Floating City and the Golden Portal

MiniMax H3Sci-FiFantasyImage to VideoEpicPost-ApocalypticSpace10s
Pushing in from a wide shot of a floating city toward the cosmic swirl of a golden portal, a 10-second image-to-video clip. A single push-in shot plus one energy-burst moment is the most reliable structure for image-to-video.
PROMPT · 视频提示词
Create with AI
MiniMax H3 10 Image-to-Video Prompt @Image 1 as the first-frame reference image. Generate a 10-second, ultra-high-quality, cinematic sci-fi fantasy video. The camera begins on a wide shot of a cluster of floating cities, slowly pushing forward through a vast sea of clouds and between floating islands. The camera gradually approaches a massive golden portal at the center, and the cosmic vortex inside the portal begins spinning at high speed, releasing radiant energy that intertwines blue-violet and gold. As the camera draws closer, nebulae drift slowly across the sky, distant planets rotate slowly, the lights of the floating cities switch on one by one, waterfalls pour down from the edges of the floating landmasses, and the clouds are gently pushed apart by the pulses of energy. On the central platform, a lone figure's cloak billows in the wind as they face the opening cosmic gate, tiny glowing particles swirling around them. In the final 2 seconds, the camera passes through the golden portal into a dazzling cosmic space, revealing an infinite galaxy and the ruins of a future civilization. Visual style: Epic sci-fi fantasy film, evoking the feel of a Hollywood blockbuster, 8K ultra-high-definition, realistic volumetric lighting, HDR, high dynamic range, complex particle effects, cinematic color grading, IMAX cinematography style. Camera language: Slow dolly in, subtle orbiting movement, strong sense of spatial depth, clearly layered foreground/midground/background, authentic camera motion. Mood: Grand, mysterious, awe-inspiring, futuristic, an exploration of an unknown civilization. Aspect ratio: 9:16 vertical cinematic composition. Avoid: Low-quality textures, distorted figures, collapsing architecture, random objects, flickering, camera shake, excessive haze. Reference image generated in GPT — see the comments section for details.
✍️ Editor’s Notes

这条开篇直接点名「Image to Video」并锁定@图片1为首帧参考,末尾又补一句「参考图在GPT生成详见评论区」,等于交代了完整的生产链路:先用GPT生成一张概念图作为起始帧,再交给MiniMax H3做图生视频的动态延展,这种分工模式下文字提示词的任务变成了描述「从这张静态图出发要怎么运动」而不是从零构建场景。镜头运动只用了一句「慢速推进镜头,轻微环绕运动」,剩下篇幅全部让位给「星云流动、行星旋转、瀑布倾泻、传送门能量漩涡」这类逐层展开的环境细节,这种写法是图生视频类提示词的典型分工——起始构图已经被参考图锁定,文字只需要负责让画面里的元素依次动起来、层层揭示纵深感。结尾避免清单简短却精准,专门点出「建筑崩坏」和「随机物体」,说明悬浮城市这种复杂场景最容易在生成过程中出现结构性穿模,这是针对该题材的常见翻车点做的提前预防。

Create with AI
Related Works
×