Cast and contact shadows declare depth better than transparency
Aliases: cast shadow · drop shadow · attached shadow
What it is
To make a panel read as sitting on a table or stopping in front of a hand, turning it translucent only makes it look like a film. A contact shadow and a cast shadow state that this volume meets that surface: the shape and hardness of the shadow pin the depth. Translucency only says light from behind still arrives. It does not say where the panel itself sits — both layers can be seen and front/back can still be judged wrong.
Why it happens
A contact shadow is the thin, dark, hard occlusion between an object and its supporting surface. It appears almost only on true contact, which is why it is visual evidence of “resting on”. A cast shadow gives light direction and height off the surface: the farther the shadow from the contact point, the more the object floats. Translucency says “the rear is not blocked” and lifts the object out of the scene’s light field instead of binding it in. People can see both layers at once and still fail to say which is nearer. A shadow in the wrong direction is worse than none — it presses the object onto the wrong plane. Under indoor bounce light, cast shadows go soft and faint; contact shadows often remain, so contact survives bad lighting estimates better than a large falloff.
Studying it
A placement match: people put a virtual panel “onto” a table or wall under contact shadow, cast shadow only, translucency only, or none.
Independent variables: shadow type, translucency on or off, surface texture. Dependent variables: height error, accuracy of a “is it touching?” judgement, adjustment time.
AR placement tasks and visuo-haptic alignment belong here. Hard shadows from a laboratory parallel light overstate how much information a shadow carries; accept in both indoor bounce and window side-light.
Where it stops holding
Optical see-through can rarely paint a shadow darker than the real environment (it adds light, it cannot subtract), so contact shadows are often simply absent and the cue is void. Video see-through can darken. With no stable lighting estimate, do not draw shadows. A white void or a skybox has no receiving surface. At long range a contact shadow shrinks below visibility and relative size plus occlusion take over.
Applying it
- To declare “on this surface”, draw a contact shadow; do not borrow translucency for the job.
- To declare “how far it floats off this surface”, use the gap in the cast shadow, not alpha.
- Keep translucency for “pass through / see behind”. Do not use it as a depth label.
- How to check: kill every shadow, leave translucency, and ask someone new whether the panel sits on the table or hangs. Unstable or systematically wrong answers mean depth was being carried by the shadow.
Related
- Same group: N3.06.1 A semi-transparent panel mixes with the background · N3.06.2 Occlusion is the strongest depth cue and must not be violated · N3.06.3 Transparency must adapt to background complexity · N3.06.5 Transparency carries a pass-through meaning, not just a visual style · N3.06.6 Stacked semi-transparent panels multiply opacity · N3.06.7 On optical see-through, black equals transparent; a dark UI cannot hide reality
- Nearby: N3.03 Depth Conflict · N5.01 See-through Modes
- Search terms:
contact shadow·cast shadow·depth cue
Cards in the same group
- N3.06.1A semi-transparent panel mixes with the background
- N3.06.2Occlusion is the strongest depth cue and must not be violated
- N3.06.3Transparency must adapt to background complexity
- N3.06.5Transparency carries a pass-through meaning, not just a visual style
- N3.06.6Stacked semi-transparent panels multiply opacity
- N3.06.7On optical see-through, black equals transparent; a dark UI cannot hide reality