Vidrush
Toggle Sidebar
AI 视频生成器
Language
领取积分
AI 视频生成器
加载中...
导航
我的创作
Vidrush 工具
Agent
AI 视频
Faceless Channel
AI 音乐
爆款预设
提示词库
图片生成器
AI 配音
AI 广告生成器
Shorts Studio
模板
价格
视频编辑
视频加音频
视频超分
视频增强
人声背景音分离
视频转 Prompt
AI 智能切片
长视频转短视频
AI 游戏高光剪辑
AI B-roll 生成器
YouTube 缩略图生成器
视频 OCR
视频摘要生成器
剧本还原
AI 漫剧转绘
长视频理解
产品视频生成器
Sora 2 视频生成器
编辑图片
图片去背景
图片放大
图片扩图
背景虚化
AI 智能扩图
黑底背景
白底背景
图片去物体
图片增强
图片去模糊
图片锐化
图片转 Prompt
AI 发型切换器
AI 发色切换器
产品图片生成器
生成器
AI 女生生成器
AI 人脸生成器
AI 表情生成器
AI 角色生成器
AI 涂鸦生成器
AI 暗黑美学生成器
音频工具
音频翻译
AI 旁白配音
文本转语音
女声生成器
AI ASMR 声音生成器
AI 有声书
播客生成器
AI 音乐生成器
AI 音效生成器
口音生成器
英式口音生成器
澳洲口音生成器
字幕工具
视频转文本
MP4 转 SRT
视频转 SRT
字幕翻译器
SRT 翻译器
MP3 转 SRT
字幕转语音
AI 写作
AI 语气改写器
中文
Eye Check
模型
Gemini Omni
10s
50 credits
提示词
模板
Ultra-photorealistic cinematic close-up portrait of a man, shoulders-up, looking straight into the camera as if it were a mirror. He gently pulls down his lower eyelid with one finger, examining his eye as if something is in it. Expression: focused, intent, natural, slightly puzzled, with subtle facial tension but no exaggeration. The pose must feel candid and believable, like someone checking their eye in a mirror. Visual style: extremely realistic, premium live-action still, ARRI / full-frame camera feel, crisp but natural detail, subtle film grain, realistic skin tones, no beauty retouching, no artificial glamour, no illustration or CGI look. Straight-on eye-level camera, centered composition, shallow depth of field with sharp focus on the eye and face. Background: realistic apartment interior, softly out of focus, unobtrusive, with believable depth and natural ambient detail. Soft, natural window light with subtle falloff and gentle shadows. Create one continuous 15-second forward zoom loop that never reverses, cuts, dissolves, fades, or pauses. The camera moves forward with smooth logarithmic acceleration, starting from the man’s eye and traveling through increasingly smaller scales (biological, cellular, molecular, atomic, subatomic), then seamlessly into deep space, then back toward Earth, through a building window, into the same room, ending on the exact opening composition so the video can loop perfectly. Camera path and scale progression (approximate timing): - 0.00–0.65 s: Exact opening portrait. Silent, mouth still, no blink. Finger already under the visible eye. He pulls the lower eyelid down only 2–3 mm more, leans forward a few millimeters, and makes one tiny micro-saccade. Around 0.45 s, begin a subtle forward creep centered on the pupil. - 0.65–1.30 s: Macro zoom into the eye. Resolve pores, lashes, scleral vessels, iris fibers/crypts/furrows, wet curved cornea, and a moving window reflection. Pupil remains a real optical aperture, never a tunnel or portal. Focus moves from lashes → cornea → iris/pupil. - 1.30–3.05 s: Pass through thin tear film, corneal epithelium, Bowman’s layer, ordered stromal collagen, keratocytes, Descemet’s membrane, and endothelium. Real refraction, translucent anatomy. Enter clear aqueous humor and pass through the pupil toward the lens. - 3.05–3.75 s: Move through transparent lens capsule and ordered lens fibers/crystallin proteins with smooth refraction, then nearly transparent vitreous with sparse collagen/hyaluronan. Retina appears ahead. - 3.75–5.75 s: Pass the inverted retina through neural layers to photoreceptors, pigment epithelium, and choroid. Real proportions. Lock onto one central cone, enter organized outer-segment discs with subtle phototransduction, through the cilium into the mitochondria-rich inner segment, then toward the nucleus. - 5.75–7.35 s: Crowded cytoplasm with proteins, ions, cytoskeleton, ribosomes, and molecular motors. Pass through a nuclear pore into 3D chromatin. Resolve nucleosomes/linker DNA, then one right-handed DNA helix with bases inside and sugar-phosphate backbones outside. Approach one carbon atom; bonds appear as shared electron density. - 7.35–8.65 s: Quantum carbon: diffuse electron probability cloud/orbitals, compact proton-neutron nucleus, brief quark/gluon field density. No planetary electrons or literal universe inside the atom. Seamless scale-match: field nodes retain screen positions while scale expansion turns them into dark space, cosmic-web filaments, and galaxy concentrations. No visible transition boundary. - 8.65–10.10 s: Real dark cosmic web, sparse galaxies, clusters, and voids. Accelerate into a barred spiral galaxy through dust lanes toward a Sun-like star. Solar System resolves naturally; planets stay widely spaced. Earth is the target. - 10.10–11.65 s: Realistic Earth with thin atmosphere, oceans, land, and volumetric clouds. Enter atmosphere without fireball: black → deep blue → pale blue, through a cloud gap. Continue surface → region → city → neighborhood, each scale emerging from real detail. Target a plausible ordinary building. - 11.65–13.35 s: Approach a window consistent with the bright frame-left window in the opening shot. Pass through real glass; exposure adapts into the darker room. Reconstruct the exact room; the man is already in the correct pose. Continue forward and decelerate into the original camera position, never pulling backward. - 13.35–15.00 s: Progressively reconstruct the exact source composition. During 14.40–15.00 s he gently releases the extra 2–3 mm eyelid tension while finger and gaze return imperceptibly to the exact opening state. At 15.00 s the frame is visually indistinguishable from 0.00 s: identical camera position, perspective, focal length, focus, head/finger/eyelid/gaze, expression, room geometry, window reflection, exposure, white balance, and depth of field. Identity and environment lock: preserve exactly from @Image1: identity, face, skin/stubble, hair, blue-green iris, white T-shirt, finger, room geometry, frame-left window, daylight, dark background object, perspective, lighting, and framing. No identity, wardrobe, environment, or eye-color drift. Look and audio: extreme photorealism; live-action opening/ending; high-end scientific microscopy; grounded quantum visualization; plausible dark space. Real refraction, scattering, transparency, parallax, exposure, and scale-dependent depth of field. No dialogue, lip movement, music, subtitles, labels, logos, or text. Strict locks: no sudden morphs, pupil portal, flesh tunnel, gore, neon biology, textbook graphics, classical atom models, fantasy space, duplicate anatomy, lighting discontinuity, or loop mismatch. Reference image: use @Image1 as the exact opening frame, exact final frame, and hero reference for the man’s identity, eye color, skin, stubble, hair, clothing, room layout, window position, lighting direction, and camera angle.audible CTA.
5957 / 50000
参考图片(可选)
参考图片
可选:最多上传 2 张参考图来引导生成(JPEG、PNG、WebP,最大 10MB)
画面比例
16:9
9:16
时长
10s
分辨率
720P
1080P
生成音频
2 倍积分
Basic+
所需积分
50
加载中...
视频预览
中文