Files
Bubberstation/code/datums/datum.dm
LemonInTheDark ab307032ed Nightvision Rework (In the name of color) (#73094)
## About The Pull Request

Relies on #72886 for some render relay expansion I use for light_mask
stuff.

Hello bestie! Night vision pissed me off, so I've come to burn this
place to the ground.
Two sections to discuss here. First we'll talk about see_in_dark and why
I hate it, second we'll discuss the lighting plane and how we brighten
it, plus introducing color to the party.

### `see_in_dark` and why it kinda sucks

https://www.byond.com/docs/ref/#/mob/var/see_in_dark

See in dark lets us control how far away from us a turf can be before we
hide it/its contents if it's dark (not got luminosity set)
We currently set it semi inconsistently to provide nightvision to mobs.

The trouble is stuff that produces light != stuff that sets luminosity.
The worst case of this can be seen by walking out of escape on icebox,
where you'll see this


![image](https://user-images.githubusercontent.com/58055496/215683654-587fb00f-ebb8-4c83-962d-a1b2bf429c4a.png)

Snow draws above the lighting plane, so the snow will intermittently
draw, depending on see_in_dark and the luminosity from tracking lights.
This would in theory be solvable by modifying the area, but the same
problem applies across many things in the codebase.
As things currently stand, to be emissive you NEED to have a light on
your tile. People are bad at this, and honestly it's a bit much to
expect of them. An emissive overlay on a canister shouldn't need an
element or something and a list on turfs to manage it.
This gets worse when you factor in the patterns I'm using to avoid
drawing lights above nothing, which leads to lights that should show,
but are misoffset because their parent pixel offsets.

It's silly. We do it so we can have things like mesons without just
handing out night vision, but even there the effect of just hiding
objects and mobs looks baddddddd when moving. It's always bothered me.
I'll complain about mesons more later, but really just like, they're too
bright as it is.

I'm proposing here that rather then manually hiding stuff based off
distance from the player, we can instead show/hide using just the
lighting plane. This means things like mesons are gonna get dimmer, but
that's fine because they suck.

It does have some side effects, things like view() on mobs won't hide
stuff in darkness, but that's fine because none actually thinks about
view like that, I think.

Oh and I added a case to prevent examining stuff that's in darkness, and
not right next to you when you don't have enough nightvision, to match
the old behavior `see_in_dark` gave us.

Now I'd like to go on a mild tangent about color, please bare with me

### Color and why `lighting_alpha` REALLY sucks

You ever walk around with mesons on when there's a fire going, or an
ethereal or firelocks down.
You notice how there isn't really much color to our lights? Doesn't that
suck?

It's because the way we go about brighting lighting is by making
everything on the lighting plane transparent.
This is fine for brightening things, but it ends up looking kinda crummy
in the end and leads to really washed out colors that should be bright.
Playing engineer or miner gets fucking depressing.

The central idea of this pr, that everything else falls out of, is
instead of making the plane more transparent, we can use color matrixes
to make things AT LEAST x bright.

https://www.byond.com/docs/ref/#/{notes}/color-matrix

Brief recap for color matrixes, fully expanded they're a set of 20
different values in a list
Units generally scale 0-1 as multipliers, though since it's
multiplication in order to make an rgb(1,1,1) pixel fullbright you would
need to use 255s.

A "unit matrix" for color looks like this:
```
list(1, 0, 0, 0,
     0, 1, 0, 0,
     0, 0, 1, 0,
     0, 0, 0, 1,
     0, 0, 0, 0
)
```

The first four rows are how much each r, g, b and a impact r, g, b and
well a.
So a first row of `(1, 0, 0, 0)` means 1 unit of r results in 1 unit of
r. and 0 units of green, blue and alpha, and so on.
A first row of `(0, 1, 0, 0)` would make 1 red component into 1 green
component, and leave red, blue and alpha alone, shifting any red of
whatever it's applied to a green.

Using these we can essentially color transform our world. It's a fun
tool. But there's more.

That last row there doesn't take a variable input like the others.
Instead, it ADDS some fraction of 255 to red, green, blue and alpha.

So a fifth row of `(1, 0, 0, 0)` would make every pixel as red as it
could possibly be.

This is what we're going to exploit here. You see all these values
accept negative multipliers, so we can lower colors down instead of
raising them up!
The key idea is using color matrix filters
https://www.byond.com/docs/ref/#/{notes}/filters/color to chain these
operations together.

Pulling alllll the way back, we want to brighten darkness without
affecting brighter colors.
Lower rgb values are darker, higher ones are brighter. This relationship
isn't really linear because of suffering reasons, but it's good enough
for this.
Let's try chaining some matrixes on the lighting plane, which is bright
where fullbright, and dark where dark.

Take a list like this

```
list(1, 0, 0, 0,
     0, 1, 0, 0,
     0, 0, 1, 0,
     0, 0, 0, 1,
     -0.2, -0.2, -0.2, 0
)
```
That would darken the lighting a bit, but negative values will get
rounded to 0
A subsequent raising by the same amount
```
list(1, 0, 0, 0,
     0, 1, 0, 0,
     0, 0, 1, 0,
     0, 0, 0, 1,
     0.2, 0.2, 0.2, 0
)
```
Will essentially threshold our brightness at that value.
This ensures we aren't washing out colors when we make things brighter,
while leaving higher values unaffected since they basically just had a
constant subtracted and then readded.

### But wait, there's more

You may have noticed, we gain access to individual color components
here.
This means not only can we darken and lighten by thresholds, we can
COLOR those thresholds.
```
list(1, 0, 0, 0,
     0, 1, 0, 0,
     0, 0, 1, 0,
     0, 0, 0, 1,
     0.1, 0.2, 0.1, 0
)
```
Something like the above, if applied with its inverse, would tint the
darkness green.
The delta between the different scalars will determine how vivid the
color is, and the actual value will impact the brightness.

Something that's always bothered me about nightvision is it's just
greyscale for the most part, there isn't any color to it.
There was an old idea of coloring the game plane to match their lenses,
but if you've ever played with the colorblind quirk you know that gets
headachey really fast.
So instead of that, lets color just the darkness that these glasses
produce.
It provides some reminder that you're wearing them, instead of just
being something you forget about while playing, and provides a reason to
use flashlights and such since they can give you a clearer, less tinted
view of things while retaining the ability to look around things.

I've so far applied this pattern to JUST headwear for humans (also those
mining wisps)
I'm planning on furthering it to mobs that use nightvision, but I wanted
to get this up cause I don't wanna pr it the day before the freeze.

Mesons are green, sec night vision is red, thermals orange, etc.

I think the effect this gives is really really nice. 
I've tuned most things to work for the station, though mesons works for
lavaland for obvious reasons.

I've tuned things significantly darker then we have them set currently,
since I really hate flat lighting and this system suffers when
interacting with it.

My goal with these is to give you a rough idea of what's around you,
without a good eye for detail.
That's the difference between say, mesons, and night vision. One helps
you see outlines, the other gives you detail and prevents missing
someone in the darkness.

It's hard to balance this precisely because of different colored
backgrounds (looking at you icebox)
More can be done on this front in future but I'm quite happy with things
as of now

### **EDIT**

I have since expanded to all uses of nightvision, coloring most all of
them.

Along the way I turned some toggleable nightvision into just one level. 
Fullbright sucks, and I'd rather just have one "good" value.

I've kept it for a few cases, mostly eyes you rip out of mobs.
Impacted mobs are nightmares, aliens, zombies, revenants, states and
sort of stands.

I've done a pass on all mobs and items that impact nightvision and added
what I thought was the right level of color to them. This includes stuff
like blobs and shuttle control consoles
As with glasses much of this was around reducing vision, though I kept
it stronger here, since many of these mobs rely on it for engaging with
the game

<details>
<summary>
Technical Changes
</summary>

#### Adds filter proc (the ones that act like templates) support to
filter transitions.
Found this when testing this pr, seemed silly.

#### Makes our emissive mask mask all light instead
This avoids dumbass overlay lighting lighting up wallmounts.
We switch modes if some turfflags are set, to accomplish the same thing
with more overhead, and support showing things through the darkness.

Also fixes a bug where you'd only get one fullscreen object per mob, so
opening and closing a submap would take it away

Also also fixes the lighting backdrop not actually spanning the screen. 
It doesn't actually do anything anymore because of the fullscreen light
we have, but just in case that's unsued.
Needs cleanup in future.

#### Moves openspace to its own plane that doesn't draw, maxing its
color with a sprite

This is to support the above
We relay this plane to lighting mask so openspace can like, have
lighting

#### Changes our definition of nightvision to the light cutoff of night
vision goggles and such
Side affect of removing see_in_dark. This logic is a bit weak atm, needs
some work.

#### Removes the nightvision spell
It's a dupe of the nightvision action button, and newly redundant since
I've removed all uses of it

#### Cleans up existing plane master critical defines, ensures
trasnparent won't render

These sucked
Also transparent stuff should never render, if it does you'll get white
blobs which suck

</details>

## Why It's Good For The Game

Videos! (Github doesn't like using a summary here I'm sorry)
<details>

Demonstration of ghost lighting, and color


https://user-images.githubusercontent.com/58055496/215693983-99e00f9e-7214-4cf4-a76a-6e669a8a1103.mp4

Engi-glass mesons and walking in maint (Potentially overtuned, yellow is
hard)


https://user-images.githubusercontent.com/58055496/215695978-26e7dc45-28aa-4285-ae95-62ea3d79860f.mp4

Diagnostic nightvision goggles and see_in_dark not hiding emissives


https://user-images.githubusercontent.com/58055496/215692233-115b4094-1099-4393-9e94-db2088d834f3.mp4

Sec nightvision (I just think it looks neat)


https://user-images.githubusercontent.com/58055496/215692269-bc08335e-0223-49c3-9faf-d2d7b22fe2d2.mp4

Medical nightvision goggles and other colors


https://user-images.githubusercontent.com/58055496/215692286-0ba3de6a-b1d5-4aed-a6eb-c32794ea45da.mp4

Miner mesons and mobs hiding in lavaland (This is basically the darkest
possible environment)


https://user-images.githubusercontent.com/58055496/215696327-26958b69-0e1c-4412-9298-4e9e68b3df68.mp4

Thermal goggles and coloring displayed mobs


https://user-images.githubusercontent.com/58055496/215692710-d2b101f3-7922-498c-918c-9b528d181430.mp4

</details>

I think it's pretty, and see_in_dark sucks butt.

## Changelog

<!-- If your PR modifies aspects of the game that can be concretely
observed by players or admins you should add a changelog. If your change
does NOT meet this description, remove this section. Be sure to properly
mark your PRs to prevent unnecessary GBP loss. You can read up on GBP
and it's effects on PRs in the tgstation guides for contributors. Please
note that maintainers freely reserve the right to remove and add tags
should they deem it appropriate. You can attempt to finagle the system
all you want, but it's best to shoot for clear communication right off
the bat. -->

🆑
add: The darkness that glasses and hud goggles that impact your
nightvision (think mesons, nightvision goggles, etc) lighten is now
tinted to match the glasses. S pretty IMO, and hopefully it helps with
forgetting you're wearing X.
balance: Nightvision is darker. I think bright looks bad, and things
like mesons do way too much
balance: Mesons (and mobs in general) no longer have a static distance
you can see stuff in the dark. If a tile is lit, you can now see it.
fix: Nightvision no longer dims colored lights, instead simply
thresholding off bits of darkness that are dimmer then some level.
/🆑
2023-02-17 18:10:39 -07:00

393 lines
12 KiB
Plaintext

/**
* The absolute base class for everything
*
* A datum instantiated has no physical world prescence, use an atom if you want something
* that actually lives in the world
*
* Be very mindful about adding variables to this class, they are inherited by every single
* thing in the entire game, and so you can easily cause memory usage to rise a lot with careless
* use of variables at this level
*/
/datum
/**
* Tick count time when this object was destroyed.
*
* If this is non zero then the object has been garbage collected and is awaiting either
* a hard del by the GC subsystme, or to be autocollected (if it has no references)
*/
var/gc_destroyed
/// Active timers with this datum as the target
var/list/active_timers
/// Status traits attached to this datum. associative list of the form: list(trait name (string) = list(source1, source2, source3,...))
var/list/status_traits
/**
* Components attached to this datum
*
* Lazy associated list in the structure of `type:component/list of components`
*/
var/list/datum_components
/**
* Any datum registered to receive signals from this datum is in this list
*
* Lazy associated list in the structure of `signal:registree/list of registrees`
*/
var/list/comp_lookup
/// Lazy associated list in the structure of `signals:proctype` that are run when the datum receives that signal
var/list/list/datum/callback/signal_procs
/// Datum level flags
var/datum_flags = NONE
/// A cached version of our \ref
/// The brunt of \ref costs are in creating entries in the string tree (a tree of immutable strings)
/// This avoids doing that more then once per datum by ensuring ref strings always have a reference to them after they're first pulled
var/cached_ref
/// A weak reference to another datum
var/datum/weakref/weak_reference
/*
* Lazy associative list of currently active cooldowns.
*
* cooldowns [ COOLDOWN_INDEX ] = add_timer()
* add_timer() returns the truthy value of -1 when not stoppable, and else a truthy numeric index
*/
var/list/cooldowns
/// List for handling persistent filters.
var/list/filter_data
#ifdef REFERENCE_TRACKING
var/running_find_references
var/last_find_references = 0
#ifdef REFERENCE_TRACKING_DEBUG
///Stores info about where refs are found, used for sanity checks and testing
var/list/found_refs
#endif
#endif
#ifdef DATUMVAR_DEBUGGING_MODE
var/list/cached_vars
#endif
/**
* Called when a href for this datum is clicked
*
* Sends a [COMSIG_TOPIC] signal
*/
/datum/Topic(href, href_list[])
..()
SEND_SIGNAL(src, COMSIG_TOPIC, usr, href_list)
/**
* Default implementation of clean-up code.
*
* This should be overridden to remove all references pointing to the object being destroyed, if
* you do override it, make sure to call the parent and return it's return value by default
*
* Return an appropriate [QDEL_HINT][QDEL_HINT_QUEUE] to modify handling of your deletion;
* in most cases this is [QDEL_HINT_QUEUE].
*
* The base case is responsible for doing the following
* * Erasing timers pointing to this datum
* * Erasing compenents on this datum
* * Notifying datums listening to signals from this datum that we are going away
*
* Returns [QDEL_HINT_QUEUE]
*/
/datum/proc/Destroy(force=FALSE, ...)
SHOULD_CALL_PARENT(TRUE)
SHOULD_NOT_SLEEP(TRUE)
tag = null
datum_flags &= ~DF_USE_TAG //In case something tries to REF us
weak_reference = null //ensure prompt GCing of weakref.
if(active_timers)
var/list/timers = active_timers
active_timers = null
for(var/datum/timedevent/timer as anything in timers)
if (timer.spent && !(timer.flags & TIMER_DELETE_ME))
continue
qdel(timer)
#ifdef REFERENCE_TRACKING
#ifdef REFERENCE_TRACKING_DEBUG
found_refs = null
#endif
#endif
//BEGIN: ECS SHIT
var/list/dc = datum_components
if(dc)
var/all_components = dc[/datum/component]
if(length(all_components))
for(var/datum/component/component as anything in all_components)
qdel(component, FALSE, TRUE)
else
var/datum/component/C = all_components
qdel(C, FALSE, TRUE)
dc.Cut()
clear_signal_refs()
//END: ECS SHIT
return QDEL_HINT_QUEUE
///Only override this if you know what you're doing. You do not know what you're doing
///This is a threat
/datum/proc/clear_signal_refs()
var/list/lookup = comp_lookup
if(lookup)
for(var/sig in lookup)
var/list/comps = lookup[sig]
if(length(comps))
for(var/datum/component/comp as anything in comps)
comp.UnregisterSignal(src, sig)
else
var/datum/component/comp = comps
comp.UnregisterSignal(src, sig)
comp_lookup = lookup = null
for(var/target in signal_procs)
UnregisterSignal(target, signal_procs[target])
#ifdef DATUMVAR_DEBUGGING_MODE
/datum/proc/save_vars()
cached_vars = list()
for(var/i in vars)
if(i == "cached_vars")
continue
cached_vars[i] = vars[i]
/datum/proc/check_changed_vars()
. = list()
for(var/i in vars)
if(i == "cached_vars")
continue
if(cached_vars[i] != vars[i])
.[i] = list(cached_vars[i], vars[i])
/datum/proc/txt_changed_vars()
var/list/l = check_changed_vars()
var/t = "[src]([REF(src)]) changed vars:"
for(var/i in l)
t += "\"[i]\" \[[l[i][1]]\] --> \[[l[i][2]]\] "
t += "."
/datum/proc/to_chat_check_changed_vars(target = world)
to_chat(target, txt_changed_vars())
#endif
///Return a LIST for serialize_datum to encode! Not the actual json!
/datum/proc/serialize_list(list/options)
CRASH("Attempted to serialize datum [src] of type [type] without serialize_list being implemented!")
///Accepts a LIST from deserialize_datum. Should return src or another datum.
/datum/proc/deserialize_list(json, list/options)
CRASH("Attempted to deserialize datum [src] of type [type] without deserialize_list being implemented!")
///Serializes into JSON. Does not encode type.
/datum/proc/serialize_json(list/options)
. = serialize_list(options)
if(!islist(.))
. = null
else
. = json_encode(.)
///Deserializes from JSON. Does not parse type.
/datum/proc/deserialize_json(list/input, list/options)
var/list/jsonlist = json_decode(input)
. = deserialize_list(jsonlist)
if(!istype(., /datum))
. = null
///Convert a datum into a json blob
/proc/json_serialize_datum(datum/D, list/options)
if(!istype(D))
return
var/list/jsonlist = D.serialize_list(options)
if(islist(jsonlist))
jsonlist["DATUM_TYPE"] = D.type
return json_encode(jsonlist)
/// Convert a list of json to datum
/proc/json_deserialize_datum(list/jsonlist, list/options, target_type, strict_target_type = FALSE)
if(!islist(jsonlist))
if(!istext(jsonlist))
CRASH("Invalid JSON")
jsonlist = json_decode(jsonlist)
if(!islist(jsonlist))
CRASH("Invalid JSON")
if(!jsonlist["DATUM_TYPE"])
return
if(!ispath(jsonlist["DATUM_TYPE"]))
if(!istext(jsonlist["DATUM_TYPE"]))
return
jsonlist["DATUM_TYPE"] = text2path(jsonlist["DATUM_TYPE"])
if(!ispath(jsonlist["DATUM_TYPE"]))
return
if(target_type)
if(!ispath(target_type))
return
if(strict_target_type)
if(target_type != jsonlist["DATUM_TYPE"])
return
else if(!ispath(jsonlist["DATUM_TYPE"], target_type))
return
var/typeofdatum = jsonlist["DATUM_TYPE"] //BYOND won't directly read if this is just put in the line below, and will instead runtime because it thinks you're trying to make a new list?
var/datum/D = new typeofdatum
var/datum/returned = D.deserialize_list(jsonlist, options)
if(!isdatum(returned))
qdel(D)
else
return returned
/**
* Callback called by a timer to end an associative-list-indexed cooldown.
*
* Arguments:
* * source - datum storing the cooldown
* * index - string index storing the cooldown on the cooldowns associative list
*
* This sends a signal reporting the cooldown end.
*/
/proc/end_cooldown(datum/source, index)
if(QDELETED(source))
return
SEND_SIGNAL(source, COMSIG_CD_STOP(index))
TIMER_COOLDOWN_END(source, index)
/**
* Proc used by stoppable timers to end a cooldown before the time has ran out.
*
* Arguments:
* * source - datum storing the cooldown
* * index - string index storing the cooldown on the cooldowns associative list
*
* This sends a signal reporting the cooldown end, passing the time left as an argument.
*/
/proc/reset_cooldown(datum/source, index)
if(QDELETED(source))
return
SEND_SIGNAL(source, COMSIG_CD_RESET(index), S_TIMER_COOLDOWN_TIMELEFT(source, index))
TIMER_COOLDOWN_END(source, index)
///Generate a tag for this /datum, if it implements one
///Should be called as early as possible, best would be in New, to avoid weakref mistargets
///Really just don't use this, you don't need it, global lists will do just fine MOST of the time
///We really only use it for mobs to make id'ing people easier
/datum/proc/GenerateTag()
datum_flags |= DF_USE_TAG
/** Add a filter to the datum.
* This is on datum level, despite being most commonly / primarily used on atoms, so that filters can be applied to images / mutable appearances.
* Can also be used to assert a filter's existence. I.E. update a filter regardless if it exists or not.
*
* Arguments:
* * name - Filter name
* * priority - Priority used when sorting the filter.
* * params - Parameters of the filter.
*/
/datum/proc/add_filter(name, priority, list/params)
LAZYINITLIST(filter_data)
var/list/copied_parameters = params.Copy()
copied_parameters["priority"] = priority
filter_data[name] = copied_parameters
update_filters()
/// Reapplies all the filters.
/datum/proc/update_filters()
ASSERT(isatom(src) || istype(src, /image))
var/atom/atom_cast = src // filters only work with images or atoms.
atom_cast.filters = null
filter_data = sortTim(filter_data, GLOBAL_PROC_REF(cmp_filter_data_priority), TRUE)
for(var/filter_raw in filter_data)
var/list/data = filter_data[filter_raw]
var/list/arguments = data.Copy()
arguments -= "priority"
atom_cast.filters += filter(arglist(arguments))
UNSETEMPTY(filter_data)
/obj/item/update_filters()
. = ..()
update_item_action_buttons()
/** Update a filter's parameter to the new one. If the filter doesnt exist we won't do anything.
*
* Arguments:
* * name - Filter name
* * new_params - New parameters of the filter
* * overwrite - TRUE means we replace the parameter list completely. FALSE means we only replace the things on new_params.
*/
/datum/proc/modify_filter(name, list/new_params, overwrite = FALSE)
var/filter = get_filter(name)
if(!filter)
return
if(overwrite)
filter_data[name] = new_params
else
for(var/thing in new_params)
filter_data[name][thing] = new_params[thing]
update_filters()
/** Update a filter's parameter and animate this change. If the filter doesnt exist we won't do anything.
* Basically a [datum/proc/modify_filter] call but with animations. Unmodified filter parameters are kept.
*
* Arguments:
* * name - Filter name
* * new_params - New parameters of the filter
* * time - time arg of the BYOND animate() proc.
* * easing - easing arg of the BYOND animate() proc.
* * loop - loop arg of the BYOND animate() proc.
*/
/datum/proc/transition_filter(name, list/new_params, time, easing, loop)
var/filter = get_filter(name)
if(!filter)
return
// This can get injected by the filter procs, we want to support them so bye byeeeee
new_params -= "type"
animate(filter, new_params, time = time, easing = easing, loop = loop)
modify_filter(name, new_params)
/// Updates the priority of the passed filter key
/datum/proc/change_filter_priority(name, new_priority)
if(!filter_data || !filter_data[name])
return
filter_data[name]["priority"] = new_priority
update_filters()
/// Returns the filter associated with the passed key
/datum/proc/get_filter(name)
ASSERT(isatom(src) || istype(src, /image))
if(filter_data && filter_data[name])
var/atom/atom_cast = src // filters only work with images or atoms.
return atom_cast.filters[filter_data.Find(name)]
/// Returns the indice in filters of the given filter name.
/// If it is not found, returns null.
/datum/proc/get_filter_index(name)
return filter_data?.Find(name)
/// Removes the passed filter, or multiple filters, if supplied with a list.
/datum/proc/remove_filter(name_or_names)
if(!filter_data)
return
var/list/names = islist(name_or_names) ? name_or_names : list(name_or_names)
for(var/name in names)
if(filter_data[name])
filter_data -= name
update_filters()
/datum/proc/clear_filters()
ASSERT(isatom(src) || istype(src, /image))
var/atom/atom_cast = src // filters only work with images or atoms.
filter_data = null
atom_cast.filters = null