Repository navigation
Presets
The same full prompts as in the Lab's «Preset ▾» menu (there they are inserted directly, in English or in your language): copy one and adapt it — the hero, the place, the actions. «Line to camera» is tested in Remiqora Video; the rest are written from the MiniMax guide and our experience, and results may vary — shoot 2–3 takes and pick one.
A new scene from text: a person, a place, three actions.
Realistic live-action cinematic look, golden hour sunlight, shallow depth of field, film grain.
Scene: an adult woman, about 30, with long wavy red hair and freckles, wearing a white linen shirt, stands on a wooden pier by a calm lake at sunset.
Camera: medium shot, slow push-in.
Timeline:
[0s-2s] She looks out over the water.
[2s-4s] She turns her head toward the camera.
[4s-5s] She smiles softly.
Audio: gentle acoustic guitar at a slow tempo. Water lapping against the pier, distant birds.
Keep her face, hair and outfit exactly the same. No text, subtitles, logos or watermarks.
The hero speaks; the words in quotes are in the language he should speak.
Realistic live-action cinematic look, warm evening lamp light, shallow depth of field, film grain.
Scene: an adult man, about 40, with short grey hair and a trimmed beard, wearing a dark wool sweater, sits at a kitchen table at night.
Camera: medium close-up, static shot, his face clearly visible.
Timeline:
[0s-1s] He looks up at the camera.
[1s-4s] He speaks in a low, tired voice, in English: "Sit down. We need to talk."
[4s-5s] He folds his hands on the table.
Audio: his low calm voice speaking English, quiet room tone, a clock ticking, no music.
No text, subtitles, logos or watermarks.
Length 8 s. The characters differ by clothes and by place in the shot, the lines come in turns, each with the speaker's feature.
Realistic live-action cinematic look, soft window daylight, shallow depth of field, film grain.
Scene: a small café by a big window. On the left, an adult man, about 35, with short black hair and a grey wool coat. On the right, an adult woman, about 30, with a red bob and a yellow raincoat. They sit across a small table with two cups of coffee. Only these two people are in the scene, nobody else.
Camera: medium shot, static shot, both faces clearly visible.
Timeline:
[0s-3s] The man in the grey coat leans forward and says in a low calm voice, in English: "So you came after all."
[3s-4s] The red-haired woman looks at him and smiles.
[4s-7s] The red-haired woman answers in a soft amused voice, in English: "I promised, didn't I?"
[7s-8s] The man in the grey coat laughs quietly.
Audio: two voices speaking English, quiet café chatter in the background, cups clinking, no music.
No text, subtitles, logos or watermarks.
No speech: each has their own features, place and action.
Realistic live-action cinematic look, golden hour sunlight, film grain.
Scene: a sunny park meadow. On the left, an adult woman, about 25, with long blond braids and a white sundress. In the middle, an adult man, about 30, tall, with a shaved head and a green T-shirt. On the right, an adult woman, about 40, with short curly grey hair and a denim jacket. Only these three people are in the scene, nobody else.
Camera: full shot, static shot.
Timeline:
[0s-2s] The man in the green T-shirt throws a red frisbee to the left.
[2s-4s] The blond woman in the white sundress jumps and catches it.
[4s-5s] The grey-haired woman in the denim jacket claps and laughs.
Audio: a light ukulele at a cheerful tempo. Birds, laughter, the whoosh of the frisbee.
No text, subtitles, logos or watermarks.
Each event's sound in its own timeline line, no music.
Epic fantasy film look, dramatic dusk light, smoke and embers, high contrast.
Scene: a stone castle wall at dusk, burning torches along the battlements. An adult knight in dented steel armor and a helmet stands on the wall with a round shield.
Camera: wide shot, the camera shakes slightly.
Timeline:
[0s-2s] The knight raises his shield and looks up at the sky.
[2s-3s] A fireball hits the wall below him and explodes with a deafening boom.
[3s-5s] Stones and burning sparks fly through the smoke as the knight crouches behind his shield.
Audio: no music. A low rumble before the impact, a deafening explosion, falling stones, crackling fire and the clang of metal.
No text, subtitles, logos or watermarks.
Length 8 s. The hero plays the music, everyone in the scene hears it; there is no background score.
Realistic live-action cinematic look, warm streetlights at night, shallow depth of field, film grain.
Scene: a narrow old-town street at night, wet cobblestones. An adult man, about 50, with a grey beard and a black hat sits on a stool and plays an acoustic guitar. Passers-by walk past, blurred in the background.
Camera: medium shot, a slow arc shot around the musician.
Timeline:
[0s-5s] The musician plays a fast flamenco melody, his fingers moving quickly over the strings.
[5s-8s] He slaps the guitar body twice in rhythm and smiles.
Audio: the loud acoustic guitar of the street musician, footsteps on wet stones, distant voices, no background score.
No text, subtitles, logos or watermarks.
Music by instruments, tempo and volume; the sounds of the place after it.
Realistic live-action cinematic look, sunrise light, warm haze, film grain.
Scene: an empty desert road at sunrise, a red vintage car drives toward the camera, dust rising behind it.
Camera: wide shot, a slow pull out as the car approaches.
Timeline:
[0s-3s] The red car drives along the road toward the camera.
[3s-5s] The car passes close by the camera and leaves a cloud of dust.
Audio: a slow synthesizer pad joined by a steady electronic drum beat that gradually gets louder. The car engine hums, gravel crunches under the tires, the wind whistles.
No text, subtitles, logos or watermarks.
The start frame is the last frame of the previous take («Continue from the last frame» under the take).
Continue seamlessly from <Picture 1>.
Realistic live-action cinematic look, golden hour sunlight, shallow depth of field, film grain.
Scene: the same wooden pier at sunset, the same woman in the white linen shirt.
Camera: medium shot, static shot.
Timeline:
[0s-3s] She sits down on the edge of the pier.
[3s-5s] She dips her bare feet into the water.
Audio: gentle acoustic guitar at a slow tempo. Soft splashing, distant birds.
Keep her face, hair and outfit exactly the same. No text, subtitles, logos or watermarks.
A landscape without people: fog, sunbeams, an animal in the distance.
Cinematic nature documentary look, soft morning light, high detail.
Scene: an empty pine forest at dawn, thick fog between the trees, sunbeams breaking through the branches.
Camera: wide shot, a slow tracking shot forward between the trunks.
Timeline:
[0s-3s] The sunbeams slowly brighten through the fog.
[3s-5s] A deer steps out from behind a tree in the distance.
Audio: no music, only birdsong, a light breeze in the pines and a twig cracking.
No text, subtitles, logos or watermarks.
Seven more presets — only in the Lab's «Preset ▾» menu:
A prompt skeleton: every part with «...» to fill in.
8 s. A woman narrator speaks, the heroine's lips stay closed (the «Speech and dialogue» section).
The camera backs away in front of the hero; night, rain, neon.
A stage, light, haze, the camera circling, a beat.
An ad: studio light, the camera circling, a drop.
Animation instead of live action.
The 1970s: faded colors, grain, flicker.
Built from the app's help · Собрано из справки приложения — Remiqora Video 0.1.0 (f96dc8e). The help lives in the app; this wiki is rebuilt with every release.
English
Getting started
Scenarios
- First run
- A video from an idea
- Fix one scene
- Your own hero: photo, look, voice
- Bring a photo to life
- A video without people
- A clip from a song
- Own prompts: a storyboard by hand
- One scene in the Lab
- A graph from a template
- Montage and export to DaVinci Resolve
- Final: 1080p, 4K, HDR, 48 fps
- Reshoot a finished video
- Your own LoRA and another model
- Disk space
Prompts
- How to write a prompt
- The parts of a prompt, in order
- Speech and dialogue
- Sounds and music
- Several characters
- Camera and light: the words
- Common mistakes
- Presets
Idea
Storyboard
Shooting and video
Graph
Tools
Русский
Начало
Сценарии
- Первый запуск
- Ролик из идеи
- Исправить одну сцену
- Свой герой: фото, внешность, голос
- Оживить фото
- Ролик без людей
- Клип из песни
- Свои промты: раскадровка вручную
- Одна сцена в лаборатории
- Граф из шаблона
- Монтаж и выгрузка в DaVinci Resolve
- Финал: 1080p, 4K, HDR, 48 к/с
- Переснять готовое видео
- Своя LoRA и другая модель
- Место на диске
Промты
- Как писать промт
- Части промта по порядку
- Речь и диалоги
- Звуки и музыка
- Несколько персонажей
- Камера и свет: слова
- Частые ошибки
- Заготовки
Идея
Раскадровка
Съёмка и ролик
Граф
Инструменты