grom/benilla-octo
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

The water already reflected the scenery; it was switched off

Browse Source 
`docs/GRAPHICS.md` said we had refraction and foam but no mirror. That was
wrong. `waterQuality` 2 has carried a depth-marched screen-space
reflection of the scenery all along (`enhanced_water.wgsl`, "HIGH"), with
bisection refinement against the real scene depth rather than the fixed
probe its WarcraftXL source took. What was missing was the switch: the
registered default was Enhanced (sky only), and of the five lighting rungs
only Ultra reached High.

The line holding it back said exactly what it was waiting for —
"Mirror reflections stay opt-in until their cost is measured." So I
measured it, `WOW_FPS_PROBE=300`:

                  Classic   Enhanced   High
    Lake Everstill  1.91 ms   1.84 ms   1.90 ms
    open ocean      1.89 ms   1.88 ms   1.89 ms

Every tier is inside the run-to-run noise — Enhanced lands FASTER than
Classic on the lake, which is how you know you are reading noise rather
than a cost. What it buys is not noise: the far bank and the mountains
appear in the water, 18.5% of the lake frame and 35.1% of the ocean frame.

So High becomes the default and the High rung. That is a weld across four
places the tests hold together, and each one bit in turn and was fixed as
it asked: `REGISTERED`, `VideoConfig`'s literal, `WaterQuality::default`,
and the rung itself — the High rung IS the registered defaults, which
`the_high_preset_is_the_registered_defaults` states and enforces.

Low and Medium keep Enhanced deliberately, and that needed saying in the
one test that assumed every rung between Off and Ultra sits AT its
default: free on the machine it was measured on is an argument for the
default, not for the rung whose whole job is hardware that cannot afford
what this one can. `water_and_lava_observers_clamp_and_water_high_is_opt_in`
asserted the old decision in its name and its body, so it is renamed and
now pins the pairing the other way: High reflects, Medium does not.

The honest remaining gap is written into the doc rather than claimed as
done: SSR gives up when the ray leaves the screen and fades toward the top
edge, which is where a shallow-angle lake wants its far bank. A planar
pass would fill that — and it is the first item on the list that draws the
scene twice, so it costs main thread and gets priced against item 1's
view-distance table before anyone builds it.

Verified: ALL GATES GREEN (492 s); `benilla-app --lib` against the install
at the exact 269-failure baseline, set-identical by `comm`.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01SgstFWtn1RxEJ5kVrpgYwn
This commit is contained in:
grom
2026-10-26 19:26:46 -06:00
co-authored by Claude Opus 5
parent 113572b8e2
commit 77fce9210e
5 changed files with 80 additions and 16 deletions
Download Patch File Download Diff File Expand all files Collapse all files
crates/benilla-app/src/cvars.rs
+40 -8
View File
@@ -799,10 +799,14 @@ pub(crate) const REGISTERED: &[Registered] = &[
"benilla's own: draw the zone skybox LightParams names for the living, 0 Off / 1 On; the \
reference draws a DBC skybox only for the ghost",
),
// GFX (water): the default moves with the High rung, because that rung IS the registered
// defaults (`the_high_preset_is_the_registered_defaults`). Scenery reflection was held opt-in
// "until its cost is measured"; it has been — see the High rung in `LIGHTING_PRESETS`.
ours(
"waterQuality",
"1",
"benilla's own: water quality, 0 Classic / 1 Enhanced / 2 High; mirror reflections are opt-in",
"2",
"benilla's own: water quality, 0 Classic / 1 Enhanced (sky only) / 2 High (the scenery \
reflects too)",
),
ours(
"lavaLightGain",
@@ -2374,8 +2378,15 @@ pub(crate) const LIGHTING_PRESETS: &[(&str, &[(&str, &str)])] = &[
("interiorShadowDynamic", "4"),
("exteriorShadows", "1"),
("spellLightGain", "1"),
// Mirror reflections stay opt-in until their cost is measured.
("waterQuality", "1"),
// GFX (water): measured, which is what this line waited for. Scenery reflection
// (`enhanced_water.wgsl`'s depth-marched SSR) costs **nothing detectable**: over 300
// frames, Lake Everstill runs 1.91 / 1.84 / 1.90 ms at Classic / Enhanced / High and
// the ocean 1.89 / 1.88 / 1.89, with Enhanced landing FASTER than Classic on the lake
// — the tiers are inside the run-to-run noise. What it buys is not subtle: the far
// bank and the mountains appear in the water, which is 18.5% of the frame on the lake
// and 35.1% on the ocean. The GPU is the resource this client has in surplus
// (`docs/GRAPHICS.md`), so High takes it; Ultra already did.
("waterQuality", "2"),
// MONKEY (volumetric fog): preset atmosphere uses the default cheap tier.
("volumetricFog", "1"),
// MONKEY (post): half-resolution halo.
@@ -3391,7 +3402,12 @@ mod tests {
// candle-lit surfaces moving under +10 % — only true at this number.
assert_eq!(d["interiorBakeFloor"], 0.12);
assert_eq!(d["spellLightGain"], 1.0, "the spell lane ships neutral");
assert_eq!(d["waterQuality"], 1.0, "mirror reflections stay opt-in");
// GFX (water): the scenery reflection ships ON. It was held opt-in pending a measurement
// of its cost, and the measurement is nothing — the lake runs 1.91 / 1.84 / 1.90 ms over
// 300 frames across the three tiers, the ocean 1.89 / 1.88 / 1.89, all inside the
// run-to-run noise, while the far bank and the mountains appear in the water (18.5% of
// the lake frame, 35.1% of the ocean).
assert_eq!(d["waterQuality"], 2.0, "the scenery reflects from High");
assert_eq!(d["lavaLightGain"], 1.0);
// MONKEY (ao): registry and renderer agree, and the lane ships Off.
assert_eq!(d["ambientOcclusion"], shadows.ambient_occlusion as f32);
@@ -3556,6 +3572,13 @@ mod tests {
} else if *name == "Ultra" && matches!(member, "waterQuality" | "volumetricFog")
{
"2" // MONKEY (presets): Ultra's two maxima over the defaults.
} else if member == "waterQuality" && matches!(*name, "Low" | "Medium") {
// GFX (water): the first row that sits BELOW its default on the lower
// rungs rather than at it. The scenery reflection measured free on the
// machine it was measured on, which is an argument for High and the
// default — not for Low, whose whole job is the hardware that cannot
// afford what this one can. So those two keep Enhanced.
"1"
} else {
cvars.default_of(member).unwrap()
};
@@ -4086,7 +4109,7 @@ mod tests {
}
#[test]
fn water_and_lava_observers_clamp_and_water_high_is_opt_in() {
fn water_and_lava_observers_clamp_and_the_reflection_rides_the_high_rung() {
let mut app = cvar_app();
for (value, water, lava) in [
("-1", 0, 0.0),
@@ -4099,11 +4122,20 @@ mod tests {
assert_eq!(res::<VideoConfig>(&app).water_quality, water);
assert_eq!(res::<VideoConfig>(&app).lava_light_gain, lava);
}
// GFX (water): High carries the scenery reflection, and a rung below it does not — the
// pairing this used to assert the other way round, when the lane was opt-in pending a
// measurement of its cost.
let mut cvars = fresh_registry();
cvars.set("waterQuality", "2");
cvars.set("waterQuality", "1");
assert_eq!(derive_lighting_quality(&cvars), LIGHTING_CUSTOM);
apply_lighting_preset(&mut cvars, "High");
assert_eq!(cvars.get("waterQuality"), Some("1"));
assert_eq!(cvars.get("waterQuality"), Some("2"));
apply_lighting_preset(&mut cvars, "Medium");
assert_eq!(
cvars.get("waterQuality"),
Some("1"),
"Medium keeps Enhanced: the reflection is High's"
);
}
#[test]
crates/benilla-app/src/ui_script/options_tests.rs
+3 -1
View File
@@ -2613,12 +2613,14 @@ fn water_quality_writes_numeric_tiers_with_localised_labels() {
.unwrap(),
"RowLavaGlow"
);
// GFX (water): the row opens on High, the registered default since the scenery
// reflection measured free — `labels[1]` was Enhanced, when it was still opt-in.
assert_eq!(
s.eval::<String>(&format!(
"return {ADVGFX}RowWaterQualityDropdownText:GetText()"
))
.unwrap(),
labels[1]
labels[2]
);
assert_eq!(
s.eval::<String>(&format!("return {ADVGFX}RowLavaGlowLabel:GetText()"))
crates/benilla-app/src/video.rs
+2 -1
View File
@@ -404,7 +404,8 @@ impl Default for VideoConfig {
interior_bake_floor: 0.12,
fire_light_gain: 1.0,
spell_light_gain: 1.0,
water_quality: 1,
// GFX (water): High — the scenery reflection, measured free (`cvars::REGISTERED`).
water_quality: 2,
// MONKEY (volumetric fog): default to the inexpensive atmosphere.
volumetric_fog: 1,
// MONKEY (post): the shipped High graphics preset uses the full-resolution tier.
crates/benilla-assets/src/water_depth.rs
+5 -1
View File
@@ -35,8 +35,12 @@ impl Default for WaterUniform {
pub struct WaterQuality(pub u8);
impl Default for WaterQuality {
/// GFX (water): High. The scenery reflection was held back until its cost was measured, and
/// it measures as nothing — Lake Everstill runs 1.91 / 1.84 / 1.90 ms over 300 frames at the
/// three tiers and the ocean 1.89 / 1.88 / 1.89, inside the run-to-run noise. Kept welded to
/// `cvars::REGISTERED`'s `waterQuality` by `registered_defaults_mirror_the_code_truths`.
fn default() -> Self {
Self(1)
Self(2)
}
}
docs/GRAPHICS.md
+30 -5
View File
@@ -163,12 +163,37 @@ of the scene the PASS sees, not of the screenshot. The glow adds `w·blur²` dow
finished vista log-averages to 0.0931 linear and the pass's own input to 0.060. Calibrating against
the former aimed the lane 1.5x high and brightened the daylight exterior it exists to leave alone.
### 3. Planar water reflection
### 3. Water reflection — **the scenery reflection already existed; it was switched off**
Our water has refraction, foam and a scene-depth copy (`liquid/`, `water_fx/`) but no mirror. The
reference-era trick is a second scene render with an oblique near clip, armed only for visible water
and thinned of small objects, under a screen-space trace. It is the single most recognisable item on
the "original art, modern presentation" axis, and it is pure GPU — which is what we have.
The plan said "we have refraction and foam but no mirror". That was wrong: `waterQuality` 2 (High)
has carried a depth-marched screen-space reflection of the scenery all along
(`enhanced_water.wgsl`, "HIGH"), with bisection refinement against the real scene depth rather than
a fixed probe. What was missing was the switch — the registered default was Enhanced (sky only),
and only the Ultra lighting rung reached High.
The line holding it back said what it was waiting for: *"Mirror reflections stay opt-in until their
cost is measured."* Measured, `WOW_FPS_PROBE=300`:
| scene | Classic | Enhanced | High |
|---|---|---|---|
| Lake Everstill | 1.91 ms | 1.84 ms | 1.90 ms |
| open ocean | 1.89 ms | 1.88 ms | 1.89 ms |
All three tiers are inside the run-to-run noise — Enhanced even lands *faster* than Classic on the
lake, which is how you know you are reading noise. What it buys: the far bank and the mountains
appear in the water, **18.5% of the lake frame and 35.1% of the ocean frame**.
So High is now the registered default and the High rung, which is a weld of three places the tests
hold together (`REGISTERED`, `VideoConfig`, `WaterQuality::default`) plus the rung itself. Low and
Medium keep Enhanced deliberately: free on the machine it was measured on is an argument for the
default, not for the rung whose job is hardware that cannot afford what this one can.
**What a planar reflection would still add**, if it is ever wanted: the geometry SSR cannot see.
The march gives up when the ray leaves the screen (`probe.z < 0.5`) and the result fades toward the
top edge, which is exactly where a shallow-angle lake wants its far bank and trees. That is the
real remaining gap — and unlike everything above it, a planar pass draws the scene a second time,
so it is the first item on this list that costs **main thread**. Price it against the view-distance
table in item 1 before building it.
### 4. Froxel volumetrics