Inverted lyrics

Use the user's footage as the visual and audio source. Render text as a mask selecting an RGB-inverted copy of the same frame: output = original * (1-mask) + (255-original) * mask. Do not substitute opaque white lettering. Outside the mask, preserve the footage.

Approved visual language

• Heavy, aggressive, non-serif condensed capitals. Impact is the default when available; keep the font configurable. Do not redistribute installed font files.
• Approximately 10% internal safe padding on all edges, not an added outer border. Preserve source framing and aspect ratio.
• Short phrase cards, usually two rows. Alternate a tall top row with a tall bottom row; the dominant row occupies roughly twice the height of the other. Occasional balanced arrangements are appropriate when phrase length needs them.
• Fit complete phrases before revealing them. Keep word positions fixed as new words appear. Reveal whole words at their sung onset, accumulate within the phrase, then replace the card at the next phrase. Never reveal an entire verse at once or re-center on every new word.
• Scale typography to fill the safe area, allowing deliberate condensed/stretched letterforms consistent with the approved treatment. Avoid decorative outlines, panels, shadows, or transitions unless requested.

Work sequence