Add 'What's recorded' section: video specs, inputs, world state, g-buffers

- Home page: 4-card panel describing video (1280x720 @ 32fps near-lossless),
inputs, per-tick world state, and the upcoming g-buffers
- README: matching bullet list with full schema details

README.md

@@ -36,6 +36,17 @@ This dataset turns those demos into rendered, frame-accurate, fully-annotated tr
 - **Spatial-audio conditioning.** Audio is recorded per-player with HRTF positional cues, so models can learn to localize footsteps, gunfire, and callouts.
 - **Multi-agent training.** All 10 perspectives of the same round are kept aligned tick-for-tick — useful for collaborative policies, opponent modelling, and multi-player world models.
 
+## What's recorded
+
+For every chunk (≤ 1 minute, one player POV):
+
+- **Video** — 1280×720 @ 32 fps, near-lossless H.264. One stream per player; ten POVs per round all tick-aligned.
+- **Audio** — per-player stereo with HRTF positional cues (footsteps, gunfire, callouts).
+- **Inputs** — every tick: keyboard state, mouse delta, fire/jump/use, weapon switches.
+- **World state** — every tick, for all 10 players: position, velocity, view yaw/pitch, camera intrinsics, health, armor, ammo, primary/secondary weapon, alive flag.
+- **G-buffers (coming soon)** — per-pixel luminance, depth map, and vertex-ID map for self-supervised pretraining and dense prediction heads.
+- **Tick-perfect alignment** — every signal is sampled on the same CS2 tick clock. No drift, no resampling artifacts, no per-stream timestamp reconciliation.
+
 ## Dataset
 
 `blanchon/cs2_dataset_render`, derived from `blanchon/cs2_dataset_demo`. Licensed under `CC-BY-4.0`.

src/routes/+page.svelte

@@ -186,6 +186,61 @@
 		</div>
 	</section>
 
+	<!-- Recordings: what's in each chunk -->
+	<section class="mx-auto max-w-5xl pt-10">
+		<h2
+			class="font-heading text-muted-foreground text-center text-xs font-semibold tracking-[0.2em] uppercase"
+		>
+			What's recorded
+		</h2>
+		<div class="mt-6 grid grid-cols-1 gap-3 sm:grid-cols-2 lg:grid-cols-4">
+			<div class="bg-card rounded-md border p-3">
+				<div class="text-muted-foreground text-[10px] font-semibold tracking-wider uppercase">
+					Video
+				</div>
+				<div class="font-heading mt-1 text-sm font-semibold">1280×720 · 32 fps</div>
+				<p class="text-muted-foreground mt-1 text-xs leading-relaxed">
+					Near-lossless H.264, one stream per player POV. 10 streams per round, all tick-aligned.
+				</p>
+			</div>
+			<div class="bg-card rounded-md border p-3">
+				<div class="text-muted-foreground text-[10px] font-semibold tracking-wider uppercase">
+					Inputs
+				</div>
+				<div class="font-heading mt-1 text-sm font-semibold">Keyboard + mouse</div>
+				<p class="text-muted-foreground mt-1 text-xs leading-relaxed">
+					Per-tick keystrokes, mouse delta, fire/jump/use, weapon switches — the actions a policy
+					has to predict.
+				</p>
+			</div>
+			<div class="bg-card rounded-md border p-3">
+				<div class="text-muted-foreground text-[10px] font-semibold tracking-wider uppercase">
+					World state
+				</div>
+				<div class="font-heading mt-1 text-sm font-semibold">Per-tick game state</div>
+				<p class="text-muted-foreground mt-1 text-xs leading-relaxed">
+					Positions, velocities, camera intrinsics, health, armor, ammo, weapons — for all 10
+					players, every tick.
+				</p>
+			</div>
+			<div class="bg-card rounded-md border p-3">
+				<div class="text-muted-foreground text-[10px] font-semibold tracking-wider uppercase">
+					G-buffers <span class="text-amber-600 dark:text-amber-400">· soon</span>
+				</div>
+				<div class="font-heading mt-1 text-sm font-semibold">Luminance · depth · vertex IDs</div>
+				<p class="text-muted-foreground mt-1 text-xs leading-relaxed">
+					Per-pixel auxiliary channels for self-supervised pretraining and dense prediction heads.
+				</p>
+			</div>
+		</div>
+		<p
+			class="text-muted-foreground/80 mx-auto mt-4 max-w-2xl text-center text-[11px] leading-relaxed"
+		>
+			Every signal — video, audio, inputs, world state — is sampled on the same CS2 tick clock. No
+			drift, no resampling artifacts.
+		</p>
+	</section>
+
 	<!-- Stats -->
 	<section class="mx-auto max-w-5xl pt-10">
 		<div class="grid grid-cols-2 gap-3 sm:grid-cols-4">