied string, via the print() function, if so.
'level' is a number that, if supplied, dictates the lowest 'biome_lib_debug'
can be set to in minetest.conf for this message to be displayed. Both the
default log level and the default message level are 0, thus always showing the
supplied message.
Although it's not set in stone, a good practice is to use a level of 0 (or
just omit the value) for anything that generally important enough that it
ought always be shown, 1 for errors, 2 for warnings, 3 for info, 4 for verbose
spammy stuff.
=====
biome_lib.generate_ltree(pos, treemodel)
biome_lib.grow_ltree(pos, treemodel)
In the case of the growing code and the mapgen-based tree generator code,
generating a tree is done via the above two calls, which in turn immediately
call the usual spawn_tree() functions. This rerouting exists as a way for
other mods to hook into biome_lib's tree-growing functions in general,
perhaps to execute something extra whenever a tree is spawned.
biome_lib.generate_ltree(pos, treemodel) is called any time a tree is spawned
at map generation time. 'pos' is the position of the block on which the tree
is to be placed. 'treemodel' is the standard L-Systems tree definition table
expected by the spawn_tree() function. Refer to the 'trunk' field in that
table to derive the name of the tree being spawned.
biome_lib.grow_ltree(pos, treemodel) does the same sort of thing whenever a
tree is spawned within the abm-based growing code, for example when growing a
sapling into a tree.
=====
There are other, internal helper functions that are not meant for use by other
mods. Don't rely on them, as they are subject to change without notice.
======================
Fertile Ground Mapping
======================
The mod uses Perlin noise to create "biomes" of the various plants, via the
minetest.get_perlin() function. At present, there are three layers of
Perlin noise used.
The first one is for a "fertile ground" layer, which I tend to refer to as the
generic "stuff can potentially grow here" layer. Its values are hard-coded:
perlin_octaves = 3
perlin_persistence = 0.6
perlin_scale = 100
For more information on how Perlin noise is generated, you will need to search
the web, as these default values were from that which is used by minetest_game
to spawn jungle grass at mapgen time, and I'm still learning how Perlin noise
works. ;-)
===================
Temperature Mapping
===================
The second Perlin layer is a temperature map, with values taken from
SPlizard's Snow Biomes mod so that the two will be compatible, since that mod
appears to be the standard now. Those values are:
temperature_seeddiff = 112
temperature_octaves = 3
temperature_persistence = 0.5
temperature_scale = 150
The way Perlin values are used by this mod, in keeping with the snow mod's
apparent methods, larger values returned by the Perlin function represent
*colder* temperatures. In this mod, the following table gives a rough
approximation of how temperature maps to these values, normalized to
0.53 = 0 °C and +1.0 = -25 °C.
Perlin Approx. Temperature
-1.0 81 °C ( 178 °F)
-0.75 68 °C ( 155 °F)
-0.56 58 °C ( 136 °F)
-0.5 55 °C ( 131 °F)
-0.25 41 °C ( 107 °F)
-0.18 38 °C ( 100 °F)
0 28 °C ( 83 °F)
0.13 21 °C ( 70 °F)
0.25 15 °C ( 59 °F)
0.5 2 °C ( 35 °F)
0.53 0 °C ( 32 °F)
0.75 -12 °C ( 11 °F)
0.86 -18 °C ( 0 °F)
1.0 -25 °C (- 13 °F)
Included in this table are even 0.25 steps in Perlin values along with some
common temperatures on both the Centigrade and Fahrenheit scales. Note that
unless you're trying to model the Moon or perhaps Mercury in your mods/maps,
you probably won't need to bother with Perlin values of less than -0.56 or so.
================
Humidity Mapping
================
Last but not least is a moisture/humidity map. Like the temperature map
above, Perlin values can be tested to determine the approximate humidity of
the *air* in the area. This humidity map is basically the perlin layer used
for deserts.
A value of +1.0 is very moist (basically a thick fog, if it could be seen), a
value of roughly +0.25 represents the edge of a desert as usually seen in the
game, and a value of -1.0 is as dry as a bone.
This does not check for nearby water, just general air humidity, and that
being the case, nearby ground does not affect the reported humidity of a
region (because this isn't yet possible to calculate yet). Use the near_nodes
and avoid_nodes parameters and their related options to check for water and
such.
The Perlin values use for this layer are:
humidity_seeddiff = 9130
humidity_octaves = 3
humidity_persistence = 0.5
humidity_scale = 250
And this particular one is mapped slightly differently from the others:
noise3 = perlin3:get2d({x=p_top.x+150, y=p_top.z+50})
(Note the +150 and +50 offsets)
==================
Default game nodes
==================
Although this project was intended to be used with minetest_game, it can be
configured to work with something else instead. All you need to do is provide
the names of the nodes in your game you want biome_lib's internals to use.
See settingtypes.txt for a list. Any item listed there can be changed either
by adding it to your minetest.conf, or by using the "all settings" menu in
Minetest, whatever's appropriate for your particular setup.
==================
Engine Decorations
==================
If a call to biome_lib.register_on_generate() contains items and biome
definition settings that are suitable, biome_lib will pass that call on to the
engine instead, to use its built-in decorations feature, since it'll be much
faster than Lua.
For this to work, first the item to be added must either be a node, or a
table with a list of nodes that biome_lib would normally pick from randomly.
That is to say, you cannot specify an L-tree or a function here, as the engine
does not support that sort of thing (biome_lib will just switch to its normal
handling if you do).
Second, these biome definition items must not be present:
* below_nodes
* avoid_nodes
* avoid_radius
* neighbors
* ncount
* depth
* near_nodes_size
* near_nodes_vertical
* temp_min
* temp_max
* verticals_list
* delete_above
* delete_above_surround
The plantlife_limit definition item is ignored when checking if a particular
call can be routed to the engine.
The call given to the engine will use the remaining biome definition items in
the following manner:
deco_type = "simple",
flags = "all_floors"
decoration = node or table with node list
place_on = surface
y_min = min_elevation
y_max = max_elevation
spawn_by = near_nodes
num_spawn_by = near_nodes_count
param2 = \_ set to the range specified by the biome definition's
param2_max = / random_facedir table, if present, otherwise omitted
noise_params = {
octaves = biome_lib.fertile_perlin_octaves,
persist = biome_lib.fertile_perlin_persistence * (250/biome_lib.fertile_perlin_scale),
scale = ((100-biome.rarity)/100) * (math.min(biome.max_count, 6400)/6400),
seed = biome.seed_diff,
offset = 0,
spread = {x = 100, y = 100, z = 100},
lacunarity = 2,
flags = "absvalue"
}
If the biome definition's check_air setting is false, "force_placement" is
added to the decoration's flags setting.
If the biome def's spawn_replace_node is set to true, the decoration's
place_offset_y is set to -1 (otherwise it is omitted).