MMA — THE UNDERDOG. @image1 (Ria, the underdog) is dominated by @image2 (Brecka, the favorite) in an octagon cage, then lands a knockout counter and stands over her as the crowd erupts. Plug in ANY two avatars. Music only, loud. See the video step for the full prompt.

3 steps

FIRE — SPARK TO FIRESTORM @image1 snaps a flame onto a fingertip, unleashes a swirling firestorm around them like a fire-bender, then gathers it into a glowing fireball. Plug in ANY avatar. Music only, loud. See the video step for the full prompt.

2 steps

Photoreal live-action commercial, shot on a full-frame cinema camera with fast prime lenses: shallow depth of field, warm low sun raking across the frame, dust drifting in the light, true pore texture and true fabric behaviour, the camera handheld-steady with only small settling movements. Not illustration, not a 3D render, not stylised. @image1 is the reference plate and is an absolute lock for BOTH the person and the product. The face, build, apparent age, hair, garment and jewellery are taken from the large portrait in its left zone exactly, with no drift and no reinterpretation in any shot. The product is the object in its right zone, reproduced exactly — same container, same label artwork and lettering, never redrawn, substituted or mirrored. Only that product exists; no second container appears at any point. @image2 is the storyboard and is followed exactly, panel for panel: six shots in that order, matching its framing, subject scale, pose and composition. One character is on camera throughout. A second person is heard off camera and is never seen, except for one hand. Shot 1 — the character holds the product up beside their face and speaks to the lens in a warm, confident tone: "Everyone asks about my morning routine." A brief pause. The smile drops, the eyes roll, and they shrug through the next line: "Nobody asks, but I already rented the studio." Shot 2 — slow motion, close on the product standing upright. A hand enters from the edge of frame and slides a leafy branch in behind it, clumsily and in plain view. The character speaks off camera: "That's my friend holding a plant. We wanted it to feel natural." Shot 3 — the character holds the product casually at chest height, the face of the label squared to the lens, and says brightly: "It's part of my daily routine," then the tone drops to reluctance as the eyes flick up and away toward the top right, "The kitchen isn't," and the eyes come back to the lens for the flat, deadpan finish, "I can't afford this kitchen." Shot 4 — tight on the face. The character attempts a hair flip into a strong breeze and a leaf blows against their mouth. They pick it away with quiet irritation and ask: "Do we need the wind machine?" The second voice answers off camera: "You paid for it." The character shrugs and answers immediately: "Keep it on." Shot 5 — close on the product standing upright, warm sunlight, soft shadows, clean composition. A hand enters from above and clumsily rotates it until the face of the label squares to the lens. The character speaks in voiceover: "One simple part of my day." Shot 6 — the product alone, centred on a clean empty ground, lit softly, settling and holding with only the faintest drift of light across it. The second voice says off camera: "Studio time's up." The character replies: "Get the plant." End immediately. Understated lifestyle music runs throughout with the dialogue clearly audible above it, and the music stops briefly for the final exchange. No on-screen captions or subtitles are burned into the frame.

3 steps030

CHARACTER LOCK — HIGHEST PRIORITY (applies separately to @image1, @image2, @image3): Each character in every shot is exactly the character in their own reference image. Photocopy facial identity from the matching tag. Preserve facial structure, eye shape and spacing, eyebrows, nose, mouth, exact skin tone, hairline, hairstyle, hair length, texture and colour, freckles and distinctive marks, apparent age, build and body proportions — identically, shot to shot. Preserve each parent’s gender presentation exactly as shown; do not masculinize, feminize or otherwise reinterpret it. Do not beautify, slim, de-age, restyle hair, or morph any face into another. Never swap the three identities. Do not invent extra leads. Do not describe the faces — follow the images only. @image1 is the child. Photocopy @image1 exactly: same face, same hair, same skin, same body and the same clothes as in the photo. This is the child in the photo at their own age, not a recast, not a different kid. Clothing from @image1 stays on them the whole film. @image2 is the first parent (sleeping, then waking). Photocopy @image2 exactly: same face, same hair, same features. Never swap with @image3. @image3 is the second parent (pickup at the door). Photocopy @image3 exactly: same face, same hair, any facial hair, and the same clothes as in the photo. Never swap with @image2. Ignore any checkerboard behind @image3 — that is not part of the set. HANDS: always anatomically correct with five fingers each, naturally proportioned, including the child’s hands, in every shot they appear. Never open on a studio headshot, a garden still, a checkerboard, or the attached photos as frame 0. First frame is already in the bedroom, in-scene. 19 seconds. 9:16 vertical. Heartwarming cinematic 3D animated short film, high-quality family-animation style (premium animated-movie look, not live-action photoreal street). Expressive characters, realistic soft skin and hair on the locked faces, detailed facial expressions, physically accurate lighting, warm golden color palette, soft shadows, realistic cloth and blanket textures, smooth natural body movement, cinematic depth of field, subtle camera motion. AUDIO (native, synced): quiet morning household only — soft breathing, bed sheets, small feet on wood, a gentle door, a warm parent laugh. No BGM, no score, no songs, no on-screen text. SCENE 1 — 0–5s: Warm cozy bedroom. The child (@image1) sits behind the sleeping parent (@image2) on a comfortable bed and gently touches and plays with their hair. @image2 sleeps peacefully, head on a soft pillow. The child smiles mischievously, looking at them with curiosity. Warm golden morning sunlight through sheer curtains. Cozy wooden bed, bedside lamp, soft blankets. Cinematic close-ups, shallow depth of field, soft volumetric lighting. Faces match their tags. SCENE 2 — 5–9s: The child slowly crawls over the bed toward the edge. @image2 begins to wake, looking surprised as they realize the child is moving away, raising their head and reaching toward the child. The child continues crawling with an excited expression. Natural character movement. Faces still match @image1 and @image2. SCENE 3 — 9–13s: The child carefully climbs down from the bed and crawls across the bedroom floor toward the open doorway. Camera follows from behind at the child’s eye level. Warm wooden flooring, soft morning light, cozy home interior, cinematic tracking shot. The child is still @image1 in the same clothes. SCENE 4 — 13–16s: The child crawls through the hallway toward the entrance of the home, camera smoothly following from behind across the polished wooden floor. Tiny hands and knees move naturally. Emotional, playful family atmosphere. SCENE 5 — 16–19s: @image3 appears and picks up the child, smiling warmly and looking at them with affection. The child happily touches @image3's face and gives a gentle kiss on the cheek. Warm indoor lighting, emotional close-up, beautiful family moment. Both faces photocopy their tags. CAMERA: Smooth cinematic camera movement, close-ups, medium shots, low-angle child perspective, gentle tracking shots, shallow depth of field, natural focus transitions. No text, no subtitles, no logos, no burned-in watermark, no BGM, no distorted anatomy, no extra fingers or limbs, no flickering, no sudden character changes, no garden terrace from @image1, no checkerboard from @image3. NEGATIVE: age recast of @image1, swapped parents, third invented adult, studio grey still as frame 0, live-action street photoreal that drops the 3D family-film look, on-screen type.

MELT — LIQUID CHROME Cornered in a rain-slicked neon alley, @image1 melts into mirror-bright liquid chrome, pours away to escape, then re-solidifies into themselves. Plug in ANY avatar. Music only, loud. See the video step for the full prompt.

2 steps

CHARACTER LOCK — HIGHEST PRIORITY (applies separately to @image1 and @image2): Each person in every shot is exactly the person in their own reference image. Photocopy facial identity from the matching tag. Preserve facial structure, eye shape and spacing, eyebrows, nose, mouth, exact skin tone, hairline, hairstyle, hair length, texture and colour, freckles and distinctive marks, apparent age, build and body proportions — identically, shot to shot. Preserve gender presentation exactly as shown. Do not beautify, slim, de-age, restyle hair, or morph the two faces into each other. Do not invent a third lead. Do not describe the faces — follow the images only. @image1 is the first person. @image2 is the second person. Never swap them. WARDROBE (scene clothes, not the stills): @image1 wears an ivory evening dress throughout. @image2 wears an elegant black outfit throughout. Hair stays as in each reference. Clothing and hair stay consistent for the whole 30 seconds. PRODUCT LOCK — @image3 is the jewelry piece. Photocopy it exactly as shown: same metal, stones, design and any box exactly as shown. Do not regenerate, restyle, recolor, or invent a different jewel. Whenever the piece is on screen it must match @image3. BRACELET WEARER: the piece from @image3 is worn only on @image1's wrist. Never on @image2. @image2 may open the box or offer it; @image1 receives it and wears it. His wrists stay bare. HANDS: anatomically correct, five fingers each, naturally proportioned, when they appear. During the dance they are not the subject. Never open on a studio headshot or the attached still as frame 0. First frame is already on the terrace, in-scene. 30 seconds. 9:16 vertical. Photorealistic luxury jewelry commercial. Elegant seaside terrace at blue hour: ocean view, wooden terrace, dinner table, premium dinnerware, flowers, subtle warm decorative candlelight as scene dressing only (not the product). Cool blue-hour ambient mixed with warm golden candlelight. Premium full-frame camera, shallow depth of field, natural bokeh. Realistic skin, ocean, wind, jewelry sparkle, food, glass reflections, fabric movement. AUDIO (native, synced): original elegant instrumental — soft piano and strings at the start, slightly more emotional during the dance, soft premium finish. Subtle realistic ocean waves, breeze, glass clinks, and the soft click of a jewelry box opening. End with a sophisticated voice-over, spoken clearly once: “Make every moment unforgettable.” 0-5s: Close-up of hands opening a small velvet box, revealing the jewelry piece exactly as @image3, glinting in warm candlelight. Reveal the romantic seaside dinner setup around it. 5-12s: @image1 and @image2 enjoy dinner, natural smiles and eye contact. @image2 presents the open box; the piece is fastened onto @image1's wrist. A short macro of the jewelry on her wrist is allowed here only — gemstone facets, metal shine. Faces match their tags. Not on his wrist. 12-20s DANCE — NO HERO HAND: After dinner they slowly dance beside the table. Camera is a medium or full two-shot of the couple: faces, bodies, dress, ocean behind them. Ocean breeze moves hair and dress. The bracelet stays on @image1's wrist as a small detail on her arm, never large, never in the middle of the frame, never a foreground blur occupying center. Do not cut to a wrist. Do not put any hand, palm, or bracelet close-up in the center of attention. His hands stay at her waist or in a normal dance hold, small in frame. 20-25s: Camera leaves the dance and moves to the jewelry piece on the dinner table (in or beside its open box) — the table product, not a wrist. The couple recedes / softly blurs. Do not fill the frame with a hand. 25-30s: Completely separate premium hero shot of the jewelry piece alone on the dinner table (in or beside its open box), ocean softly blurred behind it. Slow camera push-in, realistic light catching every facet. No logo, no brand marking. Voice-over lands here. No on-screen text, no subtitles, no watermark, no logos, no brand name in frame. No plastic skin, distorted hands, changed product design, unrealistic dancing, excessive lens flare, CGI look, or impossible camera moves. NEGATIVE: giant hand in the center during the dance, extreme wrist close-up during the dance, bracelet on @image2, studio grey headshot, swapped identities, extra third lead, different bracelet, navy plaid suit, white crew-neck tee, on-screen type.

FIRE — WALK THROUGH FLAMES @image1 walks calmly toward camera down a dark corridor as a wall of fire erupts and rolls behind them, unburned, ending on a hero close-up. Plug in ANY avatar. Music only, loud. See the video step for the full prompt.

2 steps

CHARACTER LOCK — HIGHEST PRIORITY (applies separately to @image1 and @image2): Each person in every shot is exactly the person in their own reference image. Preserve their facial identity, facial structure, eye shape and spacing, eyebrows, nose, mouth, exact skin tone, hairline, hairstyle, hair length, texture and colour, freckles and distinctive marks, apparent age, build and body proportions — identically, shot to shot. Preserve their gender presentation exactly as shown; do not masculinize, feminize or otherwise reinterpret it. Do not beautify, slim, de-age or restyle them. No face changes, no identity drift between the two of them. ATTACHED IMAGES ARE MANDATORY AND FINAL: The attached images outrank every word written below them: wherever an attached image and this text disagree, follow the image. ASSET MAP: @image1 — the first character. Photocopy face, hair, makeup, jewelry and the outfit they are wearing in @image1. @image2 — the second character. Photocopy face, hair, makeup, jewelry and the outfit they are wearing in @image2. SCENE vs STILL: @image1 and @image2 are the two people only. The colourful backstage room, clothing racks, vanity bottles, oval frame, extra figures, and the plain studio backdrop in the stills are NOT the set. The set is a real independent art gallery opening night. HANDS: hands are always anatomically correct with five fingers each, naturally proportioned, in every shot they appear. OTHER PEOPLE: background gallery guests and a photographer may appear but are never named, aged, or described by ethnicity/gender — only by role and action. They never share a face with @image1 or @image2. 10-second photorealistic live-event footage at an art gallery opening night. @image1 and @image2 walk together through the gallery space toward a wall of large abstract paintings, where a small cluster of guests and a photographer are gathered. Unbranded: no logos, no brand names, no visible labels, monograms, gallery names or readable signage anywhere in frame — walls show only abstract artwork, no text. The atmosphere feels like a real independent gallery opening: soft track lighting on the paintings, ambient chatter, occasional camera flashes, natural crowd movement in the background. 0–3 seconds: @image1 and @image2 walk side by side through the gallery, relaxed and elegant, natural synchronized movement, outfits moving naturally with each step. They occasionally glance at each other and smile. 3–5 seconds: They arrive near the artwork wall, stop, turn toward the photographer, and naturally adjust their poses. @image1 gives a playful, charming smile while @image2 poses confidently beside them. 5–8 seconds: A handheld camcorder approaches from among the guests, slightly shaky and imperfectly framed, like authentic behind-the-scenes footage. @image1 and @image2 notice the camera, look directly into the lens, smile, wave casually, and pose together. 8–10 seconds: The camera moves closer and slowly pans from @image1 to @image2. They lean slightly toward each other and start laughing naturally and playfully, genuine spontaneous expressions. End on both of them smiling toward the camera, abstract paintings visible but unreadable in the background. CAMERA / REALISM: real handheld camcorder, documentary event footage, slightly shaky operator movement, imperfect framing, subtle autofocus hunting, natural exposure shifts, realistic motion blur, slight lens breathing, authentic camera sensor noise, occasional guest flash photography, practical gallery lighting, realistic skin texture, natural hair movement, realistic fabric physics, candid event photography, no overly smooth gimbal movement. FACIAL PERFORMANCE: natural eye movement, subtle smiles, spontaneous laughter, genuine interaction between the two of them, no frozen expressions. Do not generate any duplicated people, extra limbs, distorted hands, artificial beauty filter, plastic skin, exaggerated body movement, cartoon/anime appearance, or slow-motion effect. No subtitles, no BGM, no watermark, no logos, no brand names or readable text anywhere. Overall feeling: an authentic candid recording from a real gallery-opening photographer, elegant but imperfect, spontaneous and believable. AUDIO: native gallery-opening ambience only — soft chatter, distant footsteps, occasional camera shutter/flash. No music, no voiceover, no lyrics.