Rope, their own scarves, and a basket of eggs.
Rate this hex 1–10 🪄
(one photo and you're the witch at the table)
1
23
Make yours here. One photo, first film is free:
openclips.ai/community/v/oct…
12
Misa Rae retweeted
GPT-6.1 Sol + OpenClips | Near-Astra performance at 1/5 of the price ⚡
Pick it in ChatGPT, connect the OpenClips MCP, and make video ads, UGC clips and creatives right in the chat.
Near-Astra power. 1/5 of the price. Prompt it, post it.
#openai #chatgpt #openclips
Rebuilt this first sword-flight lesson on Seedance 2.5 and moved it to a fire kingdom: obsidian bridge, lava sea, one first-person take. The timestamped beats held, including the jump onto the flaming blade. Swap in your own face, the scene stays locked.
minimaxH3,看来对于seedance的词兼容性很不足啊
prompt:
电影级超写实质感,16:9 横屏,中文对白。
整段 15 秒严格使用男主第一人称 POV,高度约 178cm,全程不切第三人称。摄影机就是男主真实双眼,所有:
走路
冲向栏杆
低头
坠落
踩上飞剑
失衡
转弯
俯冲
爬升
都必须符合真实身体惯性。
环境设定
环境严格依据 参考图 2 构建:
一条巨大弧形白玉云桥悬于无边云海之上
右侧沿山势层层抬升的是规模极其庞大的深色木构仙宫群
拥有高耸飞檐、巨大立柱、白玉台基、宽阔长阶与连续露台
清晨暖金阳光从左侧照入
云层在建筑之间缓慢流动
人物与整座建筑群相比极其渺小
人物 ID 设定
女主
全程始终是 参考图 1 中的同一名成年东亚女性:
五官结构完全一致
眼睛一致
鼻子一致
嘴唇一致
发型一致
肤色一致
身材比例完全一致
保留真实皮肤纹理
保留自然碎发
保留清晰但克制的微表情
服装
女主全程穿同一套 月白与浅银青御剑常服:
窄袖
收束腰封
真实层叠衣料
全程同一造型,不换装。
分段结构
0–3 秒|第一帧直接抛钩子
男主第一人称正与女主并肩走在弧形白玉云桥上。
女主突然停下,直接转身向镜头伸手:
“剑给我。”
男主自己的手进入画面,将带鞘长剑递给她。
她接过后几乎没有停顿,转身直接将男主的剑越过白玉栏杆扔进云海。
第一人称立刻冲到栏杆边并低头往下看,男主脱口而出:
“你干什么?”
女主走到旁边,平静看着下方:
“剑握在手里,只能走。”
远处下坠的剑忽然减速,在云层上方自然转成水平姿态并悬停。
3–6 秒|从对白直接进入起跳
女主转头直视第一人称镜头,补完后半句:
“放下,才能飞。”
她自己的长剑从剑鞘中流畅滑出,在桥外翻转成水平飞行姿态,稳定悬于栏杆下方数米。
女主单脚踩上栏杆,身体轻盈但真实地调整重心,回头只说:
“跟上。”
随后直接向后踏出云桥。
第一人称下意识扑向栏杆并低头追随。
她先短暂自由下坠,随后双脚稳稳落在自己的飞剑上,膝盖自然弯曲吸收冲击,下一秒御剑向前加速。
与此同时,男主刚才被扔下去的剑迅速从云海中升回桥边,稳定悬到第一人称脚下。
6–11 秒|完整表现第一次御剑
男主没有继续犹豫,第一人称直接跨出栏杆。
画面出现短促但清晰的真实坠落。飞剑从下方接住双脚,身体因为冲击产生一次明显重心下沉和左右修正,镜头轻晃一次后迅速稳定。
飞剑开始加速追向前方女主。
她在数米外御剑飞行,途中回头喊:
“别看脚下,看前面!”
第一人称跟着她:
先沿巨大白玉云桥外侧向左压弯栏杆和桥面在上方快速形成强烈视差
随后降低高度从弧形桥体外缘下方掠过
雕刻白石结构从头顶高速滑过
紧接着向右切入云海上方追向右侧庞大仙宫群
女主率先贴近一座巨大宫台向上拉升。
第一人称跟随形成一条连续丝滑的爬升弧线:
右侧巨大木柱、飞檐和露台近距离掠过
左侧始终能看到极深云海
速度强烈但空间关系清楚
11–15 秒|完成第一次御剑的情绪释放
女主突然沿一座巨大飞檐转过宫殿拐角,短暂从第一人称视野中消失。
男主沿同样航线压弯追过去。
下一瞬冲出建筑阴影进入暖金晨光,女主已经在几米外与自己平行御剑飞行。
此时男主终于能够稳定站在飞剑上。
女主侧头看向第一人称,确认他没有再摇晃,嘴角轻轻扬起:
“这不是会了吗?”
男主喘着气回答:
“你管这叫教?”
她轻轻笑一下,身体向前压低:
“那第二课,追上我。”
话音刚落,她脚下飞剑立刻向前加速,沿着层叠仙宫和云海之间冲向远方。
男主第一人称随即前倾加速追上去。
最后画面保持第一人称高速飞向被晨光照亮的巨型宫殿群。
叙事理解要求
整段叙事必须第一遍就完全看懂:
女主拿走男主的剑并扔下云桥
用实际行动逼他完成第一次真正御剑
男主从不敢跳到成功起飞
最后女主开始第二课
飞行主段要求
御剑飞行必须从 第 6 秒开始持续到结尾,占据整段主要时长。
运镜全部由真实第一人称动作形成,依次包含:
低头坠落
接剑失衡
沿桥压弯
桥下掠过
云海低飞
贴宫殿爬升
绕飞檐转弯
平行高速追逐
要求整体:
丝滑
连续
可读
画面与运动要求
纯第一人称 POV
不切第三人称
不使用漂浮摄影机
不要无人机感镜头
24fps 电影运动质感
真实曝光
动作具有真实启动、重量、惯性、修正与稳定过程
云桥、栏杆、仙宫、飞檐、露台、云海必须有清楚稳定的空间连续性
Negative
blurry, bad quality, low quality, low resolution, noisy, jpeg artifacts, watermark, text, malformed hands, extra fingers, duplicated limbs, inconsistent character, face morphing, hairstyle drift, unstable architecture, broken spatial continuity, duplicated woman, duplicated swords, changing sword length, floating feet, feet detached from sword, unstable foot placement, impossible body balance, weightless body, broken gravity, random flight path, teleportation, plastic skin, over-sharpening, HDR, game-like rendering, excessive camera shake, chaotic motion blur, random camera orbit, drone-like POV, unstable eye height, incoherent speed, sliding architecture, changing bridge geometry, chaotic clouds, exaggerated facial acting
4
76
Love these videos that feel like real memories.
This is where longer AI video generations start getting interesting.
I gave Seedance 2.5 one reference image and scripted an entire 30-second portrait-session story into a single prompt.
Multiple cuts, two characters, a self-timer gag, continuity across the studio with an actual ending.
All in one generation.
@runwayml
1
2
109
Is my avatar the demon actually?
Seedance 2.5. A hammock, two lamp posts, no permits.
Prompt: "the streetlight above him turns on at dusk, spotlighting the hammock. He lifts the paper, looks straight up at it, then at the camera"
Duration: 15s. Reference: one Prism image. Rendered 480p, Topaz to 4K.
Three interruptions in escalating order: pigeon, sprinkler, then the light. The light is the one that's personal.
1
5
662
I felt real fear.
15 seconds of zombie scene with H3 Minimax.
The MiniMax H3 x Picsart Challenge is live no theme, just your best idea. $50,000 prize pool. Closes Sep 23. @Hailuo_AI @Picsart
Enter: tinyurl.com/bdfp4a6p
Try it Prompt : MINIMAX H3 — “COME TO MOMMY”
15-second photorealistic live-action horror sequence, 16:9, 7 distinct shots, new shot every ~2 seconds, premium theatrical cinematography, realistic human performances, practical lighting, grounded horror, no CGI appearance.
CHARACTER CONTINUITY:
Girl: 10–12 years old, natural appearance, shoulder-length dark hair, casual clothes, small backpack.
Infected man: late 30s, messy dark hair, gray worn jacket, dark trousers, tired face.
Shopkeeper: middle-aged man, simple store uniform.
Keep identical faces, wardrobe, proportions and appearance across all shots.
SHOT 1 — NORMAL | 0–2s
Wide establishing shot inside a small neighborhood convenience store at night. The girl casually walks through a grocery aisle with a basket, choosing snacks. Shopkeeper works behind the checkout counter. Automatic glass doors visible in the background. Everything feels ordinary and safe.
35mm cinema lens, subtle lateral camera movement.
Audio: refrigerator hum, fluorescent buzz, quiet store ambience.
SHOT 2 — THE STRANGER | 2–4s
Medium shot facing the entrance. Automatic doors open and a strange man enters. He appears heavily drunk, stumbling slightly and touching the door frame for balance. He slowly looks around. Shopkeeper notices him. Girl remains visible in foreground.
Natural handheld push toward the man.
His behavior initially looks completely believable.
SHOT 3 — SOMETHING IS WRONG | 4–6s
Close-up of the man. His stumbling suddenly stops. He becomes completely motionless. His eyes slowly focus. His head makes a subtle unnatural twitch. His breathing changes. His fingers tighten. He slowly turns toward the shopkeeper.
Locked close-up with tiny natural camera movement.
The store ambience becomes unnaturally quiet.
SHOT 4 — ATTACK | 6–8s
Fast medium-wide shot. Without warning, the man violently lunges toward the checkout. The shopkeeper recoils as the attacker grabs and slams him against the counter. Products fall onto the floor. The girl sees everything, drops her basket and runs.
Handheld camera reacts naturally to the sudden violence.
Fast realistic motion blur. No gore focus.
SHOT 5 — HIDE | 8–10s
Low camera inside a grocery aisle. The girl drops behind a shelf and covers her mouth, trying not to breathe loudly. Through a narrow gap between products, she watches the store. The infected slowly walks away from the counter. A dropped snack rolls across the floor.
Camera at the girl's eye level, partially obstructed by products.
Her frightened eye remains sharply focused.
SHOT 6 — THE SEARCH | 10–12.5s
Medium shot from behind the shelf. The infected slowly enters the aisle. His footsteps approach. He looks left, then right. The girl remains completely still. He suddenly turns toward her hiding place and stops.
Very slow camera push toward the gap between products.
Keep the infected partially obscured and human-looking.
SHOT 7 — MOMMY | 12.5–15s
Extreme close-up of the girl's face. She trembles silently, covering her mouth. From the other side of the shelf, the infected speaks using her mother's voice.
INFECTED, IN HER MOTHER'S VOICE:
“Sweetheart... come out. Mommy's here.”
The girl's eyes widen. She recognizes the voice. A tear forms. She slowly presses her hand over her mouth.
VOICE:
“Come to Mommy.”
Slow push toward her terrified expression. The infected remains a blurred silhouette through the shelf.
CUT TO BLACK.
VISUAL REALISM:
Photorealistic live-action. Natural skin texture, individual hair strands, realistic child movement, believable body mechanics, natural facial expressions, accurate clothing physics, realistic product interaction, physically correct shadows and reflections, natural motion blur, realistic depth of field.
ENVIRONMENT:
Real convenience-store materials: plastic packaging, glass refrigerators, metal shelving, paper labels, cardboard boxes and polished flooring. Maintain identical store layout across every shot.
LIGHTING:
Practical cool-white fluorescent lighting. Realistic falloff, darker spaces between aisles, natural reflections on refrigerator glass and floor. No neon lighting, colored fog, fantasy glow or excessive bloom.
CAMERA:
Professional cinema camera, primarily 35mm perspective. Each shot has a clearly different composition. Natural handheld movement during action, controlled movement during tension. No random camera movement, excessive shake, artificial zoom or slow motion.
EDITING:
Hard cinematic cuts. New visual information every ~2 seconds. Every shot advances the story. Maintain perfect character and environment continuity.
AUDIO:
Automatic door chime, refrigerator hum, fluorescent buzz, footsteps, products falling, girl's breathing and distant traffic. After the attack, reduce ambience. The mother's voice must sound warm, familiar and completely believable, making the imitation disturbing.
NON-DIEGETIC MUSIC:
Minimal low-frequency horror atmosphere after the attack. No loud jump scare or trailer-style impact.
FINAL TARGET:
A premium photorealistic horror-film scene with clear storytelling:
NORMAL STORE → STRANGER ENTERS → BEHAVIOR CHANGES → ATTACK → GIRL HIDES → INFECTED SEARCHES → MOTHER'S VOICE TRAP.
No generic AI faces, plastic skin, warped hands, duplicated people, changing wardrobe, inconsistent store layout, meaningless cuts or exaggerated monster design.
1
3
133
I made one like this, but from the perspective of just one person. Things are starting to heat up: openclips.ai/community/v/one…
Seedance 2.5 on @TapNow_AI
Prompt
PROTAGONIST: Use @[Image 1](image_1) as the exact performer identity reference. Preserve her facial features, hairstyle, skin tone, and natural body proportions throughout.
AUDIO / PERFORMANCE MASTER: Use @[Video 1](video_1) as the absolute audio, vocal, rhythm, and performance reference at its original playback speed. Every dance accent, body groove, facial reaction, camera acceleration, crowd response, and transition must respond naturally to the actual beat structure. Whenever vocals occur, the protagonist performs precise frame-accurate lip-sync with convincing phoneme articulation, syllable timing, rap cadence, breaths, pauses, and expression. During instrumental passages, she stops articulating lyrics naturally while continuing the choreography.
ENVIRONMENT / DAYLIGHT CORE: Typical details baked in: multi-story school building, paved schoolyard (운동장), benches and trees, students in recognizable Korean uniforms (blazers/vests, skirts or slacks, some in PE kits), phones
6
72
I tried to make one with olive oil. How did I do?
Video created on Veo 3.1
Prompt
{
"commercial_specifications": {
"title": "From Highland Cherry to Liquid Gold",
"duration_seconds": 10,
"aspect_ratio": "16:9",
"resolution": "4K",
"style": "Cinematic photorealistic artisanal coffee commercial",
"lighting": "Misty mountain dawn transitioning into moody, dramatic studio rim lighting",
"depth_of_field": "Extreme macro shallow depth of field",
"audio_visual_notes": "No text, no logos, hyper-realistic fluid and particle dynamics"
},
"storyboard_timeline": [
{
"timecode": "00:00 - 00:02",
"scene": "The Highland Origin",
"shot_type": "Cinematic macro push-in",
"action": "A deep crimson coffee cherry clings to a dew-covered branch on a misty high-altitude plantation. The camera rushes toward the glossy skin as early sunrise beams strike the dew.",
"visual_details": "Vibrant ruby-red skin, micro water droplets, morning mist, dewy emerald leaves in soft focus",
"lighting_atmosphere": "Crisp morning dawn with diffused atmospheric sunbeams"
},
{
"timecode": "00:02 - 00:04",
"scene": "The Roast Metamorphosis",
"shot_type": "Dynamic rotating macro shot",
"action": "The cherry splits open effortlessly in zero gravity, releasing two pale green beans that spiral upward. As they tumble through a vortex of subtle heat distortion, they darken into glossy, dark-roasted beans sweating natural oils.",
"physics_effects": "Floating bean rotation, fine wisps of aromatic vapor, heat shimmer, delicate chaff flaking off",
"transition": "Seamless elemental transformation with continuous spinning camera motion"
},
{
"timecode": "00:04 - 00:07",
"scene": "The Grind & Bloom",
"shot_type": "High-speed orbital tracking",
"action": "The roasted beans shatter in mid-air into coarse, textured grounds. A crystalline stream of near-boiling water spirals down through the grounds, triggering an immediate, airy bloom with micro-bubbles of crema.",
"visual_details": "Granular ground coffee texture, rising steam swirls, glistening golden crema bubbles, realistic liquid absorption",
"lighting_atmosphere": "Moody studio lighting with warm amber backlighting"
},
{
"timecode": "00:07 - 00:09",
"scene": "The Extraction",
"shot_type": "Smooth cinematic orbit",
"action": "Dense, viscous espresso streams downward in a thick, honey-like dual ribbon into a heavy-bottomed crystal glass, layering into a deep mahogany body topped with a dense, velvety crema.",
"visual_details": "Hyper-reflective glassware, swirling tiger-stripe patterns in crema, rich viscosity, subtle condensation on the glass base"
},
{
"timecode": "00:09 - 00:10",
"scene": "The Crema Swirl",
"shot_type": "Ultra-macro high-speed slow motion",
"action": "A single drop falls into the center of the glass in 1000fps slow motion, forming a perfect crown splash that ripples across the velvety crema without breaking it.",
"texture_focus": "Suspended liquid droplets, elastic surface tension, micro-foam cellular structure, warm light refraction"
}
],
"prompt_parameters": {
"camera": "Macro push-in transitioning into rotating tumble, dynamic spiral tracking, and ultra-high-speed slow motion",
"camera_motion": "Continuous, unbroken kinetic camera path with matched-motion transitions",
"render_qualities": [
"Photorealistic fluid and foam simulation",
"Extreme macro coffee grounds texture",
"Cinematic mountain atmosphere",
"Artisanal luxury beverage aesthetic",
"True-to-life heat distortion and vapor physics",
"Seamless organic transformation",
"Warm amber and deep mahogany palette",
"Ultra-shallow depth of field",
"Ray-traced liquid reflections",
"Ultra-detailed 4K"
]
1
52
Misa Rae retweeted
Replying to @gpumaxxer
openclips.ai seems to have lower pricing, how about you lower 20% more?
Lara Croft vibes will never get old.
Made with MiniMax-H3 ComfyUI
🎬Image-to-Video
📷“Not a Signpost, but a Trap”
She thought she had finally found the entrance; but she soon realized that it wasn’t a clue left by those who came before her, but rather some kind of living mechanism inside the temple that was “luring her closer.”
MiniMax-H3 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] A 15-second, 16:9, high-stability cinematic adventure-thriller image-to-video sequence extending only inside the same ancient temple corridor shown in <Picture 1>. Preserve the exact same young female explorer throughout: same face structure, age impression, wet skin texture, realistic hair strands, clothing, and body proportions. Keep only this corridor system: high arched stone ceiling, broken dome skylight, tall statues on both sides with circular waist medallions, damp stone floor, moss, vines, dust, low braziers, and old gears as static set dressing only. No exterior, no new halls, no extra people, no subtitles, no watermark, no face drift, no body deformation, no finger errors, no flicker, and no cut to black. Story logic: she discovers what looks like a treasure marker, follows it deeper, steps onto a selection mechanism, then realizes the trail was not leading to treasure but drawing her toward something waiting ahead. She remains in continuous motion; all emotion is attached to sliding, stepping, leaning, crouching, touching stone, braking, turning, and attempted retreat. Camera language stays intimate and grounded: tight side-follow, slight push-in, over-shoulder follow, short lateral move, and a final suspended lock. [Shot 2] At 00:00-00:02, a low-angle tight near shot from her left-front waist height moves in close side-follow as she slides out from behind a stone pillar. The first frame already contains her. She steps down from a shallow stone rise with her right leg while her left palm glides along a damp carved relief for support. Her body stays low, breathing intentionally shallow, and her gaze drops toward the floor ahead. Her first emotional layer is highly controlled vigilance: brow peaks press inward only slightly, eyes scan the dark ahead then cut quickly down, lids narrow, pupils focus sharply, nostrils move with restrained breathing, lips part by a thin line without sound, jaw stays stable, and neck muscles show mild tension. Backlight from an arch behind outlines wet hair edges, collarbone, shoulder strap, and the side of her waist, while most of the face remains in partial shadow. Dust disturbed by her slide-step drifts through the beam with directional inertia. Hair trails for a beat and settles. [Shot 3] At 00:02-00:04.5, keep the same axis and raise the camera to chest height with a gentle push into a medium close shot. She leans out farther from the pillar and locks onto a fresh golden powder trace lodged in the floor-tile seams. Her feet stop, the body tilts forward half an inch, the left hand remains on the pillar, and the right hand starts to lower toward the ground. Her second emotional layer is a brief flare of analytical excitement: vigilance remains in the eyes, but focus sharpens at the sight of the trail, the brow center loosens for half a beat and closes again, and one mouth corner lifts by the smallest amount before restraint returns. She takes one subtle inhale through the nose and the throat moves as she suppresses the excitement. She murmurs in a low breath-close voice, <d>[English] Someone got here first.</d> The golden powder is not magical glow but fine mineral dust embedded in the seam, flashing softly under grazing light. [Shot 4] At 00:04.5-00:07.5, in a close shot, begin on her fingers touching the golden powder, then smoothly lift with her hand to her side face and end in an over-shoulder composition that includes deeper corridor space. She lowers into a half-crouch, brushes the powder with two fingers, and lifts them to inspect the residue while still shifting half a step forward. Her gaze passes her fingertips and moves deeper under the far arch. Her third emotional layer is rational confirmation: eyes narrow slightly, inner corners tighten, the brow draws together with more concentration but no heaviness, lips close then loosen again, and the nose tip and nostrils show delicate tension from close inspection. When she raises her eyes deeper into the corridor, a dangerous curiosity appears for the first time. Her face falls deeper into shadow during the crouch, while the gold dust on her fingertips flashes a small warm reflection in backlight. Near the deeper arch, a second and third trace of golden dust becomes faintly visible in sequence as if her movement disturbed them. [Shot 5] At 00:07.5-00:10.5, over-shoulder medium shot with natural perspective as the camera follows her forward two steps, then stops abruptly at the key moment. She follows the golden traces and places full weight onto a central stone tile. The instant the weight settles, a very light mechanical bite clicks beneath her. Her whole body locks in place. The right shoulder stiffens, the ankle and calf freeze, the left hand reflexively starts to rise toward the nearest pillar, and her breath catches high in the chest. Her fourth emotional layer is the body recognizing danger before the mind completes the thought: eyes cut sharply downward, lids tighten, pupils contract, the brow center locks, nostrils open a fraction and then stop, all softness leaves the mouth, the lip line narrows, and the jawline hardens. It reads as a trained freeze, not exaggerated fright. As her weight compresses the tile, the boot sole drags microscopically across stone, fine dust squeezes from the seam, and tiny grains roll from the edge. Her sudden stop creates small delayed inertia in the hair, straps, fabric, and upper torso before they settle. [Shot 6] At 00:10.5-00:12.8, cut to a side half-body close shot with a short controlled lateral move that follows her turn and attempted retreat. She snaps her head back to retreat, but after only half a step she sees that two floor tiles behind her have silently risen half an inch, creating a subtle break line in the escape path. She instantly plants her left palm against the pillar edge, tightens the core to stop her body, and lifts the right arm halfway as if preparing to pivot around the pillar and escape. Her fifth emotional layer is controlled fear merged with rapid calculation: the brow presses lower, the eyes scan the retreat path and then cut back toward the dark opening ahead with precise motion, nostrils widen more visibly with faster breathing, the lips part for an instant and clamp shut again as she swallows the urge to gasp, and the cheek tension and jaw remain locked. At the same moment, both low candles shiver once in sync. Their flames dip under a brief air-pressure pulse and rebound. Fine golden dust begins to reverse-flow through the pillar seams and floor carvings toward the deeper arch, not as a violent magical blast but as a slow unseen draw. The lifted floor tiles shed dust, sand, and rubbing powder. [Shot 7] At 00:12.8-00:15, a medium close shot with a slight low angle pushes in half a step and then suspends without cutting to black. She is half-turned, right arm raised and ready to use the pillar for a vaulting escape, but the motion stops before release. In the deepest darkness beneath the arch ahead, a single tiny but razor-sharp golden reflection appears, like a beast eye, fixed directly on her. She does not recoil; she is forced into stillness by the correction of understanding. Her sixth emotional layer is cold recognition after a shattered assumption: the gaze locks fully for the first time, pupils fixed on that point of gold, upper lids slightly raised, lower lids tightened, brows lifting only by a thread as her previous judgment is overturned, a short controlled exhale slips through the nose, lips part slightly, mouth corners press down, and the jaw remains tense. She says in a quieter, colder throat-close voice, <d>[English] That’s not a trail.</d> with “not” the lightest but hardest word and “trail” cut very short. All the golden dust and fine particles continue to stream toward that single reflective point. Her face remains mostly in shadow, with only the eye corner, nose bridge edge, wet lip line, and collarbone rim caught by mixed cold reflection from the arch and warm low candle bounce. The golden point stays tiny, sharp, and singular, never expanding into a full creature or large supernatural display. Hold on this suspended threat state, with particles still moving and the corridor still alive with pressure.
overall_soundscape: Maintain a realistic ancient temple acoustic bed throughout: low hollow room resonance, distant dripping water, faint insects, candle burn, boot friction on damp stone, palm sliding across humid carved relief, fine grain contact as fingers skim the golden powder, and subtle fabric and gear movement. Emphasize shallow controlled breathing, the side-step from behind the pillar, dust disturbed into the light beam, the quiet stop as she notices the medallion, fingertip contact with the gold dust, tiny grains lifting and falling, the crisp mechanism bite click under her foot, a brief oppressive reduction in ambience under her frozen breath, the silent rise of the tiles, sand sliding from edges, her palm planting against the pillar, the synchronized candle flutter, and the faint inward draw of air carrying dust toward the arch. End with corridor resonance, low candle crackle, drifting particulate movement, restrained breathing, a tiny pressure in the air around the golden point, and her cold close line <d>[English] That’s not a trail.</d>
non_diegetic_music: Keep music minimal and subordinate to the physical environment. Use only a very thin tension bed in the first half, almost below perception, then tighten the low-frequency pressure slightly after the mechanism click. Do not swell into action music or melodrama. End suspended on unresolved tension.
1
50
Maybe a bit late, but had to try it out myself.
Seedance 2.5 on @DomoAI_
Prompt
AN EXCITING TAIWAN TRAVEL VLOG — 30 SEC MUSIC CUT
SUBJECT: one fictional adult Korean woman in her early twenties, with large round dark eyes, a small nose, coral-red lips, long side-parted black waves and small cubic-zirconia stud earrings. Keep the face established in the first shot consistent throughout. Visible pores and natural skin texture, without doll-like smoothness or a beauty filter.
OUTFIT: throughout all 30 seconds, she wears an indigo denim one-shoulder crop top and asymmetric-hem skirt, burgundy leather belt and pointed burgundy ankle boots. Keep the exposed shoulder, stitching and direction of the longer skirt panel consistent with their appearance in the first shot. Her outfit and hair do not change between locations.
CAMERA: the texture of a personally filmed phone travel video. She films the selfies. An off-screen travel companion films the two-handed chicken-cutlet handoff, her back in the market, and the restaurant food and eating shots. Changing operators does not introduce filming equipment or device UI into the image. Landscape 16:9, no black padding. Ordinary travel shots use a 26mm-equivalent phone wide angle, enough depth of field to read the surroundings, and natural iPhone colour. Handheld movement follows real footsteps; exposure and focus catch up slightly late. A fingerprint on the lens causes local light smearing beside strong light sources. No cinematic colour grade, studio lighting or added post-production glow. People in the surroundings move independently and do not stare into the lens for long.
EXCEPTIONS: only the three restaurant shots from 17 to 23 seconds use a DV-tape texture, light noise and slightly delayed focus. This deliberately differs from the travel footage; phone texture returns at 23 seconds. The food close-up at 19–21 seconds and the sunset focus change at 23–26.5 seconds follow their own shot-specific depth-of-field directions.
MUSIC: upbeat BGM carries the excitement and rhythm across the travel montage. Location sounds do not repeatedly interrupt the music. Only the gondola section has a one-beat drop. Dialogue is brief and relaxed, with audible breaths and complete word endings. Mandarin is limited to the opening and closing greetings, delivered as newly learned phrases. The ㅋㅋㅋ in the dialogue means laughter, not spoken letters.
PROPS: the bubble-tea clerk directly hands her a clear cup; the chicken-cutlet stall owner directly hands her the fried-chicken bag. In the xiaolongbao restaurant, the waiter sets the steamer with its lid on the table and she eats with chopsticks in her right hand and a spoon in her left. Food and containers are used only in their respective food scenes.
(0:00–0:02) — Guesthouse window, Taipei morning. She films a selfie from a low angle on her phone, smiling broadly. Tilt Up extends the frame above her face to the view through the window. Taipei 101's green-glass tower rises beyond low grey buildings, rooftop water tanks and air-con units. Exposure catches up with the bright sky. Dialogue: "你好~ 대만 왔다!" Transition: Hard cut on beat.
(0:02–0:04) — Main-road junction. At the crossing, she turns the phone into selfie position. A packed queue of helmeted scooter riders leaves the white stop line as the light turns green, passing behind her. Overcast daylight, Handheld Selfie and deep focus. Dialogue: "스쿠터 뭐야 미쳤어ㅋㅋㅋ" Transition: Fast cut on beat.
(0:04–0:06) — Bubble-tea counter. She points the phone in her left hand downward to film. Her right hand reaches out first. With arm and torso visible, the aproned clerk lowers the clear cup upright into it. After she supports its base and side, the clerk releases it. The sealed film lid faces up; black brown-sugar streaks line the cup. A wide straw rests on the counter. Insert, briefly Static. Transition: Snap cut.
(0:06–0:08) — Chicken-cutlet stall. The companion films from beside her, including both her hands and the owner. The large fried cutlet, seasoned with pepper and chilli, is already in a white paper bag. She extends both hands first. With arms and torso visible, the owner lowers the bag with its opening facing up. Only after her two hands support the bottom does the owner release it. Steam rises above the bag. Handheld Insert under a daylight awning. Transition: Cut at the end of the handoff.
(0:08–0:10) — Longshan Temple. She starts with a selfie and raises the phone. Tilt Up shifts the view from her face to incense smoke, red lanterns and the dragon-carved roof. The sky above the eaves is bright in natural daylight. Transition: Cut during the upward movement.
(0:10–0:12.5) — Jiufen stepped lane. She extends her arm high for a selfie and speaks. As she tilts the phone downward in a Tilt Down, her face leaves the frame and layers of red lanterns along the narrow stone stairway and wooden teahouse windows on both sides come into view. A misty afternoon; light fog veils the lower stairs. Dialogue: "여기 진짜 영화 속인데" Transition: Cut while looking down the stairway.
(0:12.5–0:15) — Dihua Street market. The companion films a waist-height Follow Shot from behind her. She raises the chicken-cutlet bag in one hand and takes a bite while walking through the passage. On either side of her indigo denim outfit, sacks of dried seafood, dried shiitake, tea tins, red envelopes and herbal-medicine drawers pass in sequence. Other people independently browse or walk by. Overcast afternoon light. Transition: Cut on forward movement.
(0:15–0:17) — Maokong gondola. In this scene she is not holding food or its bag. She films downward through the glass floor, holding the phone at knee height in one hand. Green tea fields drift beneath her feet and a cable shadow passes across them. Low Static. The music drops for one beat and returns. Transition: Snap cut.
(0:17–0:19) — Xiaolongbao restaurant, DV texture begins. She is already seated; the companion films a Static Medium from eye level across the table. The waiter sets a bamboo steamer upright on the table with its lid on, then lifts the lid. Six xiaolongbao on white paper, shredded ginger and black vinegar. She watches and speaks while the waiter performs the action. Dialogue: "이거 육즙 터진다며?" Transition: Cut to the open steamer.
(0:19–0:21) — Broth close-up. The companion films a downward-looking Macro from beside the table, with shallow depth of field. One xiaolongbao is already on the spoon in her left hand. She opens the skin slightly with the tips of the chopsticks in her right hand; broth flows into the spoon. Steam brushes past the lens. DV texture continues. Transition: Cut from the filled spoon.
(0:21–0:23) — First mouthful. The companion films a close Handheld shot from across the table. Begin immediately after her first mouthful of xiaolongbao. She lowers the spoon in her left hand and smiles toward the lens. White restaurant fluorescent light, DV texture continues. Dialogue: "미쳤다 진짜." Transition: Cut to sunset footage with phone texture.
(0:23–0:26.5) — Tamsui Fisherman's Wharf, signature. Phone texture returns. She walks beside the railing while filming an extended-arm selfie. The white Lover's Bridge and its reflection on the sea catch orange sunset light; her long waves and asymmetric denim hem move in the wind. Rack Focus: focus moves from the initially sharp bridge behind her to her face, leaving the bridge softly blurred. Actual sunset light, without added post-production glow. 60% slow motion. Transition: Cut when focus settles on her face.
(0:26.5–0:30) — Shifen railway, night. She films a selfie with the phone extended in her right hand. Her left hand releases the lower edge of a red sky lantern bearing brush calligraphy. The lit lantern leaves her fingertips and rises into the sky. Shop lights and other sky lanterns remain around her. She waves at the lens with her now-free left hand while speaking. Dialogue: "그럼 모두 再見~!" That left hand covers the lens, ending at 30 seconds. Natural night phone exposure.
CONSISTENCY: preserve the same face and single denim outfit. Contact between hands and props is visible, without smeared fingers. Do not add subtitles, graphic text, logos or watermarks to the footage. Writing that belongs to the location or props, including the lantern calligraphy, remains. No filming equipment is visible in frame.
FINALLY: skin and movement should look like real travel footage, not AI-generated imagery. Dialogue remains clear within natural breathing, without slurred syllables or clipped word endings. Expressions and gestures respond to the situation rather than posing like an advertising model. #DomoAI #DomoAICPP
1
4
144
If you want your face in it, you know what to do:openclips.ai/community/v/one…
22
This is doing some real directing. I wanted to try it out in more realistic style, because it reminded me so much of the retro cart games.
There's only one way forward, we must accelerate.
Runway + Fable 5.1
Prompt share:
Bunny Grand Prix — Seedance 2.5 prompt
Settings used: Seedance 2.5 · 30s · 21:9 · 1080p · native audio on
IDENTITY — CINEMATIC, CUT, ARRI Alexa 35 paired with a high-speed Phantom for the slow-motion inserts, 35mm anamorphic 2.39:1, motorsport broadcast meets prestige sports documentary. Golden-hour sun low on the horizon at 5200K raking across the track from frame right, long dramatic shadows, heat shimmer rising off the asphalt, rich saturated color, fine organic film grain, gentle halation on chrome and headlights, anamorphic oval bokeh and thin horizontal blue streaks on every hot highlight. Photorealistic live-action rabbits in miniature racing karts — real fur, real whisker physics, real ear weight whipping in the wind, real paw contact with tiny steering wheels. No cartoon stylization. No anthropomorphic faces. No talking mouths. The drama comes from real rabbit behavior at speed, real kart physics, and epic cinematography.
CAST — Driver 1 (Hops): adult white rabbit, medium build, long floppy ears streaming straight back in the wind, pink nose, dark glossy eyes, leaning hard into every corner. WARDROBE: tiny red satin bow tie flattened against his chest by the wind, no other clothing. KART: miniature open-wheel go-kart, glossy candy-apple red with a white racing stripe and the number 1 on the nose, chrome side pipes. Personality: the hotshot who overdrives every corner. CAST — Driver 2 (Thumper): adult chocolate-brown Holland lop rabbit, heavy droopy ears pinned back by speed, wide flat face, half-lidded eyes, sitting perfectly upright and motionless behind the wheel, jaw slowly chewing. WARDROBE: miniature black felt top hat with a thin grey ribbon band, somehow not blowing off, no other clothing. KART: matte black go-kart with a gold pinstripe and the number 2, no wasted movement, no drama. Personality: the deadpan veteran who never lifts and never flinches. CAST — Driver 3 (Nibbles): small young grey dwarf rabbit, dense soft coat, enormous ears for his body vibrating in the airflow, nose twitching at double speed, tiny paws gripping the wheel too tightly. WARDROBE: oversized black plastic novelty glasses with no lenses, slipping down his nose at every bump, no other clothing. KART: undersized sky-blue go-kart with a wobbling rear wheel and the number 3 painted slightly crooked. Personality: the terrified rookie who somehow ends up in the fight for the lead.
SYNOPSIS — Three rabbits contest the final lap of a miniature grand prix on a sprawling scale-model race circuit at golden hour. What begins as a clean three-wide sprint down the main straight escalates into an epic duel through a sweeping banked curve, a hairpin under a footbridge, a chicane lined with hay bales and tiny grandstands, and a long hill climb into a blinding sunset — with a mid-race spin, a slow-motion carrot-on-track hazard, a photo finish, and a final champagne-free podium where the deadpan lop accepts victory by simply chewing. Tone is genuinely epic: massive scale despite the miniature subjects, thunderous sound design, sweeping camera moves, and the absurdist comedy of three actual rabbits treating a kart race like the most important event in history.
LOCATION — A sprawling 20x20-foot scale-model racing circuit built outdoors on a rooftop or open field at golden hour. Real asphalt-textured track surface about 14 inches wide with painted white edge lines and red-and-white kerbing at the corners. Sections: a long main straight with a start-finish gantry, a wide banked sweeper with a tiny Armco barrier, a tight hairpin passing beneath a small wooden footbridge, a chicane lined with real miniature hay bales and small painted tire stacks in red and white, and a rising hill climb section that crests directly into the setting sun. Miniature grandstands with rows of tiny empty plastic seats catch the light. Miniature pit garages with rolled-up doors and dangling flags line the straight. A tiny race-control tower with a checkered flag mounted on top. Scattered real carrots and lettuce leaves as track debris. Real dust and small pebbles on the track edges. Beyond the circuit: hazy distant treeline and a vast sky graded from deep orange at the horizon to violet overhead, wisps of cloud catching gold. Long shadows from every object stretching across the track.
SHOT LIST — six beats, roughly 5 seconds each, hard cuts between shots, camera movement exactly as specified.
[0:00–0:05] SHOT 1 — THE FINAL LAP. Low tracking shot, camera mounted at track level about 6 inches above the asphalt, 35mm anamorphic, racing alongside the three karts down the main straight at full speed, the start-finish gantry whipping overhead as they cross the line for the final lap. Three-wide: Hops on the left in red, Thumper center in black, Nibbles on the right in sky blue. Ears streaming straight back on all three. Motion blur on the track surface and the grandstands, the karts sharp. Heat shimmer distorts the far end of the straight. Golden sun rakes across the karts from frame right, blowing out the chrome pipes into anamorphic streaks. A tiny race marshal's checkered flag waves in the foreground blur — final lap. Hops edges forward, leaning over his wheel. Thumper doesn't move a whisker. Nibbles' glasses slide down and he tilts his head back to catch them at 40 miles an hour.
[0:05–0:10] SHOT 2 — THE BANKED SWEEPER. Cut to a soaring drone shot, 24mm wide, camera sweeping in a high orbiting arc from directly behind the karts as they enter the wide banked curve, the whole circuit revealed below like a cathedral of asphalt and gold light — the hairpin, the chicane, the hill climb, the grandstands all laid out in miniature glory, shadows of the karts stretching long across the banking. The three karts lean into the curve, Hops taking the high line on the banking, Thumper the middle, Nibbles hugging the bottom. Hops carries too much speed — his rear wheels step out, a puff of dust, a slide caught with a flick of his paws on the wheel. The camera continues its arc, sun sweeping behind the trio and flaring across the lens in a wide horizontal blue streak. The scale of the circuit against the vast sunset sky makes three rabbits in go-karts look like the most important race ever run.
[0:10–0:15] SHOT 3 — THE HAIRPIN UNDER THE BRIDGE. Cut to a locked-off frame on the exit of the hairpin, 85mm long lens, camera positioned ahead of the corner looking back so the karts approach head-on through the shadow beneath the small wooden footbridge and burst into golden light as they exit. Deep compression of the long lens stacks all three karts in frame. Thumper leads now, perfectly calm, top hat level. Hops dives inside, karts almost touching, sparks of gravel kicking off the kerb. Nibbles brakes far too early behind them, his little kart's rear wheel wobbling violently. In the middle of the corner, a real carrot lies on the racing line. Rack focus snaps to it. Hops runs straight over it — the carrot bursts, orange fragments flying in slow motion at 120 frames per second, catching the sun like sparks. Snap back to real time. Hops' kart lurches, his ears flip forward over his face, he is momentarily driving blind.
[0:15–0:20] SHOT 4 — THE SPIN. Cut to a handheld chase cam, gimbal-stabilized but alive, 35mm, tracking close behind Hops as he charges into the hay-bale chicane still with his ears over his eyes. He flips them back just in time to see the first hay bale. He yanks the wheel. The red kart snaps into a full 360-degree spin — the camera whips with it, the world rotating: grandstands, hay bales, sky, sun, hay bales, grandstands — a plume of golden dust erupting around him, a hay bale clipped and exploding into drifting straw that catches the light. Through the dust, Thumper's black kart threads the chicane untouched, not deviating an inch, top hat level, chewing. Nibbles enters the chicane behind them, sees the chaos, panics, closes his eyes completely — and drives a perfect line through it by pure luck, emerging in second place with his eyes still shut, glasses hanging from one ear. Hops comes out of the spin facing the right direction and floors it, now third.
[0:20–0:25] SHOT 5 — THE HILL CLIMB INTO THE SUN. Cut to the most epic frame in the film: a low wide shot, 35mm anamorphic, camera locked at the crest of the hill climb looking back down the rising track directly into the setting sun, which sits huge and orange on the horizon at the exact top of the hill. Three tiny silhouettes charge up the incline toward the camera, backlit, dust glowing gold behind them, three long shadows racing ahead of the karts down the hill. Slow motion at 60 frames per second. Thumper leads, a black silhouette with a top hat. Nibbles second, ears enormous in silhouette. Hops third, closing fast, red kart glowing at the edges. As they near the crest the sun flares across the entire frame, a massive horizontal anamorphic streak, the karts becoming pure shapes of light and shadow. The slow-motion lets every strand of fur catch the rim light. This is the shot that makes rabbits in go-karts look like legend.
[0:25–0:30] SHOT 6 — PHOTO FINISH AND PODIUM. Cut to a hard side-on finish-line camera, 50mm, locked-off, the checkered line painted across the asphalt dead center of frame, the gantry and grandstands beyond. The three karts explode into frame from the left in real time — a blur — then a hard freeze-frame at the exact moment they cross: Thumper's nose ahead by a whisker, Nibbles half a wheel back, Hops a full kart behind. Hold the freeze for one beat. Then cut to a static wide of a miniature three-step podium beside the track, 35mm, the sun now a sliver on the horizon, sky gone deep violet and orange. Thumper sits on the top step in a perfect loaf, top hat slightly askew, chewing. He looks at the camera. He does not react. Nibbles on the second step, eyes still closed, glasses fully off and lying on the step beside him, nose twitching at triple speed. Hops on the third step with straw stuck in his fur and bow tie sideways, still panting, one ear inside out. A tiny checkered flag flutters above them. A carrot rests on the top step as the trophy. Thumper leans down and takes one slow bite. Hold on the tableau. Cut to black.
PERFORMANCE NOTES — All three rabbits move with true rabbit behavior even at speed: ears react to wind physically, noses twitch constantly, paws grip the wheel with real rabbit anatomy, bodies lean with genuine weight into corners. Karts obey real physics: weight transfer under braking, rear-wheel slides with dust, suspension compression over kerbs, spin dynamics with rotational momentum. Wardrobe behaves physically in wind: bow tie flattens, glasses slide and fall, top hat defies all logic and stays put until the podium. Thumper's stillness is absolute throughout the race — his calm under chaos is the anchor of every joke and the reason he wins.
CAMERA NOTES — Anamorphic characteristics throughout: oval bokeh, horizontal blue flares on the sun and every chrome highlight, slight barrel breathing on rack focuses. Consistent fine film grain. Speed ramps exactly where specified: 120fps on the carrot burst, 60fps on the hill climb, hard freeze at the finish line. Motion blur honest to shutter angle at 180 degrees during real-time shots. Golden-hour continuity: the sun sits low throughout and dips to a sliver by the podium.
SFX — Thunderous, epic, sports-broadcast sound design scaled up as if these are real racing engines: three distinct high-revving miniature engine notes — Hops' red kart a screaming high buzz, Thumper's black kart a deep steady growl, Nibbles' blue kart a sputtering uneven whine with a rhythmic wheel wobble. Doppler shifts on every pass. Tire scrub and squeal on the banking, gravel spray hitting Armco barriers, wind roar in the tracking shots. Shot 3: a low reverberant echo as they pass under the footbridge, then a wet crunch and slow-motion sub-bass thump as the carrot bursts, fragments pattering on asphalt. Shot 4: shrieking tire spin, hay bale impact as a dry explosive whump and rustling straw, engines howling through the dust. Shot 5: engine notes stretch and deepen into slow motion, wind becomes a low roar, a sub-bass swell building to the crest. Shot 6: engines scream to a peak, a hard cut to total silence on the freeze-frame, then quiet evening ambience on the podium — distant birdsong, a light breeze fluttering the checkered flag, three rabbits breathing, one slow crunch of a carrot bite. No music. No voices. No crowd. The engines are the orchestra and the silence is the applause.
1
39











