A 1970s studio take lip-synced to an audio reference
Shot list · MiniMax H3 per the author · Oct 28, 2026
Two references get separate jobs: the image fixes the singer's face, with hair and wardrobe written out, and the audio supplies the vocal he must lip-sync from the first second to the last in every shot where his mouth is visible. Five three-second shots each have one framing and one step of rising intensity, and the period look is defined as a film stock with a separate clean-print rule, so the 16mm feel does not bring scratches or flicker with it. Handheld behaviour is written as specific imperfections (slight wobble, small reframes, focus hunting) rather than as a mood.
Reference-driven charactersSport, dance and rhythmCamera moves and transitions
H3 takes several references at once, and the cases that hold up treat them like a cast list: each asset gets a label and a job. One image locks the face, a turnaround locks the outfit, a third sets the location, and a storyboard is followed panel by panel as a sequence of beats rather than copied as one picture. The other half of the job is saying what a reference must not pass on. A portrait passes on its pose and camera angle unless told otherwise, a studio sheet passes on its grey backdrop, and a reference with two people needs each person named by appearance. One author identified a police officer and a thief only by which side of the reference image they stood on, then placed them on the opposite sides of the frame, and reported that their roles came back swapped. MiniMax's documented full-reference layout (subject definitions, a summary, what is kept, then the shot-by-shot description) appears in several cases and is worth copying.
Write these in
One line per reference: its label, what it controls, what it must not pass on
Characters identified by appearance, never by their position in the reference image
A rule on whether pose, camera angle and background are inherited
Storyboards declared as a sequence of beats, one panel per beat
A sentence saying which parts the text controls: camera, action, sound
Which image, if any, is the exact first frame
For a reference video, what it controls (motion, camera, timing) and what the text replaces (materials, light, sound)
What breaks it
Uploading references without saying what each is for. The model will guess and blend them.
Telling characters apart by left and right in the reference image. Use hair, clothing and build.
Forgetting to ban grid lines, panel borders and duplicate figures that come with boards and sheets.
Restyling a scene with the original footage as the reference video. H3 tends to copy faces and props from it; a blurred greyscale copy keeps the motion without the content.
The template
References:
<Picture 1>: face and hair of [CHARACTER A] only. Do not copy its pose, framing or background.
<Picture 2>: outfit and proportions of [CHARACTER A], from a [turnaround / single view].
<Picture 3>: the location [place], with [landmarks]. No people from this image appear.
[<Picture 4>: storyboard. Follow each panel as one beat, in order; no panel borders or sketch lines in the video.]
The text below controls camera, action and sound. [No image is / <Picture N> is] the exact first frame.
Summary: [one sentence: who does what, where, ending in what state].
Shot 1, 0-[X]s: [CHARACTER A, identified by [hair / coat colour]] [action] in [location]. Camera [move].
Shot 2, [X]-[Y]s: [...].
Keep [face, outfit, landmarks] unchanged throughout. One instance of each character only.
Ours to use freely. Replace everything in brackets.
27 clips that use it
Each plays from the creator's own post on X. Our note sits under it; the prompt belongs to them.
Labelled references · MiniMax H3 per the author · Aug 27, 2026
Three references are defined as three subjects (explorer, orc observer, location), with a rule that the two characters never share a frame. Summary, retention analysis and shot list follow MiniMax's full-reference layout, and suspense comes from the observer appearing only in fragments behind a column.
Reference-driven charactersPerformance and emotionCamera moves and transitions
Read @ou_zhen599's prompt
<Subject 1> is Alin from <Picture 1>, with her locked identity, facial structure, long light-blonde straight hair, white sleeveless top, blue jeans, brown boots, body proportions, and cool restrained temperament. Her emotional progression must remain clear: wonder and quiet delight at first, then uncertainty, then subtle alertness and the feeling of being watched.
<Subject 2> is the female orc observer from <Picture 2>, with her locked green skin, white dreadlocks, pointed ears, bone ornaments, muscular structure, tribal clothing, bone hammer, bare feet, and wild but controlled presence. She is never an attacker here. She remains hidden in shadow and is shown only through partial fragments such as half a face, one eye, a shoulder line, or the edge of the bone hammer. She must never share the same frame with <Subject 1>.
<Subject 3> is the semi-outdoor ancient cave antechamber / temple forehall from <Picture 3>, with a giant natural cave opening above, cool natural light falling from overhead, towering carved stone columns, a right-side temple portico with steps, a broad central stone path, a darker damp edge zone on the left, scattered torches, wet stone ground, slight mist, and deep extending darkness beyond.
summary:
[reference generation] The target video is a 10-second dark-fantasy exploration fragment in which <Subject 1> enters <Subject 3> and is gradually drawn from awe into unease while <Subject 2>, hidden in shadow near the right-side columns and portico, silently observes and judges her. The visual style is photoreal and restrained, with IMAX 70mm film grain, Nolan-inspired cinematic realism, and a natural palette of charcoal black, rich brown, dark ochre, smoke grey, and pale amber. The dramatic core is not confrontation but the tightening relationship of Alin watching the deeper temple space while the orc watches Alin.
retention_analysis:
<Subject 1> (appears in [Shot 1], [Shot 2], [Shot 4]): fully_preserved - Alin's identity, clothing, body proportions, and emotional progression from wonder to alertness are retained.
<Subject 2> (appears in [Shot 3], [Shot 4], [Shot 5]): fully_preserved - the female orc's identity, shadow-hidden position, tribal details, and fragment-only presentation are retained.
<Subject 3> (appears in all shots): fully_preserved - the cave-temple forehall, overhead opening, carved columns, right-side portico, central path, damp stone, torches, mist, and deep spatial extension are retained.
detailed_description:
The target video uses hyper-real dark-fantasy exploration realism with UE5.5-grade environmental fidelity, grounded camera logic, and restrained suspense. The image carries IMAX 70mm film grain and Nolan-style naturalistic contrast, with cool daylight from the cave opening mixing with warm torchlight across charcoal-black shadow, stone brown, dark ochre, smoke grey, and pale amber highlights. The dramatic structure is attraction, uncertainty, and quiet relational tightening. <Subject 1> and <Subject 2> must never appear in the same frame.
[Shot 1] A low establishing shot begins on wet stone, scattered rubble, shallow reflective dampness, and torch bases in <Subject 3>. The camera slowly pushes forward and slightly tilts upward, revealing the giant cave opening above, the scale of the carved stone columns, the right-side temple portico and steps, the broad central stone path, and the darkness extending deeper into the complex. <Subject 1>, Alin, enters from the left-rear side and walks along the central path until she stops in the central-left area of the forehall. She first looks up toward the broken cave ceiling and the monumental architecture, then lets her gaze travel toward the deeper temple interior and the right-side portico. Her expression opens into visible wonder and quiet excitement, as if she has finally discovered a hidden place of real significance.
[Shot 2] At 00:03.000, the camera moves into a medium-close observation shot of Alin, keeping her on the left third of the frame while preserving a large amount of architectural space to the right and behind her. She takes one small half-step forward. Her gaze moves from the right-side portico and column shadows toward the deeper darkness beyond the central path. The initial delight begins to fade. Her breathing grows lighter, her fingers tighten slightly, and her face shifts into uncertainty, as if this is no ordinary ruin but a place that still feels inhabited or aware. Torchlight brushes one side of her face while the other side remains in cool cave shadow, turning curiosity into unease.
[Shot 3] At 00:05.200, the perspective shifts to a hidden peeking view from behind a massive right-side stone column near the portico. The foreground is dominated by the rough pillar edge, blocking most of the frame. In the dark midground, only part of <Subject 2> is visible: perhaps half a face, one eye, the line of a shoulder, and a faint glint on bone ornament or the edge of the bone hammer. In the far background, Alin remains on the central-left stone path, still studying the temple space. The female orc does not move. She simply watches with a calm, assessing gaze, as if she has already been judging the intruder for some time. <Subject 1> and <Subject 2> are never shown in the same frame.
[Shot 4] At 00:07.000, the edit begins a restrained cross-close pattern without ever bringing the two subjects into the same shot. First, Alin in close view: she suddenly feels attention on her from somewhere inside the shadowed architecture and slowly stops. Her head turns toward the right-side column-shadow area, and the wonder has fully become alert uncertainty. Then the cut returns to <Subject 2> in her hidden position. She does not retreat. She only tilts her head slightly and keeps watching, as if deciding whether Alin should be allowed to continue deeper into the place. The tension comes from mutual awareness without formal contact.
[Shot 5] At 00:08.800, the camera settles on a final hook shot focused only on <Subject 2> in the right-side shadow. The foreground right edge is blocked by a rough stone pillar. In the midground, only the female orc's half shoulder line, part of her side face, one eye, and the upper section of the bone hammer are visible. The background falls away into soft torch blur and cave darkness. The camera makes a slight probing move forward and then stops. <Subject 2> keeps watching toward camera direction, as if she has already finished judging the intruder. She tilts her head just slightly, then slowly steps back by half a step while keeping hold of the bone hammer, letting her face recede into shadow again until only the weapon edge and a faint glint in the eye remain. Torchlight briefly skims her eye, the bridge of her nose, and the edge of the hammer, then is swallowed by darkness again. She never fully reveals herself. The ending feeling is clear: she has seen Alin completely, but she has not yet chosen to emerge.
overall_soundscape:
No music or only an almost inaudible low ambient pressure. The soundscape is built from cave reverb, soft torch crackle, distant water drops, small rubble footfalls, faint cloth movement, and low air flowing in from the cave opening. The space must feel quiet enough that every small movement becomes legible.
non_diegetic_music:
N/A
Shot-by-shot timeline · MiniMax H3 per the author · Aug 24, 2026
The author posted this as a failure. The two men are identified only as the left and right figures in the reference image, and the prompt then places them on the opposite sides of the frame; the post reports that officer and thief came back reversed, a clear argument for binding identity by appearance.
Reference-driven charactersDialogue and native sound
sections · MiniMax H3 per the author · Aug 22, 2026
Two prompts use the same board in opposite ways. One animates the board in place with a fully static camera and rules for each panel (portrait poses, turntables, small actions inside frames); the other uses it as reference only and builds a cinematic intro, showing how one sentence changes a reference's role.
Reference-driven charactersAnimated styles and art direction
Read @aimikoda's prompt
Bottom Video
MiniMax H3 Prompt for Board Animation:
Use the provided character board @[board reference] as the sole visual source. Animate the existing static board itself while preserving the exact 16:9 composition, panel positions, character design, illustration style, color palette, and solid off-white background.
Keep the camera completely static for the entire video. No zoom, pan, tilt, reframing, parallax, or camera movement of any kind. The board itself must remain perfectly locked in place.
The large waist-up portrait on the left is the main animated focal point. Let the character naturally change pose within the existing portrait area: subtle head turns, gaze changes, shoulder and upper-body shifts, small hand or arm movements when visible, expression changes, blinking, breathing, and gentle hair and clothing motion. Avoid repeating a single idle loop. Make it feel like a short sequence of natural portrait poses.
The 3 full-body characters in the upper-right must remain in their exact panel positions and each perform a smooth full 360-degree turn in place, like a character turntable. Preserve their original proportions, outfit, identity, and panel boundaries.
The 3 environment panels in the lower-right must remain fixed inside their frames. Animate each scene subtly and differently, preserving its original setting and lighting. Let the character perform small natural actions appropriate to the environment, such as walking a few steps, shifting posture, sitting naturally, leaning, looking around, or reacting gently to the surroundings. Add restrained environmental motion such as rain, wind, moving clouds, reflections, foliage, drifting particles, or subtle lighting changes where appropriate.
Keep the bottom color palette completely static. Keep the solid off-white board background unchanged. Do not rearrange panels, expand scenes beyond their frames, add new elements, duplicate characters, redesign the character, or alter the layout. The result should feel like a polished animated artbook character board.
Top Video
MiniMax H3 Prompt for Character animation:
Use the provided character board @[board reference] as the sole visual reference for character identity, design, facial features, hair, proportions, outfit, colors, environments, and illustration style. Use it as reference only. Do not animate the board itself.
Create a short cinematic character introduction where the character comes to life through a sequence of polished animated shots inspired by the board’s content. Begin with the large waist-up portrait as a living shot, then move into the 3 full-body poses as brief character-focused moments, then transition into the 3 environment scenes as cinematic shots where the character naturally exists in those spaces. End on a clean composed hero shot of the character.
Let the character be the animated focus in every shot. Use natural motion such as blinking, breathing, gaze shifts, head turns, subtle pose changes, small gestures, gentle hair and clothing motion, walking, sitting, leaning, resting, or looking out depending on the scene. The 3 full-body moments should clearly show the character’s silhouette, outfit, and attitude. The 3 environment moments should show the character in different natural settings, different lighting conditions, and different natural poses that fit each location.
Use elegant cinematic camera movement, smooth transitions, and refined artbook-inspired presentation. Preserve strong character consistency and the board’s color language throughout. Minimal typography, atmospheric, polished, artistic, no redesign, no watermark.
storyboard · MiniMax H3 per the author · Aug 22, 2026
The whole video prompt is two instructions: treat the storyboard as a sequence of beats rather than one image, and use the character image for character and style. The post says the storyboard itself came from a one-line request for a rough sketch, so the structure lives in the board, not in the text.
Reference-driven characters
Read @aimikoda's prompt
Use the storyboard @[storyboard ref] as sequential shot guidance, not as a static image. Do not treat the storyboard as one image. Follow each panel as a separate beat. Use @[character ref] as the character and visual style reference.
storyboard · MiniMax H3 per the author · Aug 20, 2026
The storyboard handles the shots and the text handles the speech: four short lines in Japanese for her to say while making the latte. Each line matches a stage of the process, which puts the spoken script in the same order as the panels.
Reference-driven charactersDialogue and native sound
Ballet portrait where the reference sets only the look
sections · MiniMax H3 per the author · Aug 16, 2026
The reference is declared visual only: it sets the person, clothing, location and light, but not the camera, choreography or sound, which the text defines in separate sections. Five named ballet positions each get a framing note, and the camera holds a fixed portrait distance, widening only when the feet need to be seen.
Reference-driven charactersSport, dance and rhythmCamera moves and transitions
Read @LoveUolanda's prompt
REFERENCE
<Picture 1> is the ONLY visual reference image.
Use <Picture 1> to preserve:
the exact person
the person's appearance
the clothing
the hairstyle
the visible makeup
the environment
the background
the lighting
the exposure
the color appearance
the time-of-day appearance.
Do not redesign or reinterpret these visual elements.
The person and environment remain visually consistent with <Picture 1> throughout the video.
IMPORTANT:
<Picture 1> provides VISUAL REFERENCE ONLY.
<Picture 1> does NOT define the camera movement.
<Picture 1> does NOT define the choreography.
<Picture 1> does NOT define the soundtrack.
Camera movement, choreography, and audio are defined independently below.
==================================================
SUBJECT
==================================================
The exact person shown in <Picture 1> is the only dancer.
Keep the same person throughout the entire sequence.
Do not replace the person.
Do not redesign the person.
Do not add new appearance characteristics.
Preserve the person's appearance according to <Picture 1>.
==================================================
ENVIRONMENT
==================================================
Use the environment shown in <Picture 1> as the continuous location.
Keep the same visible background and spatial atmosphere.
Do not replace the environment.
Do not transform the location into another setting.
==================================================
LIGHTING AND COLOR
==================================================
Preserve the lighting and color appearance shown in <Picture 1>.
Keep the same apparent illumination.
Keep the same exposure.
Keep the same white balance.
Keep the same color relationships.
Keep the same time-of-day appearance.
The visual lighting remains consistent with <Picture 1> from beginning to end.
Do not introduce a new lighting design.
Do not introduce a new color grade.
Do not reinterpret the reference image with a different visual atmosphere.
==================================================
PHOTOGRAPHIC STYLE
==================================================
Real human photography.
Professional real-life ballet dancer / coser portrait photography.
Natural photographic rendering.
The realism comes from believable human movement, natural optics, and professional camera operation.
Do not use facial close-up photography.
Do not use macro photography.
Do not make the face fill the frame.
==================================================
CAMERA — TEXT DEFINED
==================================================
The camera movement is independently created by this prompt.
Use a professional portrait camera positioned at a comfortable medium distance.
The dominant composition is a natural half-body portrait.
Frame primarily from approximately chest to waist while keeping the head clearly visible.
The camera remains physically separated from the dancer.
Use restrained, elegant camera movement:
gentle lateral tracking
→ subtle three-quarter movement
→ small smooth arc
→ slight reframing
→ controlled portrait hold.
The camera moves around the dancer rather than directly toward the face.
Maintain a consistent comfortable camera distance.
Do not suddenly approach the face.
Do not push the lens toward the face.
Do not switch to face-only framing.
Do not use extreme close-up framing.
When a ballet movement requires the feet or more of the body to be visible, temporarily widen the composition only enough to clearly show the movement.
After the movement, naturally return to the medium portrait composition.
The camera movement should feel like a real professional photographer physically operating the camera.
==================================================
BALLET ACTION — TEXT DEFINED
==================================================
The choreography is independently defined by this prompt.
Create a refined five-part classical ballet portrait sequence.
The dancer performs five distinct classical ballet poses.
Each pose transitions naturally into the next.
The movement is graceful, controlled, continuous, and anatomically believable.
SHOT 1 — RELEVÉ
The dancer rises naturally into relevé on demi-pointe.
Both heels lift.
The feet and ankles remain anatomically correct.
The dancer maintains controlled balance and elegant ballet posture.
Temporarily allow a slightly wider composition when necessary to clearly show the relevé and feet.
The camera remains smooth and controlled.
SHOT 2 — PORT DE BRAS + CAMBRÉ
Transition to a rear three-quarter composition.
The dancer raises both arms overhead into a classical bras haut / fifth-position en haut line.
She performs a controlled cambré backward.
The spine follows a natural ballet curve.
The head follows the arm line naturally.
The movement remains elegant and physically believable.
The camera makes a subtle three-quarter movement while maintaining a comfortable portrait distance.
SHOT 3 — EFFACÉ + PORT DE BRAS
Transition into a three-quarter side composition.
The dancer performs an elegant effacé standing position.
One hand lightly touches the wall when appropriate to the visible environment.
The opposite arm extends upward and slightly backward.
The supporting leg remains extended.
The dancer maintains stable classical posture.
The camera performs a gentle lateral movement.
SHOT 4 — THREE-QUARTER TURN
The dancer begins partially turned away from the camera.
She smoothly rotates her shoulders and head toward the camera.
The chin leads naturally.
The eyes gradually meet the camera.
The arms settle into bras bas.
The camera subtly follows the body rotation.
Keep the camera at a comfortable portrait distance.
SHOT 5 — FINAL PORTRAIT
Finish with a natural frontal half-body portrait.
The dancer stands calmly with elegant ballet posture.
Her hands rest naturally around waist level.
She looks calmly toward the camera.
The camera settles into a stable portrait hold.
The final expression becomes a very subtle closed-mouth Mona Lisa-like smile.
The smile is restrained, quiet, natural, and almost imperceptible.
No dramatic emotional change.
==================================================
EXPRESSION
==================================================
Keep the expression understated throughout.
The final smile is:
very subtle
closed-mouth
quiet
natural
slightly enigmatic.
No visible teeth.
No broad smile.
No laughing.
No exaggerated facial movement.
No theatrical expression.
No dramatic emotional transition.
==================================================
MOVEMENT REALISM
==================================================
Maintain realistic human biomechanics.
Natural balance.
Natural weight transfer.
Natural joint movement.
Natural ballet foot articulation.
Natural body momentum.
Natural transitions.
No floating.
No distorted anatomy.
No stretched limbs.
No sudden body deformation.
==================================================
AUDIO — TEXT DEFINED
==================================================
The soundtrack is independently created by this prompt.
Use ONLY instrumental background music.
Create a refined, elegant ballet-inspired instrumental composition.
Soft piano with subtle strings.
The music remains understated and supports the portrait sequence.
NO HUMAN VOICE.
NO SINGING.
NO VOCALS.
NO LYRICS.
NO SPEECH.
NO DIALOGUE.
NO CHANTING.
NO HUMMING.
NO VOCAL SAMPLES.
The dancer remains completely silent.
The dancer does not sing.
The dancer does not speak.
The dancer does not lip-sync.
The mouth does not move in response to the music.
Only instrumental background music is present.
==================================================
FINAL STRUCTURE
==================================================
Picture 1 = visual reference only.
Text choreography = ballet movement.
Text camera instructions = camera movement.
Text audio instructions = instrumental background music.
Keep these four functions independent.
The final result is a real human ballet portrait sequence using the exact person and visual world from <Picture 1>, with independently generated classical ballet movement, restrained professional camera movement, natural half-body portrait framing, and instrumental background music only.
Continuous action chain · MiniMax H3 per the author · Jul 31, 2026
Three references cover three performers, and the prompt directs them by relationship rather than timing: the woman never acknowledges the man, the horse treats both as part of its routine, and all three fall into sync for single beats. A closed list of words flashes through as editorial type, and the clip ends on one whispered line.
Reference-driven charactersSport, dance and rhythmKinetic type and title sequences
Read @egeberkina's prompt
Use @ Image1 as the man reference.
Use @ Image2 as the woman reference.
Use @ Image3 as the horse reference.
Create a surreal 15-second fashion-film sequence inside a minimalist monochrome blue studio. The woman performs an elegant contemporary dance. The man casually walks into frame wearing a completely serious expression. Without explanation, he immediately joins the choreography. His movements are awkward, strangely confident, almost ritualistic. The woman never acknowledges him. The horse behaves as though both dancers are part of its routine. It wanders through the studio. Runs behind them. Appears from impossible places. Stops. Looks directly into the camera. Leaves. Returns again. Sometimes all three performers accidentally synchronize for one beat before drifting apart again. The choreography becomes increasingly absurd but remains perfectly cinematic. Every movement is intentional. Every frame feels designed. Throughout the sequence, premium kinetic typography appears as part of the composition, not subtitles, but editorial motion graphics inspired by high-end After Effects title design. Animated words flash for only a few frames: THE GARDENER. THE DRIVER. THE HORSE. THE WOMAN. KNIFE. MOVE. FASTER. AGAIN. THE EXIT. These words slam into frame, wrap around the dancers, disappear behind bodies, stretch with motion blur, rotate in 3D, fragment into pieces, reveal through masks, pulse with the bass, smear across cuts, become oversized, tiny, vertical, upside down, and constantly transform using premium motion-design techniques. Add abstract editorial graphics: animated arrows, circles, blueprint lines, tape labels, interface markers, halftone textures, scribbles, grain, tracking graphics, geometric overlays, chromatic aberration, liquid masks, flash frames, split screens, typography trails, speed ramps and graphic wipes. Music evolves into strange electronic art-pop. Analog synths. Mechanical breathing. Heavy bass. Everything is synchronized to the rhythm. Final close-up. The woman whispers: "Again." The man quietly smiles for the first time. The horse suddenly sprints directly toward the camera. The word AGAIN. fills the screen before an instant cut to black.
A Korean magazine edit in thirteen cuts and seven looks
Shot list · MiniMax H3 per the author · Oct 23, 2026
Each of the thirteen numbered cuts carries a look number, one action, one set element and one exact line of type, so outfit, framing and text are scheduled in the same row. Outfits change on the cut and never by morphing, one model is fixed from the reference, and the last half second is held completely still on the finished cover. A long avoid list closes the prompt, covering face and hair drift, garbled text and extra people.
Kinetic type and title sequencesReference-driven charactersProduct, cars and trailers
Read @Yuki26856052's prompt
MiniMax H3 × Ima StudioTHE RED EDIT — BOLD LOOK. SOFT POWER. STYLE: Korean fashion magazine × K-pop comeback teaser × cinematic motion typography. A luxurious red magazine comes alive through 3D editorial installations, paper transitions, and dynamic layouts. CHARACTER: Use the uploaded reference image. Only ONE adult Korean female model, preserving exactly the same face, features, skin tone, slim proportions, and long dark brown-black hair throughout. Mature, glamorous, confident Korean idol aesthetic. No identity drift. WORLD: One unified red editorial studio with giant cream magazine pages, burgundy torn paper, red silk, geometric panels, chrome lettering, silver reflections, film strips, and contact sheets. Palette: scarlet, burgundy, cream, black, silver. 55% live-action fashion film, 45% dimensional motion graphics. Different typography and camera compositions across cuts. 7 LOOKS: Burgundy tweed jacket, cream top, red mini skirt. Scarlet fitted mini dress, black heeled boots. Cream top, high-waisted red skirt, ruby necklace. Black corset top, burgundy skirt, black boots. Red cropped leather jacket, black top and mini skirt. Silver-red metallic top, fitted black skirt. Scarlet off-shoulder couture gown, structured waist. 13 CUTS: 01 | 0–0.8s: Extreme eye close-up, raises gaze, burgundy paper. Large serif: "THE RED EDIT". 02 | 0.8–1.7s: Look 1, waist-up shoulder turn, cream page. Condensed text: "SCARLET". 03 | 1.7–2.6s: Look 1, beauty close-up, brushes hair, red editorial line. "SEOUL MUSE". 04 | 2.6–3.7s: Look 2, full-body, steps forward, giant scarlet panel. "RED SIGNAL". 05 | 3.7–4.8s: Look 3, folding magazine page reveal, adjusts necklace. Thin serif: "CRIMSON TOUCH". 06 | 4.8–5.9s: Look 4, seated pose, weight shift, black platform, burgundy silk. "BOLD MOOD". 07 | 5.9–7s: Look 5, three-quarter turn, brushes hair, silver reflective set. Chrome: "ICON GIRL". 08 | 7–8.1s: Look 5, chest-up, lowers chin then raises gaze, giant red typography wall. "강렬하게". 09 | 8.1–9.2s: Look 6, knee-up, subtle waist shift, black-and-white contact sheets of herself. "AFTER DARK". 10 | 9.2–10.3s: Look 6, burgundy torn-paper reveal, hand on waist, cream background. "OWN THE LOOK". 11 | 10.3–11.4s: Look 7, steps toward camera, minimal red-white editorial grid. "SEOUL MUSE". 12 | 11.4–12.7s: Look 7, slow knee-up push-in. Magazine pages, film strips, red silk, contact sheets, and chrome frames assemble around her. "THE RED EDIT". 13 | 12.7–15s: FINAL COVER. Graphics snap into an elegant Korean luxury magazine cover. She stands slightly right, turns, raises her chin, and looks into camera. Text: "THE RED EDIT / SEOUL MUSE / BOLD LOOK. SOFT POWER. / VOL. 2026". Final 0.5s completely still. MOTION: Slow, elegant model gestures; fast rhythmic graphics. Hard cuts, page flips, paper folds, typography slams, and contact-sheet flashes. Change outfits instantly between cuts, never morph. AUDIO: 120–124 BPM premium K-pop instrumental, punchy kick, clean bass, hi-hats, paper snaps, metallic clicks. No vocals or dialogue. AVOID: Extra people, face drift, hair changes, outfit morphing, outdoor scenes, dancing, static slideshow, repetitive framing, flat stickers, cheap graphics, cyberpunk, garbled text, logos, watermarks, subtitles. RESULT: One girl, seven looks, thirteen dynamic red luxury magazine spreads, ending in a polished THE RED EDIT cover.
One ribbon that reveals a character from feet to face
Continuous action chain · MiniMax H3 per the author · Sep 25, 2026
One effect gets one job: a luminous ribbon that never disappears is the reveal mechanism, passing up the body region by region in a named order while the camera stays in a single take. The reference sets identity and rendering, and palette, effects and music are left for the model to derive from the character, so the same text can be reused with other characters. Light and sparkle are told to peak at the face before a pull-back to the first full-body view, which gives the one-take a build and a payoff.
Camera moves and transitionsReference-driven characters
Read @aimikoda's prompt
Create a high-impact single-take transformation-style character PV using @[character ref]. Preserve identity, proportions, outfit and original rendering style exactly. Let the character determine the full visual language: palette, lighting, effect design, motion style and music. Treat the sequence as a spectacular stylized reveal of the existing character design, not a redesign or costume change.
The entire PV must be one continuous shot with no cuts. Use one flowing camera move that travels around the character in a stylish, dynamic way: smooth orbiting, rising motion, push-ins, slight reframes, parallax shifts and elegant angle changes, all within a single uninterrupted take. The movement should feel premium, controlled and cinematic, gradually building energy.
Use one continuous luminous ribbon as the main effect. The ribbon should remain alive for the whole sequence, never disappearing, constantly flowing around the character like a single uninterrupted stream of energy. It should travel across and around the body, tracing the silhouette, wrapping the legs, circling the waist, sliding past the torso, orbiting the shoulders and rising toward the face. Its path should feel intentional and graceful, as if it is guiding the viewer through the reveal.
Let the ribbon function as the reveal mechanism. As it passes each body region, it illuminates and reveals details in sequence: feet, legs, hips, waist, torso, shoulders, neck, jawline, eyes, then the full figure. The ribbon should leave behind glowing contour accents, colored reflections, spark trails and brief graphic echoes, creating a chain of revealed detail as it moves upward. Make the effect feel continuous and elegant, not fragmented.
Push the visual richness with more color and more light play. The ribbon should shift through layered colors that suit the character, such as vivid pinks, cyans, golds, violets, reds, teals or other fitting tones. Add dynamic lighting interactions throughout: sweeping highlights, moving reflections, lens flares, glossy gleams, rim-light pulses, flare streaks, refracted sparkles and soft color blooms. Let the light dance across skin, fabric, hair and accessories in a way that constantly refreshes the image.
Keep the character subtly active within the shot. Use confident body language and small elegant movements: a slight step, a turn of the hips, a hand gesture, a shoulder roll, a head turn, a lift of the chin, hair and fabric reacting with graceful follow-through. The ribbon should respond to these motions, tightening, whipping, looping or widening as the character moves. Make the effect feel driven by the character’s presence.
Build steadily toward the face. As the ribbon rises, let the camera move closer and the lighting become more intense. Around the eyes and face, increase sparkle density, glow intensity and color layering. This should feel like the emotional peak of the reveal before the final payoff.
At the end of the shot, pull or arc outward just enough to reveal the first full head-to-toe hero view. The ribbon completes one final elegant loop around the full body, then detonates into a spectacular colorful burst of light: expanding ribbons, flare bursts, sparkling fragments, radiant arcs, glowing rings and a brilliant energy bloom. This final explosion should be the visual climax, but the full figure must remain clearly readable inside it. After the burst, briefly hold the completed hero image.
Sync the entire take to a rising, glittering, high-energy musical build with layered whooshes, shimmering accents, bright flare sounds and one huge final impact. Keep the aesthetic graphic, luminous, colorful and stylized rather than realistic fire, smoke or destruction.
No cuts, no text, no typography, no captions, no logos, no redesign, no outfit replacement, no unrelated props.
Continuous action chain · MiniMax H3 per the author · Sep 17, 2026
The right hand is the camera's controller: the lens matches its direction, timing, acceleration and stopping, and the prompt walks through each gesture and the move it triggers, from an inward palm that draws the lens closer to a raised hand that lifts it overhead. Each move has a stated end, such as finishing the rise exactly as the hand reaches its highest point with no extra ascent afterward, which guards against overshoot. The bans name the usual escapes from a one-take: cuts, hidden edits, lens-covering transitions, teleporting and digital zooms.
Camera moves and transitionsSport, dance and rhythmReference-driven characters
Read @aimikoda's prompt
Use @[character ref] for the character's identity, anatomy, clothing and visual style. An unarmed character stands in an empty charcoal studio under steady, soft continuous light. Fluid dance-like weight shifts and natural clothing movement. No flashes, strobing, props, visible camera equipment.
One uninterrupted take. The right hand leads every camera movement: match its direction, timing, acceleration and stopping. Keep the gesture visible while framing the face. The character watches the lens throughout, adjusting the head and torso continuously as its position changes. No cuts, hidden edits, lens-covering transitions, teleporting, digital zooms.
Begin at eye level in a frontal medium view. The character extends the right hand toward the lens, then draws the open palm inward. The camera follows that invitation, advancing toward the hand at the same height. Stop short of contact while the body shifts back gracefully to receive the approach.
The hand sweeps sharply downward. Immediately follow with a fast camera descent to ankle height, braking above the floor. Tilt upward during the descent to retain the body and hand in view. The character bends the knees and lowers the commanding hand beside them, ready to redirect the lens.
The character straightens and raises the right hand from beside the knees to high above the head. Begin the camera's ascent with this lift. Follow its upward rhythm through a smooth rising arc in the open space in front of the character, progressively pitching downward as the lens passes above eye level. Reach the overhead position just as the hand reaches its highest point: the camera is now ABOVE the raised hand, looking vertically down, with the hand and upturned face visible together. The character lifts the chin continuously to follow the lens. Complete the rise and tilt together, with no extra camera ascent after the hand stops.
The raised hand now draws one clockwise horizontal circle around the space above the crown. Follow that moving hand with a corresponding circular camera path above it, maintaining a fixed vertical clearance and the top-down angle. Keep the frame wide enough for the hand, face and shoulders. The hand circles outside the face's outline, leaving the eyes visible. Small foot pivots turn the character beneath the travelling lens; the face stays tipped upward and the eyes follow the camera. Match the circle's speed and finish to the hand, without continuing the orbit after the gesture ends.
The hand lowers forward. Follow with a continuous descending arc outside the head, gradually tilting back to eye level. The character lowers the chin with the lens. The palm then presses toward camera, driving a rapid retreat into a full-body view as the character follows with one flowing step.
The hand draws inward toward the chest, pulling the camera back toward the character. As the hand rises beside the face, ease into a tight face close-up. Finish with clear eye contact, the hand beside the cheek and subtle breathing.
Sound: clothing movement, soft footfalls, restrained air movement. No impact sound, dialogue, text, overlays, additional characters.
Restyling a grey type animation through a reference video
Shot-by-shot timeline · MiniMax H3 per the author · Sep 14, 2026
The post says the grey source animation was built in Blender with another model, and the H3 prompt makes that video the sole reference for type geometry, motion, camera and editing, leaving the text to handle materials, light and sound. The six words are listed in order with a 2.5-second section each, and every section gets its own palette, material and sound effects tied to motions already in the source. It is a clear split of jobs between a reference video and the written prompt.
Kinetic type and title sequencesReference-driven charactersAnimated styles and art direction
Shot-by-shot timeline · MiniMax H3 per the author · Sep 12, 2026
Every stage has a cause on screen: a cursor drives each action, and the prompt says twice that nothing may appear on its own and that each stage grows out of the one before. Five timed stages (guides, sketch, line art, flat colour, shading) lead to a reveal that must closely match the reference, so the reference image works as the end state rather than only as a style guide. Small undo-and-redo strokes are written in to keep the process looking human.
Transformation and style shiftsReference-driven charactersAnimated styles and art direction
Read @AIwithJessica's prompt
Video Prompt:
Aspect Ratio: 16:9 | Duration: 15 seconds
Title: Digital Character Build — Blue-Haired Cyber Idol Speedpaint
Prompt:
Capture a fast, energetic 15-second screen-recording style timelapse showing an artist building an anime character from scratch inside a modern digital painting program, framed in widescreen 16:9. Keep a visible cursor driving every action throughout — nothing should pop into existence on its own; each stage must visibly grow out of the one before it.
0:00–0:02 — Empty Canvas, First Marks
Open on a bare white workspace inside a sleek illustration app. The cursor darts across the screen laying down rough guide shapes — a basic head outline, shoulder line, and loose gesture for a half-body female portrait pose.
0:02–0:05 — Building the Sketch
Pace picks up as the cursor fleshes out facial structure, layered twin-tail hairstyle with twin topknots, front-swept bangs, and a rough outline of a fitted sleeveless mock-neck top. Small pauses for redo/undo strokes keep it grounded and human-feeling.
0:05–0:08 — Ink Pass / Clean Lines
The messy sketch tightens into crisp linework. Cursor toggles between brush sizes, sharpens eye shape, mouth, dangling earrings, stacked necklaces, and a chunky wrist chain. Guide sketches fade out of view.
0:08–0:11 — Flat Color Blocking
Speedy color-fill segment: pale aqua hair, soft ivory skin, vivid blue irises, deep black glossy garment, warm gold-toned accessories, and pink floral hair clips. Show the cursor swapping palettes, sampling colors, and using mask layers to fill precisely.
0:11–0:13 — Light, Shadow & Texture Pass
Cursor lays in subtle shading beneath the hairline and chin, adds glassy shine to eyes and lips, gives the gold jewelry a metallic gleam, and adds soft fabric creases to the top. Push in for a close-up on the face and accessories mid-detail, then pull back out.
0:13–0:15 — Reveal
Toolbars and sketch guides disappear from view. The finished portrait now closely mirrors the reference art — icy blue twin-tails with rose-covered buns, sharp blue-eyed gaze, glossy black top, gold jewelry, all set against a striking black-and-crimson angular backdrop. One last minor cursor tweak, then the frame holds steady on the completed piece.
Style: Crisp digital-timelapse aesthetic, believable creative-software interface, fluid and legible cursor movement, brisk but trackable pacing, clean panel and layer transitions, no on-screen text, no invented dialogue, no random pop-ups.
Motion: Maintain constant forward momentum for all 15 seconds — no dead air. Condense a long real-world drawing session into a tight, readable sequence: blank page → loose sketch → refined linework → base colors → shading/polish → final piece.
Camera: Treat this as a screen-capture with light digital zoom/pan following the cursor and active work area. Interface stays steady and legible — no cinematic camera cuts breaking the software frame.
Note: The character should never materialize in one instant jump — every phase must show clear cursor-driven progress building on the last.
Shot-by-shot timeline · MiniMax H3 per the author · Aug 29, 2026
Ten poses are numbered and placed in short windows, and the camera rule is to accelerate between poses and brake on every lock. A character reference covers appearance and rendering style while its background is ignored, and the last two seconds are reserved for a full-body hold.
Sport, dance and rhythmReference-driven charactersCamera moves and transitions
Read @aimikoda's prompt
A sensual anime fashion film follows a woman flowing through ten precise poses. @[char ref] defines her complete appearance, outfit, accessories, and anime rendering style; ignore the background and add no new accessories.
The film unfolds in one uninterrupted 15-second shot at 128 BPM inside a minimal white cyclorama with hard strobes, faint haze, glossy floor reflections, a satin sheet, and scattered proof sheets. The camera follows one physically continuous path, accelerating between poses and braking sharply on every lock.
0:00-0:01.5: Pose 1, centered frontal MCU. She faces forward with one hand brushing the hair beside her ear as the camera makes a restrained push-in.
0:01.5-0:03.0: Pose 2, she turns three-quarter and lifts her chin. Pose 3, she lowers her gaze as both hands settle softly near her collar and hair. The camera descends in a fast arc, then pauses on her eyes.
0:03.0-0:04.5: Pose 4, she rotates into a strict left profile. Pose 5, she rolls one shoulder forward while the robe edge catches a sharp side light. The camera whips past her cheek and settles close.
0:04.5-0:06.0: Pose 6, she gathers a fold of the robe at her waist. Pose 7, she releases it and turns her mouth toward the lens in an over-the-shoulder look. The camera dives to torso level and rises into a close facial pass.
0:06.0-0:07.5: Pose 8, she lowers onto the satin sheet with one knee raised. Pose 9, she extends one bare foot toward the lens, dominating the foreground. The camera rockets forward at floor level and briefly hangs on the foreshortened pose.
0:07.5-0:09.0: Pose 10, she rises into a three-quarter stance with one hand at her collarbone and the other touching her hair as the robe skims her thigh. The camera slides rapidly across the waistline, then eases into a portrait hold.
0:09-0:11.0: Without repeating a pose, she folds inward and closes her eyes for one breath, then opens into a long upward stretch. The camera circles in a tight orbit, slow on the locks and fast through the transition.
0:11.0-0:13.0: She twists into a rear three-quarter shoulder silhouette, then turns just enough for her jawline and robe neckline to catch the same flash. The camera skims continuously from shoulder to face with strong parallax.
0:13.0-0:15.0: The camera grazes past her eyes, follows the robe neckline, then arcs outward and lowers as she lands in a dominant full-body pose looking down into the lens. End on a clean wide hold under a final white flash bloom.
Use a smooth 24-85mm zoom feel, hard frontal flashes, sharp side rim light, shutter clicks, breath, hair movement, and fabric rustle. Keep one continuous character with stable identity, anatomy, hairstyle, outfit, accessories, and anime style throughout.
Forcing an overhead phone angle against the reference
Labelled references · MiniMax H3 per the author · Aug 29, 2026
The prompt opens with a camera rule marked as the highest priority, repeats it inside the reference analysis, and lists every forbidden fallback, including the reference's own eye-level angle and any correction of face perspective. The author's post says the model tends to pull the camera back to eye level at high resolution to keep the face flattering, and that the template took nine versions to stabilise.
Camera moves and transitionsReference-driven charactersSport, dance and rhythm
Read @LoveUolanda's prompt
MANDATORY SHOT CONSTRAINT, HIGHEST PRIORITY ABOVE ALL ELSE: FORCE 45–55° HIGH-DOWN OVERHEAD SHOT, 30–60 cm above subject head. ACCEPT NATURAL HIGH-ANGLE FACE PERSPECTIVE DISTORTION. DO NOT CORRECT CAMERA TO EYE-LEVEL TO FIX FACE PROPORTION. DO NOT auto-adjust camera perspective for more flattering face. REJECT flat eye-level frontal camera in high-resolution rendering. THIS IS NOT A REQUEST: the camera must look DOWN at Subject 1 from above her head. subject_definitions Subject 1: The young Asian female coser shown in <Picture 1>. Subject 1's face, facial structure, eye shape, makeup, hairstyle, hair color, body proportions, and full costume (every garment, every accessory, and every jewelry piece) are exactly as shown in <Picture 1>; do not describe any appearance detail in text — treat <Picture 1> as the sole source of truth for Subject 1's identity and look. CRITICAL HARD CONSTRAINT: COMPLETELY DISCARD <Picture 1> original camera angle, original framing, original background, original lighting, original pose. NO inheritance of Picture1 shooting perspective is allowed. If any element is not visible in <Picture 1>, it must not appear in the generated video. summary [keyframe completion, reference generation] retention_analysis Subject 1 (from <Picture 1>): fully_preserved — face, facial structure, eye shape, makeup, hairstyle, hair color, body proportions, and full costume (every garment, every accessory, and every jewelry piece) are exactly as shown in <Picture 1>; do not describe any appearance detail in text. CRITICAL HARD CONSTRAINT: COMPLETELY DISCARD <Picture 1> original camera angle, original framing, original background, original lighting, original pose. NO inheritance of Picture1 shooting perspective is allowed. GLOBAL_CAMERA_LOCK (HIGHEST PRIORITY, APPLIES TO EVERY FRAME 0.00–6.18s): ABSOLUTELY OVERHEAD SHOT for entire video. Camera sits 30–60 cm directly above and in front of Subject 1’s head, pitch angle 45–55 degrees LOOKING DOWN onto Subject 1. 24–28mm wide-angle lens, subtle handheld phone selfie feel. FORBIDDEN: NO eye-level shot, NO camera aligned to subject face height, NO low angle, NO side angle, NO flat frontal shot. Never revert to Picture1’s original eye-level perspective. CRITICAL FOR HIGH-RES RENDERING: Do NOT auto-compensate perspective to flatten facial geometry. Preserve overhead-shot facial perspective distortion even in 4K high-detail output. Never lift subject’s head to fake overhead shot while camera remains eye-level. If the face appears slightly foreshortened or the forehead appears larger due to high-angle perspective, this is correct and must not be fixed. Subject 1’s upward tilted face / lifted chin / upward gaze is a direct result of this high-down camera; DO NOT produce a version where camera stays eye-level while subject tilts her chin upward. Background: high-key seamless all-white photo studio, seamless white floor merging into white backdrop, small white cubic stool on lower-left floor area, no visible walls or corners. Lighting: large overhead softbox shadow-free studio top light + heavy white-surface bounce light, clean soft even slightly-cool daylight, pure white-on-white palette, zero yellow color cast. 0.00 second starting frame IS this overhead studio setup, NOT Picture1’s original eye-level image. SWAY RHYTHM LOCK (HARD CONSTRAINT): All hip-waist sway sections are LARGE-AMPLITUDE but SLOW, SMOOTH, GENTLE, and CONTROLLED. The hips and waist trace a clearly wide S-curve from side to side, but the movement must look slow and relaxed, not fast, jerky, snappy, or violent. The sway should feel like a slow BGM-following body wave, not a quick hip shake. Even with wide hip movement, the speed must stay calm, continuous, and flowing. Do NOT reduce the amplitude to make it look slower; instead, make the movement smoother and more leisurely. detailed_description Base pose lock (applies to the entire video). Subject 1's chin is lifted and her face is tilted up toward the high overhead camera lens. Her eyes are directed straight at the overhead camera lens, and her head does not tilt or drift downward at any point in the video. Subject 1 is not looking down, not looking to the side, not looking off-axis — her gaze is on the overhead camera lens throughout. This is the inheritance override: the base pose from <Picture 1> is NOT carried over; the base pose from the prompt text (chin lifted, eyes on overhead camera lens, head up) is the active pose. Every action segment in the video — entrance, turn, hands-on-hips, hands-behind-back, main sway section, closing sway section, final held pose — must preserve this base pose. The base pose is a hard lock, not a passing phrase; the eyes stay on the overhead camera lens the entire video, the chin stays lifted the entire video, the head does not tilt down the entire video.The target video is a 6-second 9:16 vertical ultra-realistic live-action fashion short. All camera rules inside GLOBAL_CAMERA_LOCK are strictly enforced for every timestamp. From this high overhead position, camera looks DOWN at Subject 1, Subject 1 actively lifts her chin, tilts her face up and looks UP into the overhead lens. The camera begins at 0.00 second in a wide near-full-body shot from this high-overhead position, then performs a fast push-in over the first 0.30 second to a large-half-body composition as Subject 1 walks in and approaches. After 0.30 second, the camera holds the high-overhead position for the rest of the video and follows Subject 1 with very subtle handheld drift and a very gentle orbital motion (handheld phone feel), ALWAYS maintaining 45–55 degree downward pitch, always 30–60cm above subject head. As Subject 1 approaches the camera, the natural perspective of the wide-angle lens produces the typical high-angle fashion selfie look: her eyes appear slightly closer to the camera than the rest of her face (a mild eye-pop), and her body visually extends downward in the frame (torso and legs lengthen). The whole video has 4K ultra-high-definition real-smartphone-photography aesthetic, high dynamic range, mild depth of field, soft highlights, real skin texture, real hair strands, editorial TikTok-fashion-short quality, typical high-angle fashion selfie effect. [Shot 1] begins at 0.00 second with a wide near-full-body shot from the high-overhead camera position defined in GLOBAL_CAMERA_LOCK. Subject 1 is already in frame at the right side, body leaning forward, walking toward the camera in mid-stride — the small white cube stool is visible on the floor to the lower left. The camera holds the high overhead position looking downward and begins a fast push-in with low amplitude as she approaches. Her right leg steps forward, her hair swings gently with the stride, her chin is lifted, her eyes are angled upward toward the overhead camera looking down at her — the looking-up axis is established from the very first frame. By 0.30 second, the framing has shifted from near-full-body to a large-half-body composition, and her face is now clearly tilted up toward the overhead lens with a natural relaxed expression, eyes directed up into the overhead camera. 0.30–0.50 second — the music drop lands. Subject 1 has now reached close-up range. She plants both hands on her hips, elbows pointing outward, in a cute, slightly playful hands-on-hips pose. Her body tilts very slightly to her left, her chin is lifted, her face is angled up toward the overhead camera, her eyes look directly up into the overhead lens, and the corner of her mouth turns up in a soft gentle smile. Her hair falls over one shoulder. The hands-on-hips pose with elbows out is the defining gesture of this segment and gives the shot a confident, slightly cute look-at-me energy. The high-overhead camera continues to look down at her, maintaining the looking-down-at / looking-up-into axis with only very subtle floating / hovering motion. She holds this hands-on-hips, soft-smile, look-up pose with a small natural body sway. The hands-on-hips hit is brief — a quick energy-release beat before the turn. 0.50–1.20 second — Subject 1 begins to turn to her own left. Her shoulders and upper torso lead, the face follows, her hands release from the hips and drop naturally to her sides as the rotation starts. The camera briefly sees her profile and then the back of her head. Her hair swings in a clear but natural arc with the rotation. Even mid-turn, her head and neck stay at a slight upward tilt because the camera is still high above her. The motion is light, smooth, and not a fast spin. By 0.90 second she is fully back-to-camera for roughly half a second, hands at her sides, the back of her hair spread in a long drape, body in a clean vertical line. The white seamless background and the high-angle camera make this back-to-camera moment feel like a confident, slightly cheeky pause. By 1.20 second she has begun rotating back to face the overhead camera, her hair still in the air from the spin. 1.20–1.50 second — Subject 1 turns back to face the overhead camera, the hair fling arcs wide and settles. As she returns to facing the overhead camera, her hands move behind her back into a hands-clasped-behind-back pose, and her chest lifts slightly — chest-out, hands-behind-back, facing the overhead camera. Her face is back in view, eyes lifted to the overhead camera, soft smile appearing. 1.50–3.30 second — Subject 1 performs the main sway section facing the overhead camera, hands clasped behind her back, chest out, soft smile, base pose preserved. The motion is a smooth, slow, BGM-following hip-and-waist sway: the body weight gently drifts from one hip to the other and back, following the BGM beat, with the hips and waist tracing a small S-curve arc on each shift. The sway is 3 to 4 gentle side-to-side shifts, each shift occupying about 0.5 to 0.6 s, so the shifts run across the 1.80 s window. The motion is hip-and-waist-only: the head, eyes, and face do not drift with the body weight shift. The head stays level (no tilt, no dip), the chin stays lifted, the eyes stay on the overhead camera lens, and the base pose is preserved throughout the main sway section — chin lifted, eyes on the overhead camera lens, head not tilted down. The hands stay clasped behind the back, the chest stays lifted. The motion is a smooth, slow, flowing, gentle BGM-following wave through the hips and waist — not a hip-jerk, not a pump-and-reverse, not a hard stop-and-go, not a sharp snap, not a fast jolt, not a forced hard shake, not a small subtle sway. The 3 to 4 shifts run from 1.50 s (first drift left) to 3.30 s (last drift right). The body extends and sways on each shift, then settles back to the vertical line. 3.30–3.50 second — The high-overhead camera continues to look DOWN at Subject 1 from above, holding the 45–55 degree downward pitch. Subject 1's hips settle back to the vertical line as the main sway ends. Her head does not tilt down. Her eyes stay on the overhead camera lens the entire 0.20 s — she does not close her eyes, she does not look down, she does not look to the side. Her soft smile is held throughout. Her hands stay clasped behind her back. The body is in a clean vertical line. 3.50–4.40 second — The high-overhead camera continues to look DOWN at Subject 1 from above, holding the 45–55 degree downward pitch. Subject 1 stands relaxed facing the overhead camera, hands still clasped behind her back, head up, soft smile, eyes stay on the overhead camera lens the entire 0.90 s — she does not close her eyes, she does not look down, she does not look to the side, she does not drift off-axis. The soft smile is held throughout. A long held relaxed breath between the main sway and the closing sway. 4.40–4.60 second — The high-overhead camera continues to look DOWN at Subject 1 from above, holding the 45–55 degree downward pitch. Subject 1 stands in place, eyes stay on the overhead camera lens the entire 0.20 s — she does not close her eyes, she does not look down, she does not look to the side. The head stays level. A small anticipation beat before the closing sway, marked by the filter cutoff in the music narrowing the top end to signal the outro. 4.60–6.10 second — The high-overhead camera continues to look DOWN at Subject 1 from above, holding the 45–55 degree downward pitch, holding the OVERHEAD position throughout this closing sway section. Subject 1 performs the closing sway section from the hands-clasped-behind-back plus chest-out pose, base pose preserved. The motion is a smooth, slow, gentle, BGM-following hip-and-waist sway: the body weight gently drifts from one hip to the other and back, slowly swaying the hips following the BGM beat, with the hips and waist tracing a small S-curve arc on each shift. The sway is 2 to 3 gentle side-to-side shifts, each shift occupying about 0.7 to 0.9 s, so the 2 to 3 shifts run across the 1.50 s window. The motion is hip-and-waist-only: the head, eyes, and face do not drift with the body weight shift. The head stays level (no tilt, no dip), the chin stays lifted, the eyes stay on the overhead camera lens, and the base pose is preserved throughout the closing sway section — chin lifted, eyes on the overhead camera lens, head not tilted down. The hands stay clasped behind the back, the chest stays lifted. The motion is a smooth, slow, flowing, gentle BGM-following wave through the hips and waist — a continuous, soft, undulating drift, slowly swaying the hips from one side to the other, not a series of sharp impacts, not sudden reversals, not a fast hip-shake, not a hard stop-and-go, not a forced hard shake, not a small subtle sway. The closing sway runs from 4.60 s (first drift left) to 6.10 s (last drift right). The closing sway runs through the entire BGM reverb tail outro — the kick and bass drop out at 5.20 s but the swaying continues, matching the lingering energy of the chopped vocal reverb tail. This is the visual climax of the video: 1.50 s of smooth, slow, BGM-following hands-clasped-behind-back plus chest-out hip-and-waist sway facing the overhead camera, fully synchronized with the BGM outro. The body extends and sways on each shift, then returns to the vertical line. 6.10–6.18 second — The high-overhead camera continues to look DOWN at Subject 1 from above, holding the 45–55 degree downward pitch throughout the held final pose. Subject 1's sway settles into a held final pose. Her body returns to the vertical line, hands stay clasped behind her back, chest stays lifted, head returns to center. Her eyes are open, her eyes look directly up into the overhead camera lens, her chin stays lifted, her head does not tilt down, her head does not drift downward, and her soft smile re-engages with the lens. This is a hard lock on the end pose — eyes open and on the overhead camera lens, chin up, head not tilted down — not a passing phrase. The video ends on this held pose at 6.18 second. The whole video emphasizes the interaction between Subject 1 and the high overhead camera. Two hand-position states structure the video: hands on hips at 0.30–0.50 s as the brief opening hit framing; hands clasped behind back at 1.20–6.10 s as the back-to-front return plus the main sway and the closing sway; hands at sides during 0.50–1.20 s turn. The body rhythm has two smooth slow BGM-following hip-and-waist sway sections (the 1.50–3.30 s main sway in the second quarter of the BGM main drop, and the 4.60–6.10 s closing sway covering the entire BGM reverb tail outro) and a brief opening hands-on-hips hit at 0.30 s. The high-angle overhead camera follows the subject with very subtle drift and gentle orbit. As Subject 1 approaches, the wide-angle lens makes her eyes look slightly larger and closer to the overhead lens, and her body visually extends downward in the frame. The closing sway is the visual climax of the video and runs parallel with the BGM outro for full music-to-action synchronization. The base pose is held throughout the entire video — chin lifted, eyes on the overhead camera lens, head not tilted down. overall_soundscape A real indoor photo-studio room tone, soft and clean, with the very faint ambient hum of overhead studio lights. The soundscape is built from small physical diegetic sounds: the soft tap of shoes on hard studio floor as Subject 1 walks in during the opening, the gentle rustle of fabric as her outfit shifts with the turn and as her hands move onto and off her hips, the soft whoosh of hair moving through the air during the 0.50–1.20 s turn, a faint soft click as any ear-worn accessories sway against her neck, and the small rhythmic fabric rustle of the two smooth slow hip-and-waist sway sections (1.50–3.30 s and 4.60–6.10 s). No dialogue, no spoken words, no on-screen text, no subtitles. non_diegetic_music A high-energy Melbourne Bounce / EDM dance track at approximately 128 BPM in 4/4. The track has three clearly separated stages that reflect a typical short-form-video BGM structure designed to loop seamlessly back into the build. (1) 0.00–0.30 second — Pre-Drop Build: no full drum kit plays yet; instead a quick rising drum roll is layered with an upward white-noise whoosh and a high-frequency filtered clap / snare that climbs in volume, pulling the listener's energy upward. A tiny high-frequency inhale vocal cue sits at the very tail of the build. (2) 0.30–5.20 second — Main Drop (Melbourne Bounce): the energy releases hard at 0.30 second into a hard, elastic bouncy sub-kick on every quarter beat, an off-beat rolling bass that lands just after each kick to create the bouncy one-up one-down feel, a sharp clap on beats 2 and 4, a tightly clipped hi-hat, and a gated, pitch-shifted chopped vocal sample (high-frequency, slightly sliding) that acts as the hook. The main sway section at 1.50–3.30 s follows the BGM beat — 3 to 4 smooth slow hip-and-waist shifts, each shift about 0.5 to 0.6 s, follow the rolling bass rhythm, so the 3 to 4 shifts run across this window. From 4.40 to 4.60 second a subtle filter cutoff begins to narrow the top end, and around 4.50 second the chopped-vocal melody nudges downward, preparing the listener for the outro. (3) 5.20–6.18 second — Reverb Tail Outro: the kick and bass drop out suddenly at 5.20 second, only the last chopped vocal and its long reverb tail remain, and a low-pass filter makes the sound feel like it is fading through a wall. The closing sway section at 4.60–6.10 s runs through this entire outro — 2 to 3 smooth slow hip-and-waist shifts, each shift about 0.7 to 0.9 s, continue even after the kick and bass cut out, matching the lingering energy of the chopped vocal reverb tail. The track is short, punchy, hype, and designed to loop back to the start of the build for seamless repeat. Music to action beat map (downbeat-to-pose and stage-boundary synchronization):
0.00–0.30 s: Pre-Drop Build, Subject 1 walks in toward the overhead camera, no full beat yet
0.30 s: DROP, hands-on-hips lands, soft smile, look-up toward overhead lens, kick plus bass plus clap plus vocal-chop all enter
0.30–0.50 s: hands-on-hips held, soft-smile, look-up toward overhead lens (brief opening hit)
0.50–1.20 s: turn begins, hands release from hips, hands drop to sides, back-to-camera moment, half-turn back
1.20–1.50 s: facing front toward overhead camera, hands move behind back, chest out
1.50–3.30 s: main sway section — hands clasped behind back, chest out, smooth slow BGM-following hip-and-waist sway, 3 to 4 gentle side-to-side shifts each about 0.5 to 0.6 s, base pose preserved (chin lifted, eyes on overhead camera lens, head not tilted down)
3.30–3.50 s: hips settle back to vertical line, eyes stay on overhead camera lens, no eyes-closed, no head-down, soft smile held
3.50–4.40 s: relaxed facing-overhead-camera hold, hands stay clasped behind back, eyes stay on overhead camera lens, no eyes-closed
4.40–4.60 s: small anticipation beat, eyes stay on overhead camera lens, no eyes-closed, filter cutoff narrows
4.60 s: closing sway section begins, hands clasped behind back, chest out, smooth slow BGM-following hip-and-waist sway, 2 to 3 gentle side-to-side shifts each about 0.7 to 0.9 s, base pose preserved, continuing through the reverb tail outro
5.20 s: kick and bass cut out, only vocal-chop reverb tail remains, closing sway continues
6.10 s: closing sway settles, body returns to vertical line, held final pose (eyes open, eyes on overhead camera lens, chin up, head not tilted down)
6.18 s: final held pose, end
h3-native · MiniMax H3 per the author · Aug 29, 2026
Written in H3's native dialogue form: each speaker has a label and a delivery note, every line is quoted in English inside dialogue tags, and four shots cut at fixed times from a medium close-up to a held two-shot. Small hand actions (adjusting a strap, holding a forearm) carry the relationship between lines, and soundscape and score are listed separately.
Dialogue and native soundPerformance and emotionReference-driven characters
Read @ou_zhen599's prompt
For the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.
integrated_multimodal_description: [Shot 1] Live-action, cinematic realism. Both women fully match <Picture 1>: same faces, hair, age, body proportions, dresses, palace interior, window light, bench, noble atmosphere, and class contrast. Left woman is the servant: dark-haired, standing, gentle, careful, attentive. Right woman is Princess Emma: blonde, seated, noble, fragile, deeply worried. IMAX 70mm film grain, Nolan-inspired cinematic realism, photorealistic natural palette with charcoal black, umber, dark ochre, ash gray, and pale amber interwoven. Medium close-up at elegant eye level, soft daylight from the right windows, subtle handheld breathing, slow physical push-in. The servant adjusts Emma’s shoulder strap and smooths the fabric at her upper arm while Emma looks down and lightly tightens the fabric in her lap. The servant (S1), soft, warm, respectful, says: <d>[English] Princess Emma... the king is on his way home.</d> [Shot 2] At 00:04.000, the camera cuts tighter and drifts from the servant’s hand toward Emma’s face. The servant keeps one hand lightly on Emma’s shoulder. Emma finally looks up, but the worry does not soften. The servant (S1) says: <d>[English] He will be here soon. Please don’t fear so much.</d> Princess Emma (S2), refined, quiet, emotionally strained, replies: <d>[English] I am trying.</d> [Shot 3] At 00:08.000, the camera cuts to a gentle moving two-shot. Emma rises from the bench and steps toward the window on the right. The servant follows immediately from the left and lightly holds Emma’s forearm in support. The camera softly follows, keeping both women clear and Emma emotionally central. The servant (S1) says: <d>[English] You must trust he will return safely.</d> Princess Emma (S2) replies with slightly rising urgency: <d>[English] I want to... but something feels wrong.</d> [Shot 4] At 00:12.000, the camera cuts to a medium shot with a slight pullback and lift, then holds. Emma stands near the window on the right facing outward. The servant stands just left and slightly behind, smoothing the back of Emma’s sleeve and leaving her hand there in quiet support. The servant (S1) says softly: <d>[English] Emma...</d> Princess Emma (S2) answers: <d>[English] I’m afraid.</d> Final feeling is shared waiting, not relief. Only these two women are visible. The servant remains the comforter. Princess Emma remains the emotional center.
overall_soundscape: Quiet palace room tone, faint fabric rustle, light footsteps on stone floor, soft air near the window, and a distant chandelier creak.
non_diegetic_music: Quiet piano and soft low strings at a very slow tempo, restrained beneath the room tone, with only a slight rise in the final seconds.
h3-native · MiniMax H3 per the author · Aug 27, 2026
A prop rule is stated and then repeated at every point where it could fail: the small jug never touches or rests on the large barrel. The shots move from pouring to a walk toward the lens to an offered glass, and the background is kept blurred so the server remains the only readable person.
Dialogue and native soundPerformance and emotionReference-driven characters
Read @ou_zhen599's prompt
For the target video, at 0.00 seconds into the target video, <Picture 1> (from [Shot 1]) is fully referenced.
integrated_multimodal_description: [Shot 1] Live-action, cinematic realism, Jones fully matches <Picture 1>: same face, hair, age, body proportions, tavern maid outfit, sensual but grounded tavern presence, medieval tavern interior, large serving barrel, small pouring jug in her right hand, beer glass, warm firelight, candles, smoky air, and rustic wooden atmosphere. IMAX 70mm film grain, Nolan-inspired cinematic realism, photorealistic natural palette with charcoal black, umber, dark ochre, ash gray, and pale amber interwoven. Medium close-up from a slightly low front-left angle near the barrel, Jones slightly right of center, large barrel and flowing beer foreground left. Subtle handheld breathing and a gentle physical push-in. Background stays heavily blurred and non-readable so Jones is the only clear visible person. Jones pours beer carefully with believable foam control. The small jug stays in her right hand and never touches, rests on, leans against, or gets placed on the large barrel. The nearby off-screen man with a mature low sexy voice (S2) says in an off-screen voiceover: <d>[English] So this is the place you were saving for me?</d> Jones (S1), mature, warm, sensual, relaxed, confident, playful, answers with a faint smile while still pouring: <d>[English] Mm. Best drink in the house.</d> [Shot 2] At 00:03.500, the camera cuts to a moving medium close-up follow shot. Jones finishes pouring, does not place the jug on the barrel, and either sets it only on the front wooden bar counter or another separate wooden surface in front of her, or keeps it with her briefly away from the barrel. She lifts the filled glass and walks toward the unseen man. The camera tracks backward and slightly sideways to stay physically close, with no zoom. Mid-walk, she briefly tucks loose hair behind her ear, then continues. S2 says in an off-screen voiceover: <d>[English] That confident, huh?</d> Jones replies with a knowing half-smile: <d>[English] You’ll see.</d> [Shot 3] At 00:07.500, the camera cuts to an intimate close-up as Jones approaches, slightly left of center, the glass rising into the lower foreground. The camera retreats slightly, then softly stops. Jones leans in and offers the drink into his space without fully releasing it at first. She says in a lower intimate tone: <d>[English] Drink this first... then maybe I’ll show you the better part of the night.</d> The off-screen man (S2), closer now, says in an off-screen voiceover: <d>[English] Jones... you always talk like that to strangers?</d> Jones gives a tiny breath of a laugh. [Shot 4] At 00:11.500, the camera cuts to a close-up resolving into a medium close-up with a subtle pan and shift as Jones passes slightly across frame and settles at the bar edge or a nearby wooden post. The glass is now in his space or softly set onto the wooden bar. If the small jug is placed down later, it goes only onto the front bar counter or another separate wooden surface in front of her, never onto the large barrel. She shifts her weight, turns back toward the unseen man, and says quietly: <d>[English] Stay for one more.</d> Final beat holds on warm tavern ambience, soft glass touch on wood, low fire crackle, and her off-camera gaze toward him. Only Jones is visible as a clear readable person for the full video. No fantasy effects, no modern props, no exaggerated flirting, no overacting, no music-video energy.
overall_soundscape: Warm tavern ambience with low room chatter, distant laughter, fireplace crackle, beer pouring, soft glass-on-wood contact, fabric movement, and light footsteps on wooden floor. Everything stays intimate, grounded, and natural.
non_diegetic_music: Quiet low strings and subtle plucked acoustic textures at a slow tempo, kept restrained beneath the room tone.
Shot list · MiniMax H3 per the author · Aug 25, 2026
The reference is limited to facial identity while hair and styling are rewritten, and the cursor gets its own overlay rule: flat, locked to screen space, never scaled by the zoom. Each of the three shots carries its own audio line, and every gag is written with weight and recovery (the skin stretches, snaps back, she settles), so the effect has physical consequences.
Product, cars and trailersReference-driven charactersPerformance and emotion
Read @AIwithJessica's prompt
Video Prompt:
@Image 2 is a reference of one single person. Use only her facial features and identity — eyes, nose, lips, bone structure — but restyle her hair into a short bob cut. Styling: small gold hoop earrings, chunky yellow sunglasses (same chunky frame shape as the reference pink pair, recolored yellow), all-yellow outfit — yellow wool overcoat, yellow knit tee, wide pleated yellow trousers — deep blue sky, hard direct sun, low angle, high-contrast saturated photographic grade. Do not reproduce any grid or more than one instance of her. One person only, styling unchanged, uninjured, clothing intact; sky, sun direction and grade constant. Cursor: standard flat white arrow with hard black outline, overlaid on the footage — no perspective, lighting, shadow, depth of field or grain. Locked to screen space, always in front of her, direct snappy motion, click-down state on grab.
Shot 1 — Cheek tug
Live-action. Tight low-angle close-up of her face and shoulder against the sky, hard sun raking her cheek, looking past camera, short bob catching the wind. Slow imperceptible push in. The cursor enters from the right, snaps to her cheek and pinches, dragging it outward and down. Real physical weight: skin stretching taut, mouth deforming open on that side, head tilting with the pull, sunglasses shifting on her nose. It holds, releases, and the cheek snaps back with a soft wobble as she settles, blinking, jaw resetting. She glances toward the cursor. No marks left.
Audio: light wind and open air, a soft rubbery stretch and release, a small mechanical click. No dialogue, music or subtitles.
Shot 2 — Lifted by the collar
Live-action, locked-off. Low-angle wide of her standing full-length on the edge of a plain white ledge against blue sky, hands in pockets, coat open, bob swaying, looking off to one side. The cursor drops in from above, clamps onto the back of her collar, and hoists her straight up — body heavy and passive, shoulders riding up around the grip, arms loose, feet dangling, a slow pendulum swing as her weight settles. She looks down at her own feet. The cursor then drags her sideways and up out of the top corner, her swinging body trailing behind. Ends on the empty ledge and sky.
Audio: wind and open air, fabric strain and rustle, a mechanical click on the grab. No dialogue, music or subtitles.
Shot 3 — Dropped, crash zoom
Live-action, locked-off. Opens empty — white ledge along the bottom, blue sky above. She's dragged in from the left hanging by her collar from the cursor, swinging with real weight, coat flapping, legs skimming the ledge. The cursor hauls her to centre, stops, releases with a click. She drops and lands on her feet in a heavy crouch, knees absorbing it, coat billowing and settling. She straightens, brushes her coat down, adjusts her sunglasses with one finger, and looks into the lens. On that look the camera crash zooms hard into a tight close-up, landing with a small jolt and settle, holding on her expression — unimpressed, jaw set, sunglasses catching the sun. The cursor leaves frame after the release and does not scale with the zoom.
Audio: wind and open air, fabric rustle, a mechanical click on the release, a heavy thud on landing, a low whoosh on the zoom. No dialogue, music or subtitles.
Labelled references · MiniMax H3 per the author · Aug 25, 2026
The frame is permanently split into three panels, and the rule that only one copy of the subject may be visible at any moment is restated at every entry. Each appearance has an exact beat time, a named emotional state and a clean exit before the next panel fills, which turns a stacked layout into a timed sequence.
Sport, dance and rhythmReference-driven characters
Read @LoveUolanda's prompt
subject_definitions:
<Subject 1> is the adult woman in <Picture 1>; her identity, facial structure, hairstyle,
body proportions, outfit, the surrounding room, light direction, color palette, and
photographic texture all come from <Picture 1>.
summary:
[reference generation] Create an original 15-second 9:16 vertical triptych relay radiating
bright genki Japanese idol energy: identical half-body copies of <Subject 1> occupy three
fixed horizontal panels, popping down and up strictly one at a time in the fixed order
TOP, BOTTOM, MIDDLE, TOP, each entrance performing a different bouncy cute state on the
beat, ending with all three copies leaning in toward the lens together while each flashes
a clearly different cute hand gesture.
retention_analysis:
<Subject 1>: fully_preserved — preserve facial structure, hairstyle, body proportions,
outfit, environment, light direction, and photographic texture from <Picture 1>, while
allowing new poses, gestures, and performance inside the panels.
detailed_description:
[Shot 1] Locked-off vertical 9:16 frame permanently divided by two thin fixed horizontal
seams into three equal stacked panels; the image never collapses back into a normal
fullscreen composition. Until 00:12.000, strictly one copy of <Subject 1> is visible at
any moment, and no part of her may ever be visible in two panels at the same time; every
panel not currently hosting her stays completely empty and motionless. Overall performance
style is high-spirited genki idol pop: buoyant, springy, cheeks lifted, eyes sparkling,
every motion snapping onto the upbeat 145 BPM grid with a small perky recoil after each
accent. The top panel holds <Subject 1> in half-body framing — head, shoulders, and upper
chest only — centered against the photograph's own backdrop, which continues seamlessly
across the empty middle and bottom panels under the same light. Curious discovery: her
eyes find the lens with a quick delighted widen, brows pop up, her head tilts with a
springy bob, she gives two fast happy shoulder bounces, and her lips press into a bright
closed smile while the fingertips of one hand drum twice on the panel's lower edge. She
finishes this gesture and holds her pose; nothing else moves, and neither of the two
lower panels contains anything at any point in this segment. No legs or feet ever appear;
nobody crosses a seam; nothing fades or teleports; the camera never moves. [Shot 2] At
00:02.400, the top-panel copy is still the only presence in the entire frame. She drops
in one brisk crouch straight down, without pausing, and slides fully out through the
bottom edge of the top panel; by the end of this action every trace of her is gone. From
this moment until 00:03.000 all three panels stand completely empty and motionless, and
no new figure rises anywhere in the frame during this span. [Shot 3] At 00:03.000,
exactly on a clear musical beat, the top and middle panels are still completely empty and
motionless; only then does <Subject 1> pop upward from below the bottom panel's bottom
edge into the same half-body framing with a lively overshoot and settle. Shy peek, played
quick and tickled rather than timid: her eyes flick to the lens, dart away, sneak back,
she claps both hands softly over a growing smile, her shoulders draw inward with one
excited wiggle, then she drops halfway so only her sparkling eyes and hairline peek above
the panel's bottom edge, blinks once playfully, and drops fully out, completely gone
before 00:05.800; hold all three panels empty and still until 00:06.000. [Shot 4] At
00:06.000, exactly on the next strong beat, the top and bottom panels remain empty and
motionless; only then does <Subject 1> pop upward from below the middle panel's bottom
edge into the same half-body framing with the same springy overshoot. Feigned innocence,
played bouncy and comic: caught mid-turn, her eyes widen for one beat, she spins the rest
of the way with a light hop, folds both hands neatly behind her back, rocks once on her
heels, gazes up and away whistling-innocent, then rolls one quick mischievous side-eye
back to the lens with a lopsided grin. Show the complete side-eye before any exit begins;
she sinks down in one brisk move and is fully gone before 00:08.800; hold all three
panels empty and still until 00:09.000. [Shot 5] At 00:09.000, the middle and bottom
panels remain empty and motionless; only then does <Subject 1> pop upward from below the
top panel's bottom edge into the same half-body framing. Secret invitation at full
energy: she rises with her chin dipped and eyes already crinkling, raises one hand low
near the panel's bottom edge and gives three quick come-here finger crooks landed exactly
on three successive beats, each with a snappy elbow bounce, then flashes a proud little
nod; she holds the pose without leaving, and the other two panels stay empty and
motionless through 00:12.000. [Shot 6] At 00:12.000, only after the top copy settles do
the middle and bottom panels each receive another copy of <Subject 1> — the same face,
hairstyle, and outfit as every other copy — both rising from below their own bottom edges
into matching half-body framings with bright overshoot; no copy crosses a seam at any
time. From the instant they appear, the three copies must strike visibly different poses
and never mirror one another. On the strong accent at about 00:13.000 all three lean
inward toward the lens together — shoulders lifting, faces coming closer, eyes beaming —
and then their hands split into three unmistakably different gestures landing on three
rapid successive accents within half a second. The sweet top copy alone flashes double V
signs beside her eye corners with a huge bright smile and her head tipped slightly left.
The cheeky middle copy alone holds up one OK sign beside her right cheek, winks her left
eye, and grins with her head straight. The playful bottom copy alone pops a small finger
heart under her chin with one last springy shoulder bounce and her head tilted right. At
no moment do any two panels share the same hand position, head tilt, or expression; read
together the trio is sweet above, winking in the middle, heart-throwing below. All three
keep leaning slightly toward the lens, beaming and holding their own gestures with tiny
rhythmic bounces through the final frame; seams, framing, light, and outfit remain
identical across panels, hands stay close to the faces without covering them, and
nothing crosses a seam.
overall_soundscape:
Soft room ambience matched to the photograph's environment kept low, one light fabric
sweep synchronized exactly to each downward exit, a short silent beat with no movement
after every exit, one soft rising whoosh synchronized to each entrance, a gentle triple
tap synchronized to the three finger crooks at 00:09.000, and clean dynamics with no
dialogue.
non_diegetic_music:
Cute Japanese summer idol pop, kawaii J-pop, 145 BPM, bright major key, sparkling synth
plucks, bubbly electric piano, crisp claps, punchy kick, playful bass, cheerful genki
anime-idol energy, extremely catchy and rhythmic, strictly instrumental with relentless
forward momentum and no breath gaps. One clear drum hit lands exactly on each entrance
at 00:03, 00:06 and 00:09, a stronger accent hits at 00:12 when the second and third
copies rise, three rapid successive stabs punctuate the three different gestures at
about 00:13.000, and one final clean hit closes the last frame.
Fashion clip where two text strings follow her moves
Shot-by-shot timeline · MiniMax H3 per the author · Aug 25, 2026
Two editable strings at the top are the only text allowed, and every type movement is tied to a body move: a sideways swing throws the text, a stop freezes it. Bullet-time windows freeze both subject and text while only the camera moves, and the music drops away before each freeze.
Kinetic type and title sequencesSport, dance and rhythmReference-driven characters
Ink-smoke characters with type grown from the smoke
JSON structure · MiniMax H3 per the author · Aug 23, 2026
Five reference images take one three-second beat each, and the side the smoke enters from alternates every beat. The word list is closed and the letters must form from smoke with volume and occlusion, never as a flat overlay; the camera is locked or nearly so, leaving all motion to the smoke.
Transformation and style shiftsKinetic type and title sequencesReference-driven characters
Read @techhalla's prompt
{
"archetype": "Artistic / Fashion / Conceptual",
"duration": "15s",
"prompt": {
"concept": {
"title": "Ink Smoke Art – Kinetic Title Sequence",
"description": "Pure continuous 15-second one-take motion-graphics sequence in living black ink-smoke art. Five iconic characters form from swirling black smoke, stay in constant organic motion, and fluidly morph into the next every 3 seconds on a pure white void. Formation direction alternates left-right for artistic rhythm. Kinetic uppercase typography emerges from and interacts with the smoke itself on the complementary side, never flat overlay. Everything remains ultra-fluid, never static. Final short invitation text appears at the end.",
"duration": "15s",
"rhythm_structure": {
"0-3s": "[image1] forms right-to-left, text left",
"3-6s": "Morph to [image2] left-to-right, text right",
"6-9s": "Morph to [image3] right-to-left, text left",
"9-12s": "Morph to [image4] left-to-right, text right",
"12-15s": "Morph to [image5] right-to-left + final invitation text"
}
},
"camera_direction": {
"shot_type": "Strict continuous one-take, locked or micro-drift",
"forbidden": ["No cuts", "No teleportation", "No hidden transitions", "No freeze frames"],
"camera_journey": "Centered medium-wide framing on pure white void. Extremely slow subtle drift or completely locked. All energy comes from the living smoke and character motion; camera never cuts or reframes aggressively."
},
"typography": {
"allowed_words": [
"INK SMOKE ART",
"MOTION GRAPHICS",
"MIDJOURNEY + H3",
"PROMPTS by TechHalla",
"MAKE IT YOURS"
],
"treatment": {
"not_flat_overlay": true,
"required_properties": [
"Formed from dense black smoke particles and tendrils",
"Physical volumetric depth and thickness",
"Smoke occlusion and interaction with letters",
"Subtle continuous warping and particle shedding",
"Bold uppercase, high-impact condensed sans-serif weight",
"Kinetic emergence and dissolution synchronized to morphs and directional flow"
]
}
},
"visual_style": {
"overall": "Black ink smoke / smoke-ink art, ultra-high-contrast monochrome, pure white void background, volumetric living smoke with dense solid cores and wispy ethereal filaments",
"inspired_by": [
"[image1] – exact silhouette, long flowing hair, armor texture and smoke density",
"[image2] – exact suit silhouette, gun, coat edges and frontal presence",
"[image3] – exact spiky hair, martial pose, wristbands and full-body smoke mass",
"[image4] – exact cowl, cape flow and dramatic profile silhouette",
"[image5] – exact straw hat, back view, coat tails and dynamic stance"
],
"color_palette": ["pure black", "deep charcoal smoke gradients", "pure white void"],
"materials_and_texture": "Living ink smoke with variable density, continuous filament motion, soft volumetric edges that never fully solidify, constant micro-turbulence even when character is fully formed"
},
"motion_language": {
"follows_music": false,
"core_behavior": "Never static. Smoke is always in continuous organic swirl. Characters form while already animating (hair, cape, coat, pose micro-movements). Morphs happen only through smoke densifying, dissolving and reforming – no hard cuts, no dissolves, no freezes. Formation direction alternates every beat for balanced artistic rhythm. Every frame contains visible smoke motion."
},
"storyboard": {
"total_beats": 5,
"beat_duration": "3 seconds each",
"beats": [
{
"id": "01",
"time": "0.0–3.0",
"description": "Pure white void. Dense black smoke enters from the right edge and flows leftward, coalescing into the character from [image1] in profile. Long flowing hair and cloak already moving with dense smoke trails trailing right-to-left. Subtle head turn and breathing motion. Kinetic text “INK SMOKE ART” forms on the left side from the arriving smoke tendrils into bold uppercase letters that interact with and are partially occluded by the smoke. Everything remains in fluid motion."
},
{
"id": "02",
"time": "3.0–6.0",
"description": "The character from [image1] dissolves into swirling black smoke that reverses direction, now entering from the left edge and flowing rightward, reforming into the character from [image2] in frontal view. Suit, tie and gun materialize while coat edges continue smoking. Slow forward lean and head tilt. Previous text dissolves into smoke; new kinetic text “MOTION GRAPHICS” emerges on the right side from the arriving smoke mass into solid bold uppercase, still interacting with living tendrils."
},
{
"id": "03",
"time": "6.0–9.0",
"description": "The character from [image2] dissolves into dense black smoke that reverses again, entering from the right and flowing leftward, reforming into the character from [image3] in full-body fighting stance. Spiky hair, wristbands and muscular form solidify while smoke keeps swirling around limbs. Subtle power-stance micro-movement. Text “MIDJOURNEY + H3” forms on the left side kinetically from rising smoke particles into bold uppercase, partially wrapped by residual smoke filaments."
},
{
"id": "04",
"time": "9.0–12.0",
"description": "The character from [image3] dissolves into smoke that reverses, entering from the left and flowing rightward, reforming into the character from [image4] in dramatic profile. Cowl and long cape already billowing with heavy smoke trails. Subtle cape and shoulder motion continues. Text “PROMPTS by TechHalla” emerges on the right side from the trailing cape smoke into bold uppercase kinetic letters that flow and interact with the moving fabric-smoke."
},
{
"id": "05",
"time": "12.0–15.0",
"description": "The character from [image4] dissolves into final swirling black smoke that reverses once more, entering from the right and flowing leftward, reforming into the character from [image5] in rear three-quarter view. Straw hat, coat tails and dynamic stance solidify while smoke continues to rise and trail. All previous text has dissolved; residual smoke particles briefly reform the full set of titles in layered kinetic composition on both sides, then the short invitation text “MAKE IT YOURS” emerges centered from the final smoke mass into bold uppercase as the smoke begins a slow elegant dissipation into the white void, never fully static until end frame."
}
]
},
]
}
}
Shot-by-shot timeline · MiniMax H3, inferred · Aug 22, 2026
Two images are fixed as the exact opening frame and the exact destination, and the man's pointing finger serves as the direction for the whole move. Second ranges stage a drift past the foreground, an alignment with the gesture and the flight into the screen, with bans on cuts, glitches and teleporting.
Camera moves and transitionsChases, racing and fast movementReference-driven characters
Read @CharaspowerAI's prompt
Here’s the prompt it created for this FPV sequence 👇
Use Image 1 as the exact opening frame and Image 2 as the exact destination world, creating one uninterrupted 15-second ultra-smooth cinematic FPV one-take with no cuts, no glitches, no teleportation, no jumpy motion, and no visible camera, as if an invisible stabilized micro-FPV lens is drifting naturally through a warm, lived-in living room and then physically flying into a television to emerge inside the second scene. Begin by preserving Image 1 perfectly: the man stays in the same yellow knit shirt, seated deep in the leather recliner with his boots propped up, cigarette smoke rising from his right hand, beer can still present, his left index finger urgently pointing toward the television in the direction he is already gesturing, with the same casual-but-suddenly-focused expression, the same residential living room geometry, the same pleated curtains, table lamp glow on the right, worn boot soles in the foreground, leather reflections, knit texture, and nostalgic warm low-key lighting. For the first 0–3 seconds, start almost motionless and then begin a very subtle forward FPV drift with a slight sideways arc that feels physically weighted and organic, skimming past the foreground boots and the armrest edge while the camera naturally aligns itself with the line of his pointing finger; use the gesture as the guiding vector for the whole move, letting the viewer feel the television is the true target of the composition. Between 3–7 seconds, perform a smooth curved tracking arc around the man’s pointing side, maintaining his face, hand, smoke, chair, and curtains in consistent perspective while progressively revealing the television in the room; the motion should feel like a real flying camera easing around a seated subject, not a pan or zoom, with strong parallax between the foreground boots, the midground man, and the soft curtain plane behind him. As the television becomes visible, its screen should already contain Image 2 as a living cinematic scene, not a static frame: the same blonde man in a black tuxedo jacket, white shirt, and black bowtie, with slicked-back hair, a knowing confident smile, and a champagne coupe raised toward camera, but animated with subtle, elegant human micro-motion only—soft breathing and natural chest rise, a faint head-and-shoulder shift, natural blinking and tiny eye movement, minimal facial micro-expressions that preserve the original confident smile, slight hand and finger adjustment around the glass, a gentle wrist rotation or small lift of the coupe, delicate liquid movement inside the glass caused by the motion, tiny reflections sliding across the glass surface, a subtle weight shift through the body, and barely perceptible cloth movement in the tuxedo from breathing and posture. From 7–11 seconds, gradually accelerate directly toward the television in a smooth, elegant FPV approach, keeping the man in Image 1 and the TV in a single spatial relationship, with realistic lens behavior, soft screen reflections, faint glass depth, slight perspective distortion at the screen surface, and natural falloff in depth of field as the screen grows larger and larger until it dominates the frame. The television surface must feel physically real, with subtle room reflections and believable contrast between the screen image and the surrounding bezel, and the approach must remain seamless, continuous, and stable, with no abrupt zoom, no wobble, and no transition artifacts. As the camera gets closer, the motion inside the TV should continue naturally and remain alive, with the character’s breathing, blinking, tiny gaze shifts, and champagne glass micro-movements persisting uninterrupted so the screen feels like a real moving scene rather than a frozen image. From 11–15 seconds, the camera flies through the television screen in one uninterrupted motion, crossing the glass plane invisibly so that the screen image expands to full frame and becomes the actual world of Image 2 behind it; at the exact crossing moment, perfectly match the perspective, lighting, and framing of the TV image so the transition is imperceptible, as if the television were a real window into that scene. Once inside Image 2, preserve the tuxedoed blond man exactly: his slicked-back hair shape, confident direct gaze, black bow tie, white shirt collar, champagne coupe, shallow depth of field, sparkling bokeh background, glossy highlights on glass, and opulent celebratory atmosphere, while continuing the same minimal natural motion—breathing, blinking, tiny facial changes, minute hand and glass adjustments, and subtle liquid and reflection movement—so the final image feels alive and cinematic rather than frozen. Continue a gentle deceleration after entry so the final frame settles into a composition very close to Image 2: his face sharp, the raised glass elegant in the foreground, reflective highlights crisp, background bokeh glowing softly, and the overall mood polished, cinematic, and premium. The final image should feel like a flawless Hollywood-level FPV match transition from a warm domestic interior into an opulent toast scene, with realistic parallax, physically believable motion blur, subtle smoke, rich contrast, stable facial identity, correct anatomy, consistent clothing, faithful room continuity, seamless screen-entry logic, living motion inside the TV, and a perfectly composed hero ending frame.
Shot-by-shot timeline · MiniMax H3 per the author · Aug 22, 2026
Two images are split by job: the first sets the reaper and the desert, the second is a character sheet that locks the warrior's look. Six blocks each open with a bracketed shot type and one exchange, and the final two seconds cut hard to a close-up as the defeated figure reforms behind him.
Close combat and impactReference-driven charactersChases, racing and fast movement
Three Kingdoms title sequence in Chinese poster style
Continuous action chain · MiniMax H3 per the author · Aug 22, 2026
Each reference image becomes a title card assembled from ink, torn silk, bronze patterns and smoke instead of being cut in. A surname in calligraphy appears every two images, and the last card holds longer for the closing title.
Animated styles and art directionKinetic type and title sequencesReference-driven characters
Seven character-select screens from one design sheet
Shot list · MiniMax H3 per the author · Aug 20, 2026
The camera is banned from moving at all, so the energy comes from poses and interface animation. Seven scenes alternate character and title sides, each with its own pose, gesture and interface motif, and the final scene is told to hold until the end.
Product, cars and trailersKinetic type and title sequencesReference-driven characters
Read @aimikoda's prompt
Minimax H3 Prompt:
Create a 15-second cinematic game UI montage using the reference image @[image] as the visual authority. Show all seven sins clearly in order, with exactly one character per scene: PRIDE, GREED, LUST, ENVY, GLUTTONY, WRATH, SLOTH.
Keep the camera completely static throughout. No pans, zooms, pushes, tilts, shake or reframing. All energy comes from character posing, subtle micro-movements and animated UI elements. Preserve the dark premium cyberpunk HUD style, oversized brush titles, thin tech borders, telemetry, scan lines, counters, geometric markers and neon accent colors.
Use seven clean hard-cut scene beats of roughly equal duration. In each scene, the character should hold a stylish hero pose with slight movement, as if posing for a premium character-select screen.
PRIDE: character right, title left. Proud upright pose, slight chin lift, subtle eye movement, gold reticle pulse, crown geometry, vertical light bars.
GREED: character left, title right. Calculating pose with hand near face, slight finger motion, rising asset counters, chart bars, percentage indicators.
LUST: character right, title left. Elegant over-shoulder pose, subtle breathing and gaze shift, heartbeat waveform, rotating magenta rings, soft luminous pulses.
ENVY: character left, title right. Sharp three-quarter pose, tense expression, ocular tracking, blinking iris, comparison bars and green readout jitter.
GLUTTONY: character right, title left. Heavy imposing pose, slow breathing, pulsing feeding interface, rising orange capacity gauge, circular consumption meters and waveform.
WRATH: character left, title right. Aggressive forward-facing pose, slight jaw clench, tension spikes, warning flashes, hazard symbols and red alert flicker.
SLOTH: character right, title left. Relaxed slouched pose, slow breathing, lazy blink, low-power indicators, slow waveform and softly fading blue-violet UI glow. Hold this final scene until the end.
Keep character animation restrained but let each one feel posed and presentational. No walking, large pose changes, morphing or duplicated characters. Keep titles fixed and clearly readable. Use fast HUD-based transitions between scenes without moving the camera.
The result should feel like seven premium AAA character-select screens activating one after another, with clean stylized 3D characters, stronger posing, sophisticated UI motion and strong visual consistency.
Continuous action chain · MiniMax H3 per the author · Aug 19, 2026
The agent-written prompt describes a physical path at bed height, around a cushion, over the pistol line and through cigarette smoke, and names a focus target at each stage. It restates that the woman stays identical to the reference in pose and styling, so all the movement belongs to the camera, and it ends by slowing into a held poster frame.
Camera moves and transitionsReference-driven characters
Read @CharaspowerAI's prompt
PROMPT made by MiniMax Design
Invisible micro-FPV camera skimming less than an inch above the coarse woven bedspread, the ribbed diagonal threads rushing past in extreme macro like a textured landscape, warm lamplight grazing every raised fiber while the background dissolves into creamy noir blur; the lens glides forward with a low, weightless inertia, catching a sudden silver glint from the handgun at the right edge as the first visual hook, then banking gently left around the corner of the small textured cushion that supports the woman’s elbows, so the cushion becomes a massive foreground wall sliding across frame. Focus starts razor-thin on the bedspread threads, then racks smoothly to the metallic pistol grip in her resting left hand, revealing her red-painted nails, the extended index finger along the slide, and tiny reflected streaks of warm practical light on the gun’s polished edges. Without cutting, the camera rises in a shallow arc over the pistol muzzle line, never crossing it abruptly, drifting close enough for the weapon to sweep past the lens as a blurred silver foreground occlusion, then eases toward her hand and forearm, the black wrist band briefly framing the lower edge while the bed and white pillow behind stretch into parallax layers. The movement continues as one controlled breath: a slow clockwise orbit around her upper body, passing just above the cushion and climbing toward the black scoop-neck top, where the two red teardrop pendants swing subtly against her collarbone; focus pulls from the jewelry’s glossy facets to the soft highlights on her skin, then to the dark shine of her straight black bob. The camera slips upward beside her raised cigarette hand, threading through the thin smoke drifting from the ember, the burning tip blooming as a warm macro point of light before it smears into bokeh; the cigarette and red fingernails fill the foreground while her face remains ghosted behind them, then focus snaps elegantly through the smoke onto her eyes. She remains visually identical to the reference: prone on the bed, elbows planted, head slightly tilted, feet lifted and crossed behind her, calm direct gaze, blunt bangs, dark clothing, red pendants, pistol in hand, no redesign, no change of pose logic. The camera now floats inches from her face, almost at pillow height, drifting from her cigarette hand across the line of her cheek and lips as a thin blue-toned exhale lifts past her forehead; the smoke curls around the lens as semi-transparent veils, briefly occluding one eye, then revealing it again with a controlled rack-focus from vapor texture to the sharp reflective wetness of her gaze. Momentum carries the viewpoint upward in a graceful rising bank over the bridge of her nose and bangs, grazing the glossy black hair so individual strands catch warm highlights from the lamp on frame-left and cooler ambience from the blinds behind. As the lens crests above her head, it widens slightly, revealing the room geometry: the illuminated pleated table lamp glowing on the left, closed horizontal blinds behind her raised heels, the smooth white pillow on the right, and the bedspread plane anchoring the entire scene. The camera then dives in a smooth S-curve toward her crossed glossy black stilettos, using the lifted legs as a dramatic foreground arch; focus shifts from her face to the patent leather shoes, catching tiny specular highlights as the lamp and blinds slide in opposite directions behind them, creating strong parallax without disorienting the space. The viewpoint wraps around the shoes in a technically precise half-orbit, passing close to one heel until it becomes a dark blurred vertical blade, then descending along the diagonal line of her legs back toward the bed, letting the blinds streak horizontally across the background while the warm lamp glow expands and contracts at the left edge. The movement accelerates subtly into the final escalation: the camera swoops low beside the bedspread, races past the cushion, then climbs in a smooth spiral around her right side, threading between the foreground cigarette smoke and the reflective pistol so smoke, gunmetal, red nails, red pendants, black hair, warm lamp, white pillow, and striped blinds all align in layered depth. During this final wraparound, focus choreographs a continuous chain—pistol highlight, cigarette ember, drifting smoke, pendant facets, her lips, then her eyes—each handoff motivated by the camera’s path and changing distance. The last maneuver decelerates into a poised three-quarter frontal hero composition at bed height, slightly low and intimate, with her face and eyes razor sharp, cigarette hand lifted in the foreground, pistol resting in the opposite hand, crossed heels and horizontal blinds behind, the glowing pleated lamp forming a warm practical halo at one side, smoke rising in elegant blue wisps through the frame. The final image holds as a premium neo-noir movie-poster frame: stable, iconic, high-contrast, shallow depth of field, warm practical light against cool background ambience, all textures preserved, the woman’s expression calm and unwavering, the room intact and spatially coherent, no cuts, no visible camera, only one continuous fluid FPV trajectory ending in perfect cinematic stillness.
Shot-by-shot timeline · MiniMax H3 per the author · Aug 3, 2026
An audio reference supplies vocals, timing and lyrics, one image fixes the performer, and further images are assigned to each two-second lyric window. The lyrics are quoted with their time ranges, and each line is asked to move according to its meaning, which gives the typography a reason for every motion.
Kinetic type and title sequencesReference-driven charactersSport, dance and rhythm
Read @aimikoda's prompt
Use the audio reference Audio1 for vocals, timing and lyrics. Use Image1 as the sole performer and style reference. Create a 16:9 kinetic lyric performance video where typography is the main event and the singer is the visual anchor. Preserve the performer’s flat editorial vector look, oversized pink sunglasses, long braids, microphone, jewelry and clean light background. Keep the camera close, stylish and energetic.
Lyrics on screen in sync with the song:
[00s-02s] “Follow my steps, let’s move together” Image2
[02s-04s] “Catch the rhythm, light as a feather” Image3
[04s-06s] “Start a wave, make it last forever” Image4
[06s-08s] “One by one, we make it better” Image5
[08s-15s] She dance. Image6
Make each lyric visually behave according to its meaning with strong graphic wordplay and semantic motion design. Use bold neo-brutalist typography, fashion editorial layout, glitch accents, analog texture, sharp cuts, fast scale shifts, collision motion and premium readability. Add simple background graphics tied to the lyrics.
Keep the text fully readable, highly impactful, tightly synced to the vocals, and progressively more energetic from line to line. The singer remains present throughout, but the lyrics dominate the visual rhythm.
The week's best new dramas and AI movies, in one email.
We are putting together a weekly email: the new dramas and AI movies worth your time, and any free-coin deals we find. Leave your address and you will get the first issue when it starts. No spam.