Minimax H3

提示词*

Generate a 15-second, 16:9 horizontal, original specimen-collection-themed game promotional PV first half. The overall style combines pure 2D Japanese cel animation with abstract Editorial MG graphic design and game UI graphic design, referencing high-end game promotional posters and dynamic flat animation language. The entire video must remain purely 2D. No 3D effects, no 立体 modeling, no real spatial depth, no 3D perspective rendering, and no CG texture are allowed. All characters, scenes, props, and effects must appear flat like 2D animation illustrations. Every shot should feel like a dynamic flat poster that continuously disassembles and recomposes.

## Theme

The story is set in “Specimen Bureau,” an abstract collection institution composed of specimen jars, classification labels, ruler scales, and coordinate grids. Original characters, original world, and original identifiers only. No existing game characters, logos, or assets may appear.

## Art Style

High-contrast three-color visual system: specimen green, bone white, and pure black create a strong impact. Grayscale is only used for hair and halftone dots. Specimen green represents immersion and preservation. Bone white represents classification and naming. Pure black represents display cases and gaze.

Characters use minimal hard-edge cel shading and large solid-color silhouettes, with only one or two layers of shadow and no realistic volumetric lighting.

The institutional space is highly flattened, retaining only specimen jar rectangles, classification labels, skeleton line art, ruler scales, coordinate-grid floors, glass rings, and display-case outlines. The visual language is built from bone-white line art, solid-color rectangles, and grids.

It fuses abstract Editorial MG, game UI, ruler scales, rings, butterfly silhouettes, scan lines, halftone dots, and geometric slices. The space may use intentional distortion, inversion, compression, and alternating perspective deformation.

Small amounts of blood red appear only when new specimens or death imagery appear, usually lasting two or three frames.

## Engineering Drawing Overlay — present in every frame

A fine gray engineering drawing layer stays on top of the entire image: thin grid lines, crosshair markers, four-corner framing brackets, small ruler tick marks at the image edges, circular dials, protractor arcs, and thin reticle frames. This layer always remains present, shifts slightly with parallax during camera movement, and the dial needles and arcs keep sweeping.

## Characters

The female protagonist is a young woman with black shoulder-length hair, a slender figure, bone-white sclera, and specimen-green pupils. She wears a black turtleneck, black boots, and a bone-white robe, and carries a slender specimen needle nearly as tall as her body. Her gaze is cold and her expression restrained. Her design is flat and minimal, composed mainly of a black body silhouette, bone-white negative shapes, and specimen-green pupils.

The curator is a male character with gray-white long hair, wearing a bone-white long coat, black apron, round goggles whose lenses reflect two bone-white dots, and black gloves. He has an elegant yet dangerous presence, a calm expression, and a faint mocking quality.

## Shot Breakdown

0–2.5 seconds
On a bone-white background, only the female protagonist’s black boots and the tip of her slender specimen needle appear, showing only her lower body. The camera moves quickly forward from an extremely low angle. As her foot lands, bone-white horizontal lines, coordinate grids, and ruler scales explode outward from the ground in a circular pattern. The needle tip scrapes across the frame, leaving a specimen-green geometric trail. End state: the needle tip touches the ground, and the specimen-green trail stretches across the screen.

2.5–5 seconds
The image is split by hard edges into three panels: boot close-up, half-body, and full-body silhouette, simultaneously showing the same female protagonist. The panel boundaries keep moving horizontally. She instantly inverts from black silhouette to bone-white silhouette and back again, three times in a row. End state: the three panel boundaries merge into a single symmetrical wide shot.

5–7.5 seconds
Symmetrical wide shot, camera pulls back rapidly. The female protagonist stands at the center of a coordinate-grid floor. Around her, a massive array of specimen jar rectangles rises. Each jar contains a huge skeleton-line silhouette at its center, while the rest of the jar remains blank. A giant specimen-green ring expands outward from behind her, and a black diagonal shape cuts through her body, dividing the frame into irregular sections. End state: the ring expands to full screen, and the diagonal plane splits the image into four areas.

7.5–10 seconds
The female protagonist’s cold eye appears in an independent horizontal close-up frame: bone-white sclera, specimen-green circular pupil, pure black hard-edge eyeliner, and hard-edge triangular eyelashes. No skin is drawn around the eye, no gradients, no highlights; the area is left blank. Four-corner framing brackets quickly close in. The pupil briefly flashes blood red and then returns to specimen green. End state: a bone-white specimen needle mark appears at the center of the pupil.

10–12.5 seconds
After a single-frame white flash, the shot cuts to walking legs on the coordinate grid. With each step, concentric ripples expand outward. Giant classification labels fly in from outside the frame and sweep past, while needle-shaped black shapes move horizontally across the screen. The entire image tilts and rotates. End state: the ripples fade away, and the labels come to rest in a row.

12.5–15 seconds
The female protagonist is duplicated horizontally into three, then multiplies into eight synchronized walking silhouettes. Behind her, a whole row of specimen jar rings lights up one by one. The green-white inversion frequency increases to six to eight times per second, while bone-white particles scatter heavily. End state: a single-frame white flash explodes, the entire image bursts into pure bone white, the eight silhouettes converge back into one, and the female protagonist finishes the motion holding the specimen needle still.

## High-Dynamic Hard Constraints — higher priority than individual shot descriptions

No static shots are allowed. Within each shot, at least two graphical motions must occur simultaneously, selected from:

engineering grid parallax shift / circular dial needle sweeping / panel boundaries moving horizontally / diagonal shapes sweeping across the frame / ring expanding from the center / bone-white particles scattering / silhouette black-white inversion / halftone dot density changing / ruler scales lighting up frame by frame.

Camera movement must be large: fast push-ins, horizontal whip pans, full-image rotation, and alternating tilt. Avoid uniform slow panning.

Each drum hit triggers one visible graphical event: single-frame white flash, green-white inversion, geometric disassembly, or brief glitch.

For this first half, the ending must land explicitly on the final state “image bursts into pure bone white.” Do not fade out.

## Motion Requirements

All transitions follow drum beats. Use diagonal wipes, ring expansion, silhouette inversion, graphic disassembly, single-frame white flashes, specimen-green afterimages, and brief glitches. Transitions are completed through character silhouettes, needle-shaped blocks, diagonal masks, black-white inversion, and graphic matching.

Character actions are expressed through key poses and freeze-frame imagery, not continuous realistic motion.

Avoid ordinary fades, meaningless fast cuts, realistic 3D, warm sunlight, excessive neon cyberpunk, and cluttered text.

## Strict Restrictions

The entire video must maintain a 2D abstract MG and three-color game UI aesthetic. No realistic lighting, no 3D environments, no cinematic realistic perspective, no ordinary cel-shaded battle animation style, no complex continuous fighting, no large neon areas, no cyberpunk realistic cityscapes, no skin tones or skin texture, no gradients, and no highlights.

Characters, architecture, props, and effects must all be flattened, silhouetted, and graphic.

No text, letters, numbers, subtitles, identifiers, or watermarks may appear in the image.

Except for the two or three frames of blood red used for new specimen imagery, no fourth color is allowed beyond specimen green, bone white, pure black, and hair grayscales.

参考图片

拖拽到此处，或点击上传

0/6

画面比例*

模型*

分辨率*

时长*

高级

Create a Hip-Hop Music Video

MiniMax H3 is an advanced multimodal AI video generation model that transforms text, images, and references into high-quality videos. With strong prompt understanding, consistent visual control, and dynamic motion generation, it enables creators to produce animations, visual effects, brand content, and creative videos efficiently.

发布于 2026年8月10日 · 更新于 2026年8月12日

高级

AI Video GeneratorText to VideoImage to VideoMultimodal AI

输出规格

画面比例16:9 / 1:1 / 9:16

案例展示

使用该模板创作，点击查看

查看全部

使用方法

For the best result, upload a clear image or keep the text prompt simple and descriptive.

1Upload your reference images
2Describe your album concept and visual style
3Click "Generate" to create your album cover

提示词*

参考图片

画面比例*

模型*

分辨率*

时长*

输出规格

案例展示

使用方法

相关模板

3D 动漫风格

剪纸定格动画

赛博朋克动漫视频