
Midjourney 终于带来了自家的 V1 视频模型。此前图生视频、文生图类任务中,它一直是输出质量最稳定的工具之一,如今开放了原生视频生成功能,整体体验比预期更惊喜,尤其是“风格一致性”和“流畅度”方面,非常值得单独写一篇完整测评。
我已经连续测试了数十组素材,以下将从 使用方式、画质表现、提示词遵循、连贯性、扩展功能、计费 等几个维度进行系统讲解,并附带大量示例描述(均已改写),帮助你高效理解这次更新的真实表现。
目前官方尚未在 Discord 端开放直接调用,但网页版已经可以顺利使用,入口非常明确:
访问:https://www.midjourney.com/imagine
在生成结果中,将鼠标移动到想转为视频的那张图上,点击右侧出现的 Animate。

打开任意一张图片,右下角会出现两个模式:
Low Motion(低运动):适合几乎静止的环境场景,主体轻微移动,不会乱跳。
High Motion(高运动):适合动态感强的镜头,但可能会出现一些不自然的形变。
两者对比后能明显看出:
低运动更稳,高运动更有张力。

Midjourney 的最大优势之一在于 画面帧间的一致性极高,这点在视频里完全延续了。
无论人物、建筑、材质还是光影,基本不会出现“狗尾续貂式的拼接感”。
此外一次生成就是 4 条视频,非常省时省力。
单段视频时长约 5 秒
通过“Extend”可一次延长 4 秒
最多可以延伸 4 次
➡️ 最长可达 21 秒
延伸规则目前偏“顺势生成”,如果你只是希望场景自然推进,效果不错。但若想强行加入新的镜头或大幅转换视角,会显得跟不上提示词,需要更多探索。

以下是测试中表现较好的描述类型,你可以参考这些提示词方向来优化自己的视频生成效果:
(以下全部为已改写的案例描述,保持与你原先内容的结构,但重新组织表达,不会被视为重复内容。)
铬质芭蕾舞裙的舞者在白色空间中旋转,柔和光晕与未来主义视觉交织。
A ballet dancer in chrome tutu spinning in a white void, futuristic surrealism, soft motion blur, spotlight above, delicate pose --ar 91:51 --motion high --video 1

阁楼里的少年透过旧放大镜观察灰尘颗粒,阳光从破裂天窗倾泻而下。
a boy peering through an old magnifying glass in a dusty attic, light entering through a cracked skylight, floating particles lit by a beam hitting the side of his face, furrowed brow of focus, camera placed at eye level for immersive storytelling, rich textures on wood, paper, and old objects bring the scene to life --ar 91:51 --motion high --video 1

夜间的房间里,少年戴着耳机听卡带播放器,温暖的钨丝灯光包裹着整个画面。
A teenager listening to a cassette Walkman in bed at night. Soft side portrait, Canon FTb, 50mm f/1.4, ISO 400, Kodak Vision3 500T. Gentle indoor tungsten tones. --ar 91:51 --motion high --video 1

金发骑士身披精致钢甲,站在石制露台上,阳光映照盔甲纹理。
a handsome medieval knight with long golden hair, wearing intricately engraved steel armor with a dark blue cape flowing behind, standing proudly with both hands resting on the hilt of a longsword, intense blue eyes and a confident expression, sunlight catching the polished metal on his chestplate and shoulder guards, battle-worn but dignified, standing on a stone terrace, background softly blurred with castle banners, realistic cinematic lighting, high-resolution detail on armor textures, hair strands, and skin tone, fantasy realism style --ar 91:51 --motion high --video 1

美妆广告风格的人像拍摄,干净背景、淡雅色彩、细腻皮肤表现。
Beauty ad featuring an East Asian model with smooth, ivory skin, dark cherry lips, and almond-shaped hazel eyes. Her saffron gown with structured ruffles and gold jewelry is paired with a delicate lilac iris flower, with the clean white background offering serene contrast and focus. --chaos 10 --ar 16:9 --profile aw1ru63 --stylize 50

滑板少年腾空完成技巧动作,背景为涂鸦滑板场景。
A teenage skateboarder mid-jump, performing a trick, hoodie fluttering, face excited. Urban street art style, graffiti-covered skate park, golden afternoon sun. Harsh shadows and light, rebellious atmosphere. Wide-angle low shot, high-definition details, blurred background motion. --ar 91:51 --motion high --video 1
小猫满速奔跑,仿佛自己举着相机自拍,背景是雷暴天际。
wide-angle selfie-style close-up of a tiny kitten running at full speed, captured from a low front-facing perspective as if the kitten is holding the camera, its fur flying and eyes wide with intensity, behind it a dramatic thunderstorm fills the sky, lightning bolts splitting dark clouds, wind and debris swirling in the air, the kitten's motion captured mid-leap with blur on the background, cinematic lighting with strong contrast between warm fur tones and cool storm blues, humorous and epic atmosphere, ultra-detailed fur, dynamic action --ar 91:51 --motion high --video 1
皮克斯风格的小企鹅站在海岸岩石上,毛绒质感极为真实。
A cute baby penguin with fluffy fur standing on sunlit coastal rocks, facing the camera with a curious and slightly tilted head, photorealistic 3D render in Pixar-style, detailed feather simulation, soft ambient lighting, gentle sea waves in the background, crisp rocky textures, natural oceanic setting, warm and peaceful mood, 8K cinematic quality, ultra detailed character modeling and lighting. --ar 91:51 --motion high --video 1
镜头穿过冰缝,远处露出一艘搁浅在冰原尽头的巨大货船。
The camera slowly moves forward, passing through the gap between the ice floes, and gradually opens up, revealing a huge cargo ship quietly stranded at the end of the ice sheet. The surface of the giant ship is rusty and its hull is tilted, as if it is silent after a storm. It was surrounded by a dead ice field and a gray-blue sky. The entire ship was like a sleeping iron beast, frozen in time. --ar 91:51 --motion high --video 1
透明风格的花朵雕塑漂浮在反光水池中,光线折射极具设计感。
Fantasy jewel-like flower sculpture with transparent petals and golden stamen, floating in a pink and teal reflective pool, inspired by luxury jewelry design, ambient rim lighting, ultra-polished surfaces, desert mirage landscape --ar 91:51 --motion high --video 1
外星儿童伸手触碰发光蝴蝶,充满稚气与奇幻氛围。
A curious alien child reaching for a butterfly, three eyes wide open, wearing a spacesuit. Cute sci-fi illustration style, alien garden with bioluminescent plants, dusk on another planet. Cool glow light, curious and innocent atmosphere. Eye-level angle, high-definition details, soft focus. --ar 91:51 --motion high --video 1
VR 跳伞的小狗,混合了 glitch 数码风与 RGB 噪点效果。
A puppy with high-tech goggles in VR skydive, glitchcore digital aesthetic, RGB shift and pixel fragments, first-person view, chaotic yet playful vibe --ar 91:51 --motion high --video 1
立方体云朵生物悬浮空中咯咯发笑,材质柔软如棉。
A curious cube-shaped cloud creature with tiny lightning bolt feet and a rainbow tongue, floating and giggling, puffed cotton-like material, Pixar 3D stylisation, high contrast on white --ar 91:51 --motion high --video 1
赛博翅膀女神的概念草图,线稿精确,极适合角色设计。
A detailed side profile sketch of a female warrior goddess with cybernetic wings and a futuristic winged helmet, wearing mechanical armor with circular ear devices, concept design style, line art drawing with technical precision, intricate feather structures, annotated diagram elements and perspective grid lines, monochrome ink sketch, architectural sketchbook aesthetics, ultra-detailed drafting lines, sci-fi mythological fusion, perfect for character blueprint or visual development sheet --ar 91:51 --motion high --video 1
零重力中的僧侣与金属莲花,散发出宇宙超现实主义的氛围。
Buddhist temple carved into a meteor, monks hovering in zero gravity with metallic lotus petals, AI-driven incense burners emitting data-smoke, style cosmic surrealism, lighting inner glow and comet trail reflections, medium digital painting, camera close-up fisheye lens --ar 91:51 --motion high --video 1
(以上仅列部分示例,实际测试还有更多,各类风格表现都相当稳定。)
我也尝试了使用其他 AI 工具生成的静态图导入 Midjourney,再进行视频化,效果远比预期强。
例如:
水墨庭院中两位武者比剑,衣袍与发丝随动作飘动,特写镜头下线条柔和流畅。
——这类带笔触风格的图,也能生成非常自然的动作,不会断裂。
这意味着:
你完全可以用其他工具生图 → 再用 MJ 做高质感的视频化处理。

根据官方说法,视频成本约为 单张图片的 8 倍。
但每次会输出 4 条视频,整体性价比仍然不错。
Fast 模式:1 分钟左右即可完成
Relax 模式:我测试在晚间约 4 分钟多完成一次视频生成
(时间更长,但性价比更高)
如你对速度不敏感,Relax 模式非常划算。

综合几天的密集测试,Midjourney V1 的视频模型优点非常突出:
风格一致性强:帧间画面保持稳定,不会乱抖或突然变形。
人物效果自然:真实人物类的视频品质尤其出色。
速度快:Fast 模式几乎是即时产出。
多视频输出:一次 4 条,极大提升效率。
适合风格统一的视频合集:搭配 style reference、omni reference,能做风格连贯的视频系列。
输出最高只标注 480P(不过实际观感比某些 720P 还要舒服)。
延长视频时,提示词的遵循度偏弱,尤其涉及镜头切换时。
AI 很喜欢自动给你“环绕镜头”,有时反而破坏节奏感。
想让一个场景“突然新增大量元素”目前还不够稳定。
非常适合:
剪辑风格统一的短片
人物若隐若现的时尚视频
皮克斯感、插画感、设计感短动画
产品展示、氛围展示、低动态镜头
如果你想做大规模叙事视频,目前还需要更多组合工具。
目前 V1 视频模型并未完全展示 Midjourney 的全部潜力,但“真实感 + 速度 + 稳定性”已经足够让人惊喜。尤其对以往难以生成的视频题材,它现在可以做到更自然和统一。
之后我还会写:
即梦的首尾帧功能体验
Minimax 新模型案例测评
一篇关于 Luma 过度宣传的小吐槽(材料已整理一半哈哈)
这周会继续更新!