Reviewed the diff and surrounding code in src/gui/VfoWidget.cpp. This is a clean, focused fix and it looks good. ✅

What it does, and why it's correct:

• The three QGraphicsOpacityEffects are parented to their rows and installed at construction (VfoWidget.cpp:1021-1064), so they're enabled with default opacity 1.0 out of the gate — exactly the identity-effect-renders-blank condition under the #3695 GPU compositing path.
• The new setRowOpacity lambda resolves that by setEnabled(!fullStrength): at full strength the effect is disabled (a no-op, so the row paints directly), and it's only kept active while actually dimming to 0.45. That's the right shape for the bug.
• The null guard is preserved, the setOpacity → setEnabled ordering avoids any visible flash on the dimmed transition, and toggling setEnabled doesn't affect geometry/layout, so the existing relayout logic is unaffected.

Conventions: No AppSettings/QSettings surface here; effects remain parented (RAII intact); the no-capture lambda referencing the constexpr kDisabledOpacity is fine. Nicely de-duplicates the three near-identical blocks into one helper.

CI is fully green (build, macOS, Windows, CodeQL, Qt a11y static analysis). The only thing left is the unchecked box in your test plan — visual confirmation on a real radio with multiple flags up. Worth getting that on-hardware check before merge since the failure mode is render-only and not exercised by the build, but the code reasoning is sound.

Thanks for extracting this into a focused change, @ten9876 — the provenance note about dropping the already-landed #3751/#3752/#3753 churn is appreciated and made review easy.

_{🤖 aethersdr-agent · cost: $1.3340 · model: claude-opus-4-8}

Reviewed the diff against the surrounding code in VfoWidget.cpp. This looks good — clean, focused, and correct.

What's right

• The setRowOpacity lambda folds the three near-identical blocks into one helper with a single null guard each, and the setEnabled(!fullStrength) semantics are correct: a disabled QGraphicsOpacityEffect is a no-op, so a full-strength row paints directly instead of round-tripping through the identity-opacity composite path that was rendering blank. Keeping the effect installed-but-disabled (rather than uninstalling) is the right call — it stays cheap to re-enable when dimming kicks back in.
• These three effects (m_extremesSpeedFade / m_showValuesFade / m_txMeterFade) are created and setGraphicsEffect'd once at VfoWidget.cpp:1021-1064 and touched nowhere else, so toggling their enabled state here has no other consumers to surprise. Initial state is also handled, since syncSmartMtrSettingsState() runs the same path that disables a default-enabled/1.0 effect.
• kDisabledOpacity is constexpr, so using it inside the empty-capture lambda is fine (value read, not odr-used). Scope is exactly the one file the title claims; no QSettings/RAII/error-handling concerns at any system boundary here.

One non-blocking note

• The blank-row mechanism is a visual GPU-compositing interaction (#3695), and the test plan's "verified on a real radio" box is unchecked. The code change is sound regardless, but it'd be worth confirming on-hardware that a full-strength row actually renders before this is treated as fully validated — the failure mode is exactly the kind that only shows under real QRhi composition.

Thanks for the focused extraction from #3757, @ten9876 — narrowing it to just the net-new fix makes this easy to reason about.

_{🤖 aethersdr-agent · cost: $1.4617 · model: claude-opus-4-8}