57 lines
4.0 KiB
Markdown

# Ansh Week 4 — Scene Findings Synthesis
## Executive Summary
- Thob is consistently strongest at fast visual scene setup and no-code authoring for basic 3D flows.
- Hierarchy fundamentals are solid: parent-child-grandchild inheritance is reliable when nesting is set up cleanly.
- Discrete state switching is reliable when built as pre-authored states and toggled via visibility.
- The biggest blocker across Week 2 and Week 3 is interaction orchestration, not raw rendering or hierarchy.
- Camera workflows split into two realities: single static camera framing is stable, but multi-camera preset switching is fragile.
- Guided storytelling is possible, but often workaround-based (prebuilt states + visibility toggle) instead of first-class orchestration.
- Output quality is limited by missing transition primitives, making state and step changes feel abrupt.
## Strongest Scene / Storytelling Flows In thob
1. Prebuilt scene states with visibility toggles
- Most repeatable flow across Week 3: deterministic state changes, reliable resets, and stable repeated switching.
2. Structural hierarchy flows (parent -> child -> grandchild)
- Strongest technical foundation from Week 1 and Week 3: inherited transforms stay predictable and offsets remain intact.
3. Single-camera static framing
- Reliable for default scene composition and non-dynamic camera use cases.
## Weakest / Most Fragile Scene Flows In thob
1. Direct multi-camera preset switching with Make Default
- Repeated toggles can become inconsistent; selected preset does not always reliably become final visible state.
2. One-click multi-target updates (camera + visibility + transforms)
- Button mapping for coordinated changes is hard to wire and does not always feel atomic/synchronized.
3. Camera rotation-driven storytelling
- Camera rotation behavior was reported as unstable in Week 2 (objects disappearing), blocking confidence in rotation-led narratives.
## Recurring Scene-System Pain Points
- Repeated orchestration issues
- Interaction binding model is not explicit enough for complex flows.
- Authors need to map one trigger to many targets, but discoverability and reliability are weak.
- Repeated state issues
- Native bundled state orchestration is missing, so users rely on workaround architectures.
- State switches are reliable only when represented as separate prebuilt states, not as a true shared-state transition system.
- Repeated hierarchy/grouping issues
- Core hierarchy behavior is strong, but transform debugging remains less transparent than needed.
- Local vs inherited transform visibility still needs clearer editor feedback for deeper nested scenes.
- Repeated transition quality issues
- No first-class interpolation controls for camera/object state changes; instant jumps reduce storytelling polish.
## Missing Product Primitives
- Camera preset system
- Named presets (position + target), deterministic switching, and clear active-camera feedback.
- Scene preset model
- First-class scene state objects instead of only visibility-based workarounds.
- Guided step flow primitive
- Native step bundles that can update camera, visibility, and transforms together with reset/back/next behavior.
- Atomic multi-target action system
- One action should update all mapped props in one synchronized transaction.
- Transition primitive
- Duration/easing controls for camera and transform interpolation between states.
- Transform inspector primitive
- Local/world transform clarity and inheritance path visibility for nested hierarchies.
## Most Important Final Takeaway
- If only one scene-system improvement is prioritized first, it should be a first-class orchestration layer: one trigger -> one bundled state transition -> many synchronized updates (camera + visibility + transforms) with optional duration/easing.
- Why this first: it directly resolves the most repeated cross-week pain (interaction reliability), unlocks guided storytelling quality, and builds on existing strengths (fast setup + stable hierarchy) instead of replacing them.