Playtest 3.0.0-rc1: early release/camera/performance findings

[JoTalbot]

Raw playtest

• Build: 3.0.0-rc1
• Session: 1.1 min
• Device profile: 427×948 @1.0 (слабый)
• Smoothness: 44 fps, 210 тяжёлых кадров
• Native save: alive
• Touches: 23; gaze taken 20; growth 16; dropped 3; steps 0; void 3; pulses 0
• Drop reason: player released ×3
• Median drop: 0.43s; 1/3 almost immediate; birth at 1.35s
• Force: gaze 208, wounds 0, pulses 0, boss 0; fell to 100
• Camera distance: 0.99–1.06, avg 1.04
• Lifetime growth: 16 (all 16 alive)
• Returns: 0; anchors: 1 (1/3); rescued: 1
• Tide claimed: 0; metamorphosis carried away: 0
• Shores: 0; roads: 0; beings: 0
• Laws touched: 3; pulses 0; colors 0
• Sky opened: 0; sigil 0
• Nature: Тот, кто не повторяется · сбой
• Season: тишина, day 1, session 1

Initial interpretation

1. Critical UX/gameplay signal: 3/3 drops were caused by manual release, with median survival only 0.43s. The interaction currently allows a player to create failure too quickly and may not teach the intended timing/feedback loop.
2. Performance signal: 44 FPS on the weak 427×948 profile with 210 heavy frames is below a comfortable mobile RC target. Profile frame spikes and identify the dominant render/update path before store release.
3. Camera signal: zoom/distance is stable enough (0.99–1.06, avg 1.04), so camera drift is not the first suspect.
4. Progression signal: 16 growth events in 1.1 min, but zero steps/pulses and no later-system interactions. The opening may be over-indexed on gaze/growth before the broader loop becomes discoverable.
5. Save signal: native save is functioning in this session. Preserve this as a regression gate.

Proposed acceptance gates before final release

• Weak-device opening: remove/mitigate accidental-feeling early drops and improve feedback before the first 2 seconds.
• Weak-device performance: measure frame-time distribution, not only average FPS; eliminate the main source of heavy frames.
• First-session funnel: ensure the player can discover at least one non-gaze interaction during the opening arc.
• Native save/load remains green after performance and interaction changes.

This issue is a playtest-derived hardening item for the existing v3 roadmap, not a replacement for the long-form product backlog in issue #2.