297 gate-passed questions · 40 cells · 14 occupied · go to this type's cells
Bins were cut on this corpus at 240×135@16 (CLEAR_PX 25, motion floor 19 px). Margin bins: a near-edge band [1, 1.25) above the gate boundary, then corpus quantiles; salience bins: a visibility floor, then quantiles.
per class: pan turn/5°, orbit turn/10°, dolly |Δdist|/0.15; static: one bin · measurable on 212 questions
| margin bin | numeric range |
|---|---|
| near-edge [1,1.25) | [−∞, 1.25) |
| bin0 | [1.25, 2.448) |
| bin1 | [2.448, ∞) |
| state:static | (state / single bin) |
Ask: is this the same quantity the gate certifies on for this type? Are the states (e.g. 'same', 'gone') the right one-bin exceptions? Are the edges where a change in difficulty actually happens?
background flow magnitude (battery flow_bg_mag_mean_fw) · floor 0.0007716
| salience bin | numeric range |
|---|---|
| <floor (0.0008) | [−∞, 0.0007716) |
| bin0 | [0.0007716, 0.008436) |
| bin1 | [0.008436, 0.01394) |
| bin2 | [0.01394, ∞) |
Ask: does this axis capture what makes a question of this type perceptually hard, independently of the answer? Is the floor the right anti-gaming boundary?
answer (branch 1) · labels: moving straight toward or away from the scene, orbiting around the scene, static (not moving), turning in place (panning)
Compatible margin bins per context label (cells outside these are definitionally impossible):
| answer | n |
|---|---|
| orbiting around the scene | 92 |
| static (not moving) | 85 |
| turning in place (panning) | 65 |
| moving straight toward or away from the scene | 55 |