Execute immediately. Do not restate the prompt, explain capabilities, offer…

Execute immediately. Do not restate the prompt, explain capabilities, offer options, or ask questions. Use the uploaded Leaning Tower of Pisa construction diagram as the only input image. Generate one 15.0-second Seedance 2.5 image-to-video source clip, native 16:9, 720p. This is an architectural construction-teaching video. Preserve the warm-white studio background, marble materials, engineering-diagram aesthetic, lighting, framing, and all existing shapes from the input image. Do not redesign or invent a new tower. Asset roles The partially built circular base at the lower center is the fixed construction anchor. It includes the existing stairs, entrance, cutaway masonry, and foundation. It never moves, rotates, scales, or breaks apart. Only these existing 3D elements are allowed to assemble: the curved column module and curved arch module at mid-left; the five C-shaped masonry-and-gallery modules in the center-left column; the five matching open arcade rings in the center-right column; the upper-right transition platform, bell-chamber arch ring, and finished bell-chamber crown. Everything else is a teaching-reference layer only: the elevation, plans, sections, diagrams, callouts, labels, leader lines, and dimension marks; the upper-left material samples; the separate entrance sample and spiral-stair sample; the loose group of seven bells at lower right. The teaching-reference layer never joins the building. It only fades out and later fades back in at exactly its original position. Its own shadows fade with it. Motion rules This must feel like the Huanghelou construction video: only one construction group moves at a time. The active group stays at full brightness; all waiting construction groups remain visible but dimmed to about 60% brightness. Every moving group keeps its original material, scale, orientation, and rigid shape. It moves along one short, clean, unobstructed path toward its final interface. No random flight, free rotation, melting, morphing, stretching, duplicate copies, or geometry left behind at the original location. Each group eases in, slows down before contact, locks perfectly into place, and pauses for about 0.12 seconds before the next group starts. The final tower follows the leaning direction indicated by the red “actual axis” in the left elevation drawing. The lean is stable architecture, not a collapse: do not straighten it, increase the lean, wobble, bend, or shake it. Source-clip timeline 0.00–0.55s — Hold the first frame Keep the full diagram perfectly still so the viewer can read the exploded structure. 0.55–1.15s — Reference layer fades out Fade out all diagrams, material samples, loose entrance and stair samples, loose seven-bell group, labels, callouts, and leader lines. Do not fade the warm-white background, the fixed central anchor, or the actual construction modules. 1.15–2.05s — First arcade level Move the curved column module into the open receiving interface on top of the fixed anchor and lock it. Then move the curved arch module into the same level, completing the first arcade layer. 2.05–7.05s — Five upper arcade levels Build the remaining five levels from bottom to top. For each horizontal pair: Move the matching C-shaped masonry-and-gallery module into place first. Move the matching open arcade ring second. It first aligns on the same leaning local axis, then descends coaxially around the masonry module and locks. Complete one full level before starting the next. Do not let upper layers move early. After this stage, the tower must show exactly six open arcade levels in total. 7.05–8.45s — Bell chamber Move the circular transition platform onto the sixth arcade level, keeping its central opening aligned with the tower. Then install the bell-chamber arch ring. Finally install the finished bell-chamber crown containing the bell already visible in the reference image. The loose seven bells remain part of the teaching-reference layer and never fly into the tower. 8.45–9.20s — Clean completed tower Hold the completed Leaning Tower of Pisa completely still on the warm-white background. No diagrams or reference samples are visible. 9.20–10.30s — Teaching layer returns While the completed tower remains absolutely motionless, fade the original diagrams, material samples, loose stair and entrance samples, loose bells, labels, and callouts back to their exact first-frame positions. 10.30–15.00s — Stable display Keep the completed tower and restored teaching-reference layer perfectly still. Final 15-second loop After generating source clip A, use the same project timeline: Keep source A from 0.00 to 10.30 seconds at normal speed. Duplicate that exact 0.00–10.30-second section. Mute the duplicated section. Reverse it and speed it up evenly to 4.70 seconds. Place it immediately after 10.30 seconds. Use a direct same-frame cut at the join: no dissolve, flash, black frame, or transition. Do not generate a second AI video for the disassembly. The reverse section must naturally show: reference layer fades out → bell chamber leaves → the six arcade levels leave in reverse order → the first arcade modules return → all construction modules return to their exact first-frame positions → teaching layer restores → final frame matches the uploaded first frame. The reverse disassembly should feel fast and exciting, but never like an explosion. Every module follows its exact original path backward. Camera, background, and audio Lock the original camera position, focal length, perspective, horizon, roll angle, exposure, and frame. No zoom, orbit, pan, tilt, push-in, pull-back, or camera correction of the tower lean. Keep the warm-white background, floor, shadow direction, and brightness continuous for the entire video. Never use black, dark gray, vignetting, a new location, a white flash, or a full-frame fade-out. No voiceover, dialogue, subtitles, lyrics, logos, watermarks, platform UI, or end card. Use only a subtle continuous room tone and very light stone-contact sounds. Mute the reversed clip audio. Deliver one final 15.0-second native 16:9 video and report the result briefly in Simplified Chinese.