90s HIP-HOP BLOCK PARTY @image1 throws down a classic 90s hip-hop set at a sunny block party — the running man, the roger rabbit and the cabbage patch, bouncing on a fat boom-bap beat with old-school swagger. Music only, mixed loud. One continuous high-energy take on an outdoor court with graffiti and a boombox. Plug in ANY avatar and let them own the block.

SLOGANS — edit these words and nothing else: BEAT 1 WORD = MAKE IT LOUD BEAT 2 WORD = GOLD REVOLUTION BEAT 3 WORD = MOVE FAST HERO WORD = BOLD BEAT 5 WORD = NO LIMITS BEAT 6 WORD = CREATE NOISE HERO WORD is used twice: once in beat 4 and again on the closing frame. A polished stylized 3D animated film of the kind used in high-end animated features and AAA game cinematics, with hand-painted surfaces over believable 3D forms, physically lit, real depth of field. Premium fashion campaign pacing, energetic but controlled. @image1 is an absolute character lock. Face, body, hair, clothing, materials and colours are taken from it exactly, with no drift or reinterpretation in any frame. @image2 is the storyboard and is followed exactly: its six panels are the six beats in order, and each beat matches that panel's shot size, camera position, figure placement, colour system, type voice and type layout. This is not a film in which the background slides. The typography is the moving force: the letters themselves animate, and the walk is what triggers them. One continuous walk, no cuts. The figure walks toward the camera with their face to the lens, travelling toward the lens for the whole clip, the camera retreating ahead of them at their own pace with a slow drift to one side. Beats, each given equal weight, one word event per beat: BEAT 1 WORD writes itself on stroke by stroke, then expands outward, some letters stretching tall and others snapping into place, arriving in layers — one letter passing behind them while another crosses in front. BEAT 2 WORD bursts in as separate pieces from several directions, overscaling and rotating slightly before locking together, with a block of colour landing behind it and torn paper edges lifting. BEAT 3 WORD rushes through the frame, the letters stretching forward in perspective and snapping back, the nearest streaking past the lens while others slam into place beside them. HERO WORD assembles in segments, splitting briefly into a near layer and a far layer so part of it stands in front of them and part behind, then settling into one controlled sculptural composition. BEAT 5 WORD hits large and stacks, fragmenting like torn paper at one corner, then recomposing into a clean arrangement as photographic fragments, ink strokes and halftone texture gather around it. BEAT 6 WORD surges toward the camera, stretches across the frame, breaks apart, collides and reassembles, every earlier motif layering back in, and everything locks into one final poster composition with HERO WORD standing large and complete behind them as they keep walking. Throughout, the letters behave as physical objects with depth: expanding, compressing, rotating in space, slicing across the frame, wrapping around their path, pushing toward the lens and receding, and passing both in front of and behind them. Type events land on the footfalls. The clothing moves with the walk — the open front of the outer layer swinging, hems and cuffs settling, everything coming to rest naturally between steps. Every movement runs forward in normal time: the subject always travels in the direction their body faces, their feet lead their motion, and nothing moves in reverse. Sound is the ambience of the space in @image2. No cuts, no camera shake, no glitch effects, no identity drift, no face change, no garment substitution, no lettering in any script other than Latin, no words beyond the slogans listed above, no sliding-background-only motion, no static type sitting flat on a surface, no sliding feet, no walking backwards, no warped hands, no extra limbs, no subtitles, no logos, no watermark.

3 steps6316.5K

Your avatar runs from a giant monster in the dark, gets caught, and the clip ends on a horrifying close reveal of the creature. Upload your own avatar (@image1) and monster (@image2). Music only, loud. See the video step for the full prompt.

Your avatar becomes a pale, fully translucent ghost, drifts off the floor, and fades from view at the end. Music only, loud. See the video step for the full prompt.

Photoreal live-action capture, clean modern camera clarity, natural warm interior daylight, gentle shallow depth of field, neutral true-to-life colour grade. No stylisation, no illustration, no animation. @image1 (before-state reference) is an absolute lock for the first half: face, hair, body state, wardrobe, fabric and colour are taken from it exactly, with no drift or reinterpretation. @image2 (after-state reference) is the same absolute lock for the second half. It is the same person in both halves - the face must read as identical across the two. @image3 (room plate) sets the location, camera position, framing, lens and lighting, and these stay identical throughout, unchanged across the cut. The camera is locked off and never moves. One static medium shot, framed from mid-thigh to just above the head, the figure standing centred in the empty floor space with the bed and headboard behind. First half. The figure stands centred with its back to the lens. It begins one slow continuous turn in place, unhurried and even in speed: the spine rotates away, a three-quarter back view passes, then a clean side profile, then it arrives square to the lens looking straight down the barrel. Without pausing it carries the same rotation onward - the opposite profile passes, then a three-quarter back view, then the full back again - and rolls on to settle into a held side profile. Arms hang loose the whole way, hands relaxed, shoulders level, hair swinging slightly with the rotation. Then a hard cut. No dissolve, no wipe, no morph, no transition of any kind. Framing, camera position, lens, lighting and room are identical either side of it. Second half. The figure now stands open to the lens in a relaxed three-quarter stance, weight settled on one leg, and does not turn away again. It presents rather than inspects: small unhurried shifts of weight from one hip to the other, the head turning slightly toward the lens and easing back again, a slow settle of the shoulders, quiet natural breathing, hair moving a little with each turn of the head. The body stays open to camera from the cut to the end. On-screen text, both halves: plain white sans-serif, centred over the chest, no box, no outline, no bar behind it. In the first half the word is 69kg - it is there from the opening frame and fades cleanly once the figure has come round to face the lens. In the second half the word is 56kg - it appears on the first frame after the cut and fades cleanly once the stance has settled. Nothing else is written on screen at any point. Audio: a soft, low, even instrumental music bed only. Exclusions: no speech, no dialogue, no voice-over, no narration, no lyrics, no sound effects. No captions or subtitles beyond the two weight labels. No camera movement of any kind - no push, no pull, no pan, no tilt, no orbit, no handheld drift, no zoom. No cut other than the single hard cut between the halves. No second person in frame. No logos, no watermarks, no product, no end card.

4 steps736.4K

@image1 is a 3x3 storyboard reference. Its nine numbered frames define the order of the assembly, the studio viewpoint and the miniature model. The video plays these steps as one continuous full-frame vertical image, without the grid, gutters or numbers. @image2 is the closing-frame reference. It shows the finished full-size space exactly as the last frame must look, with the same framing, layout and lighting, and the video ends on a frame identical to it. Every element of the space, its design, material, colour and placement, comes from @image2, first at miniature scale and then at full size. In one continuous 15-second take with no cuts, one pair of realistic adult hands builds a miniature version of the space in @image2 step by step, in the order of @image1, on a seamless pure white studio sweep. Then a seamless scale match turns the miniature into the full-size space, and the video ends on @image2. Realistic live-action footage, crisp neutral whites, fine sensor grain. During the build the white sweep fills the frame edge to edge around the model, equally bright in every corner, flooded with strong, fully diffused light from every direction at once, as inside a large white light tent. Every surface of the model and the hands reads bright and evenly lit with true colours; only a faint soft darkening sits in the seams where two parts meet. Rules for the whole take: - The only body parts ever in frame are the two hands and their forearms, bare skin from the fingertips to the frame edge, with no sleeves, cuffs, gloves, rings, watches or bracelets. Each arm enters from the top, left or right edge and is cut off at mid-forearm by that edge. - Exactly one left hand and one right hand do all the work. Every part is gripped, carried and set by a single hand, and the two hands never hold the same part together. While one hand sets a part, the other carries the next part or rests at the frame edge. The hands move with quick, precise movements, and each finishes its placement before reaching for the next. - Every element enters only at the moment a hand places it. Until then its spot stays empty, and once placed it stays exactly where it was set. - The image stays clean from first frame to last: no text, titles, captions, numbers, logos, watermarks or overlays. - Every movement runs forward in normal time, and nothing moves in reverse. A wide shot opens on the plinth carrying only the bare floor, standing on the white sweep. The camera looks down at about 30 degrees, frontally and slightly from one side, matching @image1, and makes one slow, steady dolly push-in through the whole take. 1. A hand lowers the first bare wall onto the plinth edge and presses it down, and the remaining bare walls follow, one per hand. 2. The large built-in and fixed elements stand into place against the walls, one at a time. 3. The surfaces and fixed finishes that lie on top of them are lowered and pressed flush. 4. Appliances, fittings and fixtures are pushed and clicked into place, still unlit. 5. Free-standing furniture and any partitions or dividers are set down. 6. Lighting, textiles, rugs and plants are hung and placed. 7. The hands take turns placing the last loose objects and decor, each carrying one piece at a time. Then both hands withdraw past the frame edges and the model's lights glow on. 8. The frame holds only the model, with no body parts in view. The camera eases down to standing eye level and moves forward, and the miniature grows seamlessly into the full-size space. Only during this scale change does the white studio sweep give way to the real surroundings of @image2, its ceiling, its outer walls and its openings, as the light settles into the light of @image2. The shot lands on @image2 exactly and holds, with only small natural ambient motion. Steps 1 to 7 share the first twelve seconds evenly; step 8 takes the final three.

3 steps244

Your avatar fights a losing battle as they turn undead, ending as a frightening veined zombie. Music only, loud. See the video step for the full prompt.

Keep the SECOND HALF of the last Fast take. Change only the opening. BACKGROUND: soft blurred forest bokeh behind everything — green trees out of focus, dappled light. Not a white studio. Not the background from @image1. DO NOT OPEN on an empty cup. DO NOT open with a splash ring around an empty cup. @image1 is THE CUP — photocopy the cup and drink exactly as shown — cup type, layers, syrup, cream and drizzle. IGNORE any caption, text or background in @image1. 15 seconds, 9:16. Seedance 2.0 Fast. No people except the cream-can hand at the end if needed. No branded carton. No leaves garnish. AUDIO: powder, pour, cream hiss. No speech. 0.0–3.0s START DIFFERENT: a pool or stream of WHITE MILK already in frame, forest bokeh behind. Fine MATCHA POWDER falls INTO the milk (powder into milk, never milk dumped onto dry powder). Green particles sinking and blooming in the white liquid. No cup yet, or cup already filled in the background — never empty. 3.0–6.0s Milk + matcha swirl, then a ribbon of RED STRAWBERRY SYRUP joining. Macro, forest still blurred behind. 6.0–9.5s KEEP: the cup filling — the drink, layers and syrup streaks exactly as in @image1. Cup is never empty here. 9.5–13.0s KEEP: SLOW-MO whipped cream from a cream CAN / dispenser coiling onto the cup, then the drizzle from @image1 on the peaks. 13.0–15.0s KEEP: hero of the finished cup exactly as in @image1, tall whip, forest bokeh. HOLD. NEGATIVE: empty cup, splash crown around empty cup, milk poured onto dry powder, leaves, white cyclorama, branded carton, overlay caption, storyboard.

Your avatar tours a haunted mansion at night as the house comes alive with dread, ending with a ghost appearing right behind them. Music only, loud. See the video step for the full prompt.