Plugins + Scripts [ForgeLab] Glass Forge

GLASSFORGE v1.1
The fog-only render gets cheaper, a stray camera stops triggering it, and the re-fog scan stops running when there is nothing to re-fog.​

---

In one screen

	v1.0	v1.1	change
Pane with condensation + drips, refraction off (two-person field scene, 8K, over the empty floor)	+1.86 ms	+1.29 ms	−31%
The transmission render itself	1.61 ms	1.04 ms	−35%
In the whole scene (fog, hero light, water present)	15.75 ms	14.84 ms	−0.91 ms : the hero light's shadow map inside the render goes too
The whole field scene, all three Forge plugins at factory settings	36.35 ms · 27.5 fps	12.53 ms · 79.8 fps	−65% (with LumenForge v1.4 and SteamForge v1.2)

No default changed its meaning. One new switch, on by default. Saved scenes are untouched.

---

Changelog

Change	What you see
New switch : Cheap transmission behind heavy fog (ships ON). With refraction off and Fog Amount ≥ 0.6, the transmission render that feeds the fog blur runs without shadow maps and without per-pixel lights.	Nothing : the blur samples that render at a depth where neither survives. Never applied while refraction is on or the film is thin.
Re-fog decay scan runs only while some texel is still wiped	Nothing. After the first touch the four-times-a-second scan of the whole mask used to run for the life of the pane.
Orthographic cameras are not treated as viewers	Nothing on screen. LumenForge's depth capture (SkyRay, EnvRay, CUARay) could see the pane and trigger a full transmission render of its own; it is now skipped.

---

QA : the rows

Two persons, EnvFog, HeroRay, water. 8K DSR, RTX 5090, six passes of ten seconds per row with a four-second settle, VAM panel closed, camera pinned by the scene's spawn point.

GlassForge alone, over the empty floor	ms
Condensation 0.88 + drips, cheap render OFF (v1.0 behaviour)	+1.86
Same, cheap render ON (v1.1 default)	+1.29 · −0.57
Condensation 0.50, cheap ON (below the threshold : the full render must run)	+1.87 · the control row, +0.01 vs OFF
Condensation off	+0.36 : the resident screen-grab copy and the pane's own fill
Drips off (fog on)	−0.08 : drips are close to free
Refraction on, Screen Grab source (no transmission render at all)	+0.50
Transmission resolution 512 / 1024 / 2048	+0.01 / — / +0.12 : the render's cost is what it submits, not its pixels
Reflection on, 1024 · 2048 · 4× MSAA	+1.14 · +1.25 · +1.27 over the pane : about one scene render each

Cross-plugin checks	result
Fog + glass vs the sum of each alone (transmission 1024)	−0.10 : additive. At 2048 the transmission render re-marches the fog for +0.80, which SteamForge v1.2's Aux camera quality now caps.
Fog + glass with reflection on, vs the mirror render alone	−0.19 : no fog leaks into the reflection (fog sits on the engine's mirror-exempt layer).
Hero light + glass vs the sum of each alone	+0.19 : additive at 1024.

Gate : offline compile check clean; the threshold control row is the test that could have failed and did not.

The field scene, every sweep table and the raw rows are archived with LumenForge's dev logs. Nothing here was estimated.

---

Updating scenes you already published

A scene saved against Shadow Venom.GlassForge.1 keeps loading v1.0 even after you install v1.1, because VAM pins script references to the exact version. Re-point the plugin to GlassForge.2 in the atom's Plugins tab (or re-add it), then save. Nothing else is needed for GlassForge : the new switch is on by default and no saved value changes meaning.

Every ForgeLab package also ships ForgeLab_Updater.cslist, a temporary tool that moves EnvFog Quality and HeroRay Beam steps in the same scene onto their new factory values (see the SteamForge v1.2 and LumenForge v1.4 notes).

---

Known limits

• Every camera that can see the pane still gets its own transmission render (a WindowCamera, a VAM mirror atom). Gating that to the main view is on the list; it could not be measured in the field scene, which has none.
• The screen-grab copy is still taken even when the transmission render is in use. Removing it needs a second shader variant and is on the list.
• The cheap render is a threshold, not a slider. Below Fog Amount 0.6 the film is thin enough that the full render is the right one; if you want the saving with a thinner film, that is what the switch is for, at your own eye's risk.