Emoji in PIL-rendered video text silently becomes tofu boxes

When rendering on-screen text for video via PIL with fonts like Inter, emoji (trophy, eyes) render as tofu/box glyphs - PIL wont fall back to the system emoji font, and you only catch it by zooming into the settled caption frame. Strip emoji from any PIL text pipeline or composite emoji separately as images. Related iteration lesson: when a client keeps reordering shots, drive concat from an order list and derive every audio-cue timestamp from the cumulative-duration dict built off that list - reorders become one-line edits with SFX timing following automatically.