TASK 51 blur 'heavy' fix: quadratic blur radius. BlurRadius(strength) = 1 + 5s^2, MaxBlurRadius 12 -> 6 downscaled px (~24 full px max). Default 0.5 is now radius 2 small px (was the old linear 6) - ~3x lighter perceived spread, matching the creator's real-camera finding. Range from the established tools: StreamFX defaults its Box blur Size to 5 and its author stops downscale-based blurs around 8px. One integration test pins the mapping (0 -> 1, 0.5 -> 2, 1 -> 6) and step-edge behaviour (light reaches 9px at default vs 26px at full); the existing slice-2 blur test passes unchanged. 419/419 pass. Refs: https://github.com/Vhonowslend/StreamFX-Public/wiki/Filter-Blur https://xaymar.com/articles/2020/04/26/optimizing-streamfx-in-your-obs-scenes/ (Unit: bg-blur heavy fix)

ai.md

@@ -636,11 +636,14 @@ This replaces the old five-seeder cluster (`Seed{Starting,Brb,Ending,Chat}Backgr
   passthrough), `Remove` = copy with alpha = bilinear-upscaled mask × source alpha, RGB untouched,
   `IsOpaque = false` (the compositor's straight-alpha blend shows the layers below), strength =
   edge feather (mask band 0.5±(0.05+0.45·s) → alpha); `Blur` (slice 2) = background downscaled 4×
-  by box average, 3 separable box-blur passes (radius 1→12 small px by strength ≈ gaussian, up to
-  ~48 full px), bilinear-upsampled and blended by the mask — person pixels (mask = 1) are copied
-  untouched, opacity preserved. The mask is computed once per frame (`TrySegment`) and rendered per
-  effect (`Render`), so several variants share one inference. No mask / segmenter exception →
-  passthrough + one log line, never a dropped layer.
+  by box average, 3 separable box-blur passes and bilinear-upsampled, blended by the mask — person
+  pixels (mask = 1) are copied
+  untouched, opacity preserved. **Blur radius is quadratic** (`BlurRadius(strength)`: 1→6 small px,
+  default 0.5 = 2 — the "blur heavy" fix 2026-10-09; the old LINEAR 1→12 small px ≈ 48 full px was
+  far beyond what the established tools use: StreamFX defaults its Box blur Size to 5 and its author
+  stops downscale-based blurs around 8 px). The mask is computed once per frame (`TrySegment`) and
+  rendered per effect (`Render`), so several variants share one inference. No mask / segmenter
+  exception → passthrough + one log line, never a dropped layer.
 - **Config (slice 3):** per scene, on `WebcamSceneConfig` — `BackgroundEffect` + `BackgroundStrength`
   (0..1, default 0.5), persisted as two additive guarded columns on `WebcamSceneConfig`
   (`BackgroundEffect TEXT DEFAULT 'None'`, `BackgroundStrength REAL DEFAULT 0.5`, no version bump —
@@ -668,7 +671,8 @@ which also `ScheduleSave`s). Per raw frame a background worker (latest-frame-dro
   come from the established tools, not a guess: OBS's obs-backgroundremoval smoke-tests
   `temporal_smooth_factor` at its 0.85 default, Jitsi's webcam background filter (jitsi/jitsi-meet#17080)
   found 0.7/0.3 transformed best — a midpoint that still reacts to real movement; 0.5/0.5 leaves a
-  smearing edge ghost. Blur remains per-frame ("blur is heavy" = next work unit, on the radius curve).
+  smearing edge ghost. The blur "heavy" follow-up was fixed the same day: quadratic radius, 1→6
+  small px (see the `Blur` bullet above).
 - **Perf (synthetic 1280×720, Debug build):** segment 1.6 ms, Remove 3.3 ms, Blur 17.5 ms per frame
   (Release is several × faster). Pixel loops are row-parallel, capped at min(4, cores/2); the
   bilinear upsampling is separable (horizontal rows precomputed per mask/small row, one vertical lerp
@@ -689,6 +693,9 @@ which also `ScheduleSave`s). Per raw frame a background worker (latest-frame-dro
   — the ONE integration test for the jitter fix (a `SteppingMaskSegmenter` whose two consecutive
   masks disagree at an edge; asserts the second frame keeps a partial alpha where the raw mask would
   have cut to 0: x=6 → 58 upstairs of the raw 0, person core 255, background 0).
+  `Blur_default_strength_is_gentle_and_full_strength_is_strong` pins the quadratic radius mapping
+  (`BlurRadius(0.5)` = 2, `BlurRadius(1)` = 6) and the behaviour at a step edge: default smears
+  light ~9px into the dark half vs ~26px at full strength.
 
 ### Frame cadence — how a dropped frame became measurable (UNIT 0, 2026-10-05)