Fog and mist weather

[GuardianDll]

Summary

Content "add new weather: mist and fog. maybe."

Purpose of change

I want to make a fog

Describe the solution

Currently fog and mist require extremely high humidity, more than 95%, small wind, and some level of temperature drop
Also @andrei8l fixed autotests, that require clean vision to pass,

Testing

And here is the problem - i can't properly test this solution, because such high humidity happen only when it rains. Also i am not 100% sure i get how it should be done - from what i know there should be also a dew point calculation, but i still don't understand how it should be utilized, as when relative humidity is 100%, the dew point is equal the temperature itself yep, now everything works as expected

[NetSysFire]

irl fog can also happen when the air temperature is lower than the water in a nearby body of water e.g river, which happens frequently where I live. I saw some discussion on discord that specific overmap tiles can cause specific weather, but that is unfortunately out of scope right now.

[GuardianDll]

Found some info, that to generate the mist, the temperature do not need to be same as dew point (every source i was able to find says it should be lower, but it's like impossible physically), but close to it at least for 2.5 centigrade.
With new math, i plan to use a Tetens formula, which calculate a dew point, and another math will calculate a dew point factor - if the factor is bigger than 0, it can generate a mist, and if it's bigger than 1, it can create a fog

{ "math": [ "dew_point", "=", "(237.7 * log(u_val('weather: humidity')/100) + 17.27 * u_val('weather: temperature')) / (17.27 - 237.7 * log(u_val('weather: humidity')/100))" ] },
{ "math": [ "dew_point_factor", "=", "u_val('weather: temperature') - (dew_point + 2.5)" ] }

Thoughts?

[andrei8l]

Using this as a reference, the formula for dew point temperature should be

{
  "math": [
    "dprhx",
    "=",
    "log( u_val('weather:humidity') / 100 ) + 17.625 * u_val('weather:temperature') / ( 243.04 + u_val('weather:temperature') )"
  ]
},
{ "math": [ "dew_point", "=", "243.04 * dprhx / ( 17.625 - dprhx )" ] }

But the "weather": "temperature" function returns temperature in Fahrenheit (undocumented) and the formula needs it in Celsius so you also need to convert it. I'll port the weather functions after 65058 and add conversion functions too.

In the linked paper there are also several simpler formulas that might be just as suitable here.

[GuardianDll]

Ok, i can't actually test it occur naturally, but Ramza says weather code is pretty obscure and may not work properly, so i just make how it should be with hope someone refactor the weather and fix the code in some future oof i'm an idiot, now i can confirm it works

[GuardianDll]

Oh no, now the vision tests are broken Not anymore!

[andrei8l]

There's a genuine vision bug for weathers with sight_penalty not 1.0.

output.webm

but that needs to be addressed elsewhere.

This should fix the vision test for now

diff --git a/tests/vision_test.cpp b/tests/vision_test.cpp
index 51c1ecd30c..e5239c6f64 100644
--- a/tests/vision_test.cpp
+++ b/tests/vision_test.cpp
@@ -149,7 +149,6 @@ struct vision_test_case {
                       const time_point &time ) : setup( setup ), expected_results( expectedResults ), time( time ) {}
 
     void test_all() const {
-        scoped_weather_override weather_clear( WEATHER_CLEAR );
         Character &player_character = get_player_character();
         g->place_player( tripoint( 60, 60, 0 ) );
         player_character.worn.clear(); // Remove any light-emitting clothing
@@ -159,6 +158,7 @@ struct vision_test_case {
         player_character.clear_moncams();
         clear_map( -2, OVERMAP_HEIGHT );
         g->reset_light_level();
+        scoped_weather_override weather_clear( WEATHER_CLEAR );
 
         REQUIRE( !player_character.is_blind() );
         REQUIRE( !player_character.in_sleep_state() );
@@ -541,11 +541,11 @@ TEST_CASE( "vision_junction_reciprocity", "[vision][reciprocity]" )
         player_in_junction ?
         std::vector<std::string>{
             "u#  ",
-            "-- Z",
+            "---Z",
 }:
         std::vector<std::string>{
             "z#  ",
-            "-- u",
+            "---u",
         },
         player_in_junction ?
         std::vector<std::string>{

I'll take another look tomorrow if it still fails.

[andrei8l]

I haven't been able to figure out the shadowcasting bug. We can work around it for the test unit by overspecifying the setup though

This will fix `vision_junction_reciprocity` or your money back

diff --git a/tests/vision_test.cpp b/tests/vision_test.cpp
index e5239c6f64..f7caaceec1 100644
--- a/tests/vision_test.cpp
+++ b/tests/vision_test.cpp
@@ -536,25 +536,34 @@ TEST_CASE( "vision_single_tile_skylight", "[shadowcasting][vision]" )
 TEST_CASE( "vision_junction_reciprocity", "[vision][reciprocity]" )
 {
     bool player_in_junction = GENERATE( true, false );
+    CAPTURE( player_in_junction );
 
     vision_test_case t {
         player_in_junction ?
         std::vector<std::string>{
-            "u#  ",
-            "---Z",
+            "###   ",
+            "#u####",
+            "#---z#",
+            "######",
 }:
         std::vector<std::string>{
-            "z#  ",
-            "---u",
+            "###   ",
+            "#z####",
+            "#---u#",
+            "######",
         },
         player_in_junction ?
         std::vector<std::string>{
-            "4466",
-            "4446",
+            "444666",
+            "444666",
+            "444466",
+            "444466",
 }:
         std::vector<std::string>{
-            "4444",
-            "4444",
+            "666666",
+            "444444",
+            "444444",
+            "444444",
         },
         day_time
     };
@@ -562,14 +571,14 @@ TEST_CASE( "vision_junction_reciprocity", "[vision][reciprocity]" )
     monster *zombie = nullptr;
     tile_predicate spawn_zombie = [&]( map_test_case::tile tile ) {
         zombie = g->place_critter_at( mon_zombie, tile.p );
+        get_map().ter_set( tile.p + tripoint_above, ter_t_flat_roof );
         return true;
     };
 
     t.set_up_tiles =
-        ifchar( 'C', spawn_moncam ) ||
-        ifchar( 'Z', spawn_zombie ) ||
         ifchar( 'z', spawn_zombie ) ||
         t.set_up_tiles;
+    t.flags.headlamp = true;
     t.test_all();
 
     if( player_in_junction ) {

There's also a failure in the weather realism test. Wikipedia says fog and mist aren't precipitation

Potential realism fix; haven't tested in game whether fog/mist still occur

diff --git a/data/json/weather_type.json b/data/json/weather_type.json
index a948fd8ca4..df4aea49bd 100644
--- a/data/json/weather_type.json
+++ b/data/json/weather_type.json
@@ -282,10 +282,11 @@
     "sun_multiplier": 0.3,
     "sound_attn": 0,
     "dangerous": false,
-    "precip": "light",
+    "precip": "none",
     "rains": false,
     "sound_category": "silent",
     "priority": 20,
+    "required_weathers": [ "clear", "sunny" ],
     "condition": {
       "and": [
         { "math": [ "u_val( 'weather:windpower' )", "<", "12" ] },
@@ -307,7 +308,7 @@
     "sun_multiplier": 0.3,
     "sound_attn": 0,
     "dangerous": false,
-    "precip": "heavy",
+    "precip": "none",
     "rains": false,
     "sound_category": "silent",
     "priority": 20,

Sorry it took me so long to get back to this.

[GuardianDll]

> Sorry it took me so long to get back to this

Nothing to worry about!

> Wikipedia says fog and mist aren't precipitation

i was not sure for what precip is used for (Amount of associated precipitation from doc doesn't help much), so i just cargo culted it

[GuardianDll]

ok i somewhat feel that sunny weather won't work, because it require humidity less than 70% - totally opposite of ~100% for fog forming. Maybe replace it with "cloudy"?
And yes, i can't spot mist in the game anymore spot a mist, nevermind then

[bombasticSlacks]

I'm not exactly sure where people sit on this, is this good to merge or is there still things to be resolved?

[GuardianDll]

Tests looks good, in game it seems okay, so i think it good to go