# Tweet variations

All three are drafts only. None was posted.

## Variation 1

Draft—unpublished. A real LTX-2.3 Clean Plate V2V test on dense Times Square footage removed much of the crowd while keeping the scene and camera move. Dark residuals remain. No 4:3 showcase: the production kit lacks a truthful V2V template. Footage: FASTILY, CC BY-SA 4.0.

## Variation 2

Blocked showcase draft. I sent a 2.79s CC BY-SA Times Square clip through the live Lightricks clean-plate Space—no mock. It returned 65 frames at 24 fps and cleared most pedestrians and cars, but left artifacts. Publication is withheld: no supported source-video→video layout.

## Variation 3

Visually reviewed, unpublished test: crowded Times Square became mostly empty streets through the real `/generate` endpoint. Not a finished demo: 480×256 output retains visible remnants, and the renderer has no honest V2V layout. Footage: FASTILY / Commons, CC BY-SA 4.0.