New in v5 · Game IQ

Meet Game IQ

On-device learning that finds where each game keeps its HUD, then brings protection back to those regions faster the next time they appear. No waiting for risk to build back up after a goal, a menu, or a respawn.

88Game IQ · RECOGNISING3 / 3 LOCKED

Your HUD never moves. That’s the danger.

On an OLED, the pixels that wear are the bright ones that stay put: health bars, ammo counts, minimaps, score readouts. They burn in while the action behind them keeps the rest of the panel safe. The hard part isn’t dimming a static screen; it’s protecting a HUD that vanishes and returns every round, on a panel that’s otherwise in constant motion. That is exactly what Game IQ is built for.

The architecture

Four stages, one loop-free pipeline

Detect → Enroll → Recognize → Protect. Each stage answers one question with a signal robust enough to survive real gameplay.

BURN PRESSURE = LUMA × STATIC42centre: moving · stays coolHUD (static + bright)gameplay (moving)VERDICTprotect HUDs
01
Detect

Measure burn pressure, not pixels

Every few seconds the engine takes a downsampled luma + motion frame and computes, per cell, a burn pressure = brightness × how-static-it-is. A moving battlefield stays cold; a fixed health bar, ammo counter or boost ring heats up. That pressure map is the raw material, and it is derived only from screen content, never from the protection Game IQ outputs.

GPU-side downsample · loop-free by construction

PRESENCE RATIO · SEEN-PER-ACTIVE-FRAMEgateboost metertop-right textbottom-leftcrowd / lightsyour car3 real HUDs confirmed · scenery rejected
02
Enroll

Learn which spots are really HUDs

Lots of arena scenery is briefly bright-and-static too: the crowd, the goal, your own car. Game IQ separates a real HUD from scenery with a presence ratio: a true HUD is present on almost every active frame, while scenery is framed there only sometimes. Locations that clear the gate get confirmed; the rest are rejected. Identity is a screen location, so a boost meter whose number changes every frame stays one memory instead of fragmenting into dozens.

per-app · persisted locally · playtime-independent gate

CONFIDENCE THROUGH A SCENE CUTSCENE CUTnaive (motion) · dropsGame IQ (luma-stable) · holdsprotectedoff
03
Recognize

Know it is there, even across a scene cut

A goal replay or menu wipes the engine's temporal state, and a naive motion-based check reads max-motion for a moment and drops every HUD at once. Game IQ asks a different question: is the bright skeleton at this spot the same as a moment ago? It uses only luma, the one signal a scene reset can't corrupt. So protection holds through the cut and releases cleanly when the HUD genuinely leaves for a menu.

reset-immune presence · whole-screen-static menu gate

RE-PROTECTION AFTER THE HUD RETURNSHUD returnsbase detector · ~30 sGame IQ · learned floorprotectedoff
04
Protect

Re-shield as soon as it returns

When the HUD comes back after a goal, the base detector needs about 30 seconds of risk to re-accumulate before it protects again. Game IQ already knows the spot, so it re-applies a protection floor as soon as it recognises the HUD again. The floor is max-only: it can raise protection but never lower what the base engine already decided, so the memory can never weaken anything.

learned floor · max-only merge · never reduces base protection

The hard parts, solved on principle

A HUD memory is easy to get wrong: it can chase its own output, cover the wrong thing, or drop everything at a scene cut. Game IQ avoids each failure by construction, not by tuning. Why is any of this hard?

Loop-free learning

Every signal Game IQ learns from is derived from screen content: luma, motion, location, time. It never reads its own overlay output, so the system can't chase its own tail.

On-device & private

Learning runs entirely on your PC. The confirmed HUD set is stored locally as non-reversible fingerprints. Nothing about what you play ever leaves the machine.

Reset-immune by design

Presence is read from current-vs-previous luma, the only channel a goal-cut scene reset can't corrupt. A reset spikes motion high, which keeps the floor up instead of dropping it.

Per-game memory that persists

Each title gets its own learned HUD layout, saved across launches. A known game starts protected sooner next session, using what it already learned, with no fresh learning window.

One-way
safety floor
Faster
re-protection
100% local
on-device
Per-game
learned layout
Scope & safety

Built to add protection, never take it away

Game IQ is a narrow, one-way safety layer. It is gated to gameplay, it can only raise protection, and it hands back to the base engine the moment it is unsure.

Gaming preset + fullscreen only

Game IQ only learns and applies its floor in fullscreen Gaming sessions. Your desktop, media, and menus never pick up a game-HUD floor by mistake.

A one-way floor

The learned floor can only add protection on top of what the base engine already decided. It never lowers protection, so a learned region can't weaken anything.

Falls back to the base engine

When Game IQ is unsure, the base risk engine keeps protecting everything exactly as it always has. Game IQ is a bonus layer, not a replacement.

Rejects false positives

Size caps reject whole panels that aren't HUDs, and co-presence demotion retires small regions once real HUDs are confirmed, so scenery never gets locked in.

Local and private

Learning runs on your PC and the learned HUD set is stored locally. Nothing about what you play leaves the machine.

Resets cleanly

Reset and factory reset clear the learned HUD files, and a corrupt or mismatched store falls back to a backup instead of misfiring.

Smarter protection,
automatically.

Game IQ ships in OLED Guard Pro v5. Toggle it from the dashboard and it learns quietly while you play in fullscreen Gaming sessions, brings protection back to the regions that matter, and never sends a thing off your PC.

Learns per game Faster re-protection On-device & private