The lit surface has to out-brighten the object's own face
This is a ratio, not an output figure. The band reads as a gap only when what is inside it is clearly brighter than the face bounding it, and the ratio can collapse from either direction.
It collapses downward when something else lights the front face to a similar level. It collapses upward when the strip is too strong: an over-driven run blows out the surface behind and spills light onto the face that is meant to stay dark. A modest output is plenty here, which is why something like the COB 200 Series at the low output end beats a brighter tape turned down, with lumens per foot confirmed there.
The band of light cannot break anywhere along the edge
One dark inch anywhere along the perimeter reads as the object touching down at that point.
Breaks come from four places. Segments cut and rejoined leave a gap: many strips can be cut at designated points and reconnected with compatible connectors or soldering, depending on the product, so check the interval and connector fit published for that tape. Corners are worse, because a mitered channel stops each run short of the true corner by the thickness of the extrusion wall. Fixings cast shadows, and end caps stop the light half an inch short.
An LED strip and its driver produce heat in normal operation, and an aluminum profile helps dissipate heat while keeping the tape straight enough to read as one line. No channel is mandatory. It also reduces but does not guarantee the disappearance of visible dots, with diffuser type, profile or channel depth and LED density all having a say. Compare depth across the recessed and surface profiles, and see the floating shelf build sequence for the build order.
Nothing else in the room may fill the gap
The effect is conditional on the rest of the room, the part designed last and the part that ruins the most jobs. A downlight grazing the wall pours light straight into the reveal, and daylight through a window on the right axis does the same for part of the day. A floor lamp two feet away lifts the face and the wall together.
Separate control is a requirement, not a refinement, which is the case for keeping the reveal on a separate zone. The supply, the controller and the strip all carry the same voltage rating, meaning 24V throughout for a 24V strip and 12V throughout for a 12V one. Where any of it will be dimmed, read the compatibility notes issued for the specific driver, wall device, controller and tape, since three dimmable labels are not an agreement. Reveals often send cable inside a wall, and that changes who does the work: in-wall cable, a hardwired driver, a junction box and any wall plate are a licensed electrician's responsibility, governed by the NEC where adopted, by whatever electrical code applies locally, and by the instructions issued with the equipment, with the breaker isolated and tested dead beforehand. Being handy on the 24V side is not the same qualification.
Testing the floating effect from where people actually stand
Mock it up before anything is fixed. Tape the real strip at the intended setback, hold the object at its finished height, kill every other light, then check all four conditions from the far seat, the near seat and the doorway.
Photograph each at eye height, because a camera at seat level records the sightline honestly. If the emitter shows in one frame the reveal is not deep enough, and depth is cheaper to change on a bench than after the millwork is scribed.