跳到主要内容
Kyeo AI
AI 视频模型
ByteDance
Seedance

Seedance 1.5 Pro

用文字描述镜头，或加入图片测试人物、构图与运动。这里可以调整输出规格，并选择是否请求音频或固定镜头；先做小样，再根据真实结果判断是否适合当前内容。

使用这个模型查看积分价格返回模型广场
输入方式
文字，或提示词加 1–2 张图片
预计预扣积分
7–180 积分／次
提示词要求
至少 3 字符，原文最多 2,500 字符
图片数量
0–2 张图片
快速了解
  • 可从文字开始，也可加入最多 2 张图片；图片需通过格式、容量和元数据核验，上传成功后生成前仍会复验。
  • 输出可选 3 种分辨率：480p、720p、1080p；时长为 4–12 秒的整数，不能填写小数或把它理解为少数固定档位。
  • 关闭音频请求时，合法组合预计预扣 7–90 积分；开启后须按相应费率重新计算。整段金额向上取整，因此切换音频后的预计积分不一定是原金额的两倍。
  • 共有 6 项参数，文字任务有 6 种固定宽高比可选。加入图片后会隐藏画幅选项，本次图片请求不携带该比例；固定镜头请求也不保证人物、背景或镜头完全不动。

相关查询: 如何开始 Seedance 1.5 Pro 的视频任务 · Seedance 预计预扣积分怎么计算 · 图片的格式和大小需要满足什么要求

示例

Cinematic scene

{ "scene_id": 4, "title": "Starlight Shadow / 星屑のシルエット", "duration": "10s", "visual_style": "Fantasy / High-End Commercial", "prompt": "A person walks by a clean white wall in a bright, sunlit gallery. Their shadow follows them, but then gracefully stops and starts performing a rhythmic, elegant dance. The shadow is made of soft, shimmering dark-blue particles. As the person reaches out to high-five the wall, the shadow hand gently emerges into the 3D world as a translucent, glowing crystalline form, meeting the person's hand with a soft trail of light.", "vfx_technical_details": { "camera_work": "Smooth, cinematic tracking shot with a focus on the interaction point.", "shadow_FX": "Particle-based shadow simulation with soft edges and internal luminescence.", "transition_FX": "Seamless 2D-to-3D transition using a translucent glass shader and light trail effects.", "lighting": "Bright, warm afternoon sunlight to contrast with the cool-toned magical shadow." }, "key_elements": [ "Independent dancing silhouette", "Shimmering particle effects", "Translucent crystalline 3D hand", "Magical and inspiring atmosphere" ] }

Cinematic scene

CINEMATIC MORNING ROUTINE STYLE: Ultra realistic cinematic video, soft morning sunlight, cozy luxury bedroom, warm beige and cream tones, natural skin texture, dreamy atmosphere, shallow depth of field, premium commercial quality, realistic motion, elegant feminine aesthetic, 4K HDR, cinematic color grading, smooth camera movement. ENVIRONMENT: Modern minimalist bedroom with large bed @E9OLrDQ5pCd8JWLc1V6U , white fluffy blankets, soft pillows, wooden furniture, sheer curtains, indoor plants, warm golden sunlight entering through large windows, clean luxury interior, peaceful morning atmosphere. CHARACTER: Beautiful young woman @5VLsmdxxrtZIHeHOpUlw , long slightly messy morning hair, natural makeup, oversized cream sleep shirt, soft feminine appearance, sleepy but playful expression, charming smile, natural body language, cute and affectionate mood. CAMERA: 24fps, anamorphic lens, shallow DOF, cinematic lighting, smooth dolly movement, slow push-in, close-up facial shots, medium shots, subtle handheld realism, smooth orbit camera, professional cinematography. STORYBOARD BREAKDOWN: SCENE 1 (0-3s) — SLEEPING BEAUTY Wide shot. The woman sleeps peacefully beneath a soft white blanket. Golden morning sunlight slowly enters through the curtains and illuminates her face. The blanket gently rises and falls with her breathing. Dust particles float softly in the sunlight. Camera: Slow push-in toward the bed. Emotion: Warm, peaceful, intimate. SCENE 2 (3-6s) — WAKING UP Medium close-up. She slowly opens her eyes, blinking softly. A sleepy smile appears on her face. She stretches both arms above her head while still lying in bed and gently hugs the pillow. Camera: Slow cinematic push-in. Emotion: Sleepy, cute, affectionate. SCENE 3 (6-9s) — MANJA MOMENT Close-up. She rolls onto her side, pulls the blanket closer to her body, nestles into the pillow, and gazes softly toward the camera. Her lips form a playful smile while strands of hair fall naturally across her face. Camera: Slow orbit around her face. Emotion: Playful, sweet, feminine, spoiled. SCENE 4 (9-12s) — MORNING STRETCH Medium shot. She slowly sits up on the bed. Sunlight highlights her silhouette and hair. She brushes her hair back with her fingers and performs a gentle morning stretch while enjoying the warmth of the sunlight. Camera: Side dolly movement. Emotion: Fresh, elegant, relaxed. SCENE 5 (12-15s) — SWEET ENDING Close-up transitioning to medium shot. She looks toward the bright window, enjoying the morning atmosphere. Then she slowly turns back toward the camera and gives a warm, sweet smile. Sunlight fills the entire room as the scene ends. Camera: Slow pull-back. Emotion: Happy, charming, inviting. NEGATIVE PROMPT: low quality, blurry, bad anatomy, deformed face, overexposed, underexposed, cartoon, CGI look, unnatural movement, flickering, duplicate limbs, distorted hands, watermark, text, logo, oversaturated colors, low resolution, noise, artifacts. FINAL OUTPUT: Ultra realistic cinematic 15-second video, natural movement, luxury morning atmosphere, emotional storytelling, soft sunlight, premium commercial production quality, beautiful feminine expressions, smooth transitions, highly detailed 4K HDR masterpiece.

Cinematic scene

Create a highly addictive, nostalgic early-2000s Korean MiniDV home-video story designed to hook viewers within the first second and make them stay until the final reveal. The footage should look like a real forgotten MiniDV tape discovered 20 years later — authentic 4:3 consumer camcorder, imperfect handheld movement, CCD noise, soft analog sharpness, tape grain, slight color bleeding, autofocus hunting, exposure fluctuations and occasional blown highlights. Color palette: warm golden bedroom light, dusty pink highlights, soft cream skin tones, muted teal shadows and subtle faded red accents. Cozy, nostalgic, slightly dreamy but completely realistic. Main character: a 24-year-old beautiful Korean woman with long dark hair, expressive eyes and minimal natural makeup. She wears an oversized pale-blue T-shirt, loose cream pajama pants and cozy socks. Keep her face, hairstyle, outfit, bedroom, props and lighting perfectly consistent throughout. SHOT 1 — 0:00–0:03 — THE HOOK The camera is already recording from a slightly crooked position on a bedroom desk. The woman suddenly rushes into frame holding a small mysterious red gift box. She looks around the room as if checking whether anyone is watching. She whispers excitedly: “I finally found it…” She places the box on the bed. A tiny handwritten label on it reads: “OPEN WHEN YOU’RE 24.” She stares at it in disbelief. Instant curiosity. SHOT 2 — 0:03–0:06 — THE MYSTERY She sits on the floor beside the bed and carefully opens the box. Inside is an old photograph and a tiny handwritten note. Her playful expression slowly disappears. She picks up the photograph. The camera autofocuses onto it. The photograph shows a much younger version of herself standing in the exact same bedroom. She looks from the photograph to the room. Her expression becomes completely stunned. SHOT 3 — 0:06–0:09 — THE CONNECTION She slowly looks around. The camera remains completely still. She notices something behind the old photograph: a tiny folded note. She opens it. Her eyes widen. She starts laughing in disbelief while shaking her head. The note contains a simple handwritten message: “If you found this… you finally grew up.” Natural emotional reaction — not melodramatic. SHOT 4 — 0:09–0:12 — THE EMOTIONAL TWIST She sits silently for a moment. Her expression becomes soft and emotional. She looks at the old photograph again, then smiles warmly. She gently places the photograph against her chest. A tiny nostalgic laugh escapes. The warm bedroom lamp creates a beautiful glow across her face while the MiniDV camera struggles slightly with exposure. SHOT 5 — 0:12–0:15 — THE FINAL REVEAL Suddenly she notices something written on the back of the photograph. She flips it over. Her eyes go wide. She looks directly into the camera as if she has just realized something. Then she laughs and says: “Wait… YOU recorded this?” She quickly walks toward the camera. The footage becomes shaky. She reaches for the camcorder. Her fingers cover the lens. The screen turns black. CUT. Add a tiny authentic MiniDV recording-stop sound. VIRAL DIRECTION The video should feel like viewers have accidentally discovered a private memory from another person's life. The emotional progression must be: MYSTERY → CURIOSITY → DISCOVERY → SURPRISE → EMOTION → FINAL REVEAL. Make the first frame immediately intriguing. Make the photograph reveal visually obvious. Make the final line create a reason to replay the entire video. Absolutely avoid polished commercial cinematography, modern smartphone aesthetics, unrealistic facial expressions, excessive camera movement, fantasy effects, random objects appearing, character changes or costume changes.

Cinematic scene

Create a 20-second ultra-realistic premium skincare commercial using the uploaded model image and uploaded LUXEVA Radiance Day Cream product image as strict visual references. STRICT CONSISTENCY LOCK Keep everything exactly consistent with the reference images: - Same adult female model and same facial identity throughout. - Same hairstyle, makeup, skin tone and facial features. - Same white top and pink skirt. - Same bedroom/interior, flowers, lighting style and overall pink luxury aesthetic. - Use the exact same LUXEVA cream jar, packaging, rose-gold lid, logo, colors and visible label design. - Preserve the same brand name and captions/text shown in the reference. - Do not redesign, replace, recolor or distort the product. - No face morphing, outfit changes, extra fingers, warped hands, distorted product labels or random text. 0:00–0:04 — CINEMATIC INTRO Start with a beautiful close-up of the LUXEVA Radiance Day Cream surrounded by soft pink flowers. The camera makes a smooth elegant orbit around the product, with subtle reflections moving across the rose-gold lid. Transition naturally toward the model sitting in the same position as the reference image. Sound: premium soft beauty music begins, delicate glass chime, subtle flower/air ambience and a luxurious cinematic whoosh. 0:04–0:07 — MODEL + PRODUCT Camera slowly circles toward the model. She gently picks up the same LUXEVA cream jar, holds it close to her face and looks at the packaging with a soft, confident smile. Insert a quick macro product shot where the branding remains sharp and readable. Sound: subtle jar handling ASMR + elegant rising musical note. 0:07–0:10 — OPENING THE CREAM She slowly twists and opens the rose-gold lid. Show a premium macro close-up of: - fingers opening the jar, - cream texture, - lid movement, - realistic reflections. She takes a small amount of cream with her fingertip. Sound: satisfying soft lid-opening click + delicate beauty ASMR. 0:10–0:14 — APPLICATION METHOD Show the skincare application clearly and naturally. She places small dots of cream on: - both cheeks, - forehead, - chin. Then gently blends the cream using soft upward and outward circular motions, avoiding harsh rubbing. Include one close-up of her cheek while the cream blends smoothly into the skin. Keep skin texture realistic. 0:14–0:17 — BEFORE → AFTER REVEAL Create an elegant before-and-after beauty transition using the exact same model, angle and lighting. BEFORE: natural skin appearance, slightly less luminous. A soft vertical light sweep passes across her face. AFTER: skin looks fresh, hydrated, smooth and naturally radiant, with a subtle healthy glow. Do not make the transformation artificial or extreme. On-screen caption: BEFORE → AFTER Then: Hydrated • Smooth • Radiant Sound: soft shimmer + satisfying glow-reveal sound synchronized with the transition. 0:17–0:20 — HERO ENDING The model smiles naturally, softly touches her cheek and looks toward the camera. Camera pulls back slightly while the same LUXEVA jar appears clearly in the foreground, surrounded by elegant pink flowers. Finish with a premium product hero shot. Preserve the exact same LUXEVA branding/caption from the uploaded reference image. Final text: LUXEVA RADIANCE DAY CREAM Brighten • Hydrate • Protect Final beauty tagline: “Reveal Your Natural Glow.” AUDIO STYLE Create a unique luxury skincare sound identity, not generic stock music: soft feminine ambient melody + delicate piano notes + airy synth textures + subtle glass chimes + premium ASMR product sounds + elegant cinematic whooshes. Music should build gently throughout the 20 seconds and end with a memorable soft sparkle/chime signature on the final product shot. CAMERA & QUALITY Ultra-realistic 4K beauty commercial, premium UGC + luxury brand advertising style, smooth gimbal movement, slow orbit shots, macro product cinematography, shallow depth of field, realistic hands and facial movements, natural blinking, realistic cream texture, warm golden sunlight, soft pastel-pink color palette, gentle bloom, professional beauty lighting, polished commercial editing. Important: The uploaded images are the master reference. Maintain the same model, same product, same branding, same captions, same clothing and same visual identity throughout the entire 20-second video.

Cinematic scene

*A cinematic magical-realism sequence set on a quiet European-style city street during a gentle rainfall at blue hour. A beautiful young woman in a light pastel raincoat walks calmly through the rain. Curious, she raises her hand and catches a single raindrop. The instant it touches her fingertips, it transforms into a delicate blooming flower. As she continues walking, every raindrop that lands on her or touches the ground blossoms into vibrant wildflowers, roses, daisies, cherry blossoms, and colourful petals. Within seconds, the once-grey street transforms into a breathtaking floral paradise. Flowers bloom naturally from the wet pavement, climb walls, decorate lamp posts, and spread across rooftops. Soft petals swirl gracefully in the wind while warm golden streetlights reflect on the rain-soaked road, creating a dreamy, enchanting atmosphere.* *The camera opens with an extreme close-up of the raindrop touching her fingertips, then smoothly transitions into a slow cinematic tracking shot following her as flowers bloom behind every step. Include realistic character animation, subtle facial expressions of wonder, natural walking motion, flowing hair and clothing physics, volumetric rainfall, realistic water splashes, soft fog, depth of field, cinematic lens flares, HDR lighting, global illumination, highly detailed flower growth animation, seamless motion, ultra-realistic textures, magical realism, premium film quality, 4K, 24fps, masterpiece, emotionally uplifting, visually breathtaking.* **Negative Prompt:** *Low quality, blurry, low resolution, cartoon, anime, CGI look, plastic skin, unrealistic flowers, duplicated flowers, flickering, jitter, unstable motion, deformed anatomy, extra fingers, bad hands, distorted face, oversaturated colours, flat lighting, frame glitches, artifacts, text, subtitles, watermark, logo, noise, compression, camera shake, abrupt transitions, unrealistic physics.*

Cinematic scene

15-second, 16:9 cinematic luxury fashion commercial. Concept: FROM EVERYDAY TO ICON. Use the FIRST FRAME and FINAL FRAME as strict visual anchors. Keep the EXACT SAME GIRL throughout: identical face, hair color, hair length, facial features, body proportions and identity. One continuous forward walk, one continuous shot, no cuts, no teleportation, no sudden camera changes. 0–4s — EVERYDAY: Start from the first-frame reference. A young woman stands in a bright modern apartment wearing a white T-shirt, blue straight-leg jeans and white low-top sneakers. Natural daylight enters through floor-to-ceiling windows. She naturally begins walking forward with calm confidence. Camera follows smoothly using a cinematic 35mm lens, maintaining a full-body or near-full-body composition. Natural expression, no exaggerated smile. 4–8s — MATERIAL TRANSFORMATION: As she continues walking through a bright beam of sunlight, the transformation begins gradually. The cotton texture of her white T-shirt becomes progressively finer and more luxurious, naturally evolving toward a silky premium fabric. Her blue jeans gradually develop richer texture and a more sophisticated appearance. The transformation must happen continuously during her movement, never as an instant outfit change. 8–11s — COUTURE EVOLUTION: While she keeps walking, the white T-shirt progressively transforms into an elegant black couture top, while the jeans naturally evolve into a flowing black couture skirt. Her white sneakers gradually transform into refined black high heels. Fabric must maintain realistic weight, folds, tension and movement. No flickering, disappearing clothing, clipping or extra limbs. 11–13s — ENVIRONMENT TRANSFORMATION: Without cutting, the bright apartment environment gradually evolves around her. Walls, furniture and windows subtly darken and fade into a sophisticated dark fashion studio with soft cinematic lights and delicate atmospheric haze. It should feel like the space is being transformed around her, NOT like she has been teleported. 13–15s — ICONIC HERO: She is now fully dressed in a stunning black couture gown and black heels. She reaches the center of the luxury fashion studio and slows to a stop. Camera transitions naturally from tracking to a subtle elegant arc as the gown flows around her. She turns slightly toward camera with a calm, confident fashion-editorial expression. End with a premium hero composition matching the final-frame reference: dark studio, elegant rim lighting outlining her hair, shoulders and couture silhouette. Camera slowly pulls back and settles into a refined luxury fashion campaign shot. STYLE: Photorealistic luxury fashion campaign, cinematic 35mm photography, premium editorial lighting, realistic skin and fabric texture, sophisticated black-and-white visual evolution, subtle atmospheric haze, shallow depth of field, physically accurate fabric movement, elegant restrained camera choreography. STRICT CONSISTENCY: Same girl, same face, same hair, same body proportions, same identity. One continuous forward walk. Clothing transformation must be gradual and physically believable. Environment transformation must be gradual and seamless. Correct anatomy, two arms, two hands, two legs and realistic feet.

一句话了解

这个模型怎么样？

没有图片时用提示词开始；已有画面时，可加入 1 张或 2 张图片，并说明希望参考的人物、构图或动作。图片会影响输入路径，但不会自动说明两张图之间的关系，提示词仍需填写。

模型定位
这是一条围绕文字或图片组织的视频工作流。图片任务不发送文字模式的宽高比选项；音频、固定镜头和内容检查则分别作为请求提交，实际效果要看结果。
适合谁用
适合先试一个清楚的镜头任务：看文字能否带出目标动作，比较单图与双图的影响，或在相同设置下切换音频与固定镜头请求。人物、商品或广告是否可用，仍要检查真实结果。
本站使用说明
先用分辨率、输出时长和音频请求核对预计积分。Kyeo 受理后会先预扣；成功后仍须确认实际用量有效且可用于结算。用量尚未确认、出现异常或超过预扣额时，会继续保留预扣并核对。失败或结算差额的退还，应查看明细和账户可用余额；积分有效期及欠额抵扣也会影响最终可用金额。

关键参数一览

根据这些信息，判断模型是否适合你的任务。

模型类别
AI 视频模型
厂商
ByteDance
模型家族
Seedance
任务方式
提交后等待处理完成
提示词要求
去除首尾空白后至少 3 字符，原文最多 2,500 字符
图片要求
JPEG／PNG／WebP，每张不超过 10 MiB，最多 2 张

它还叫什么

同一个模型在公开资料中可能有不同名称，这里整理相关称呼，方便查找和比较。

Seedance 1.5 Pro
Seedance V1.5 Pro
ByteDance Seedance 1.5 Pro

用户常见问题

从任务、输入条件和结果要求出发，整理选择模型时需要确认的问题。

如何开始 Seedance 1.5 Pro 的视频任务
Seedance 预计预扣积分怎么计算
图片的格式和大小需要满足什么要求
输出时长可以填写哪些整数
音频开关会怎样改变预计积分
固定镜头选项表示什么请求
单图与双图应该如何测试
上传图片后为什么看不到宽高比选项
与 Seedance 2.0 比较前应核对什么
如何选择视频分辨率并核对预算

选型指南

根据输入素材、任务要求和积分成本，逐项确认模型是否适合。

1
先决定图片负责什么

先说明图片用来参考人物、构图还是动作。比较 1 张与 2 张图片时，固定提示词和其他设置，再检查第二张图是否带来需要的变化；不要预设它一定代表结尾画面。

2
切换音频时重新核对预算

固定提示词、图片、分辨率、时长和镜头设置，只切换音频请求并重新计算预计积分。查看结果是否含所需声音；若含声音，再检查对白、节奏与画面是否合用，并记录还需修改的部分。

3
从低预算小样开始

可先选 480p、4 秒，关闭音频和固定镜头请求，测试一段清楚的动作。之后一次只改分辨率、时长、声音或图片数量，每次重新核对预计积分和实际结果。

模型对比

把常见候选放在同一任务下比较，帮助你确认输入方式、控制项和成本差异。

与 Seedance 1.0 Pro Fast 比较前，输入和时长都能对应吗？
与 Seedance 2.0 比较时，怎样保持素材和设置可比？
切换音频请求后，应该重新核对哪些费用与结果？
与 Kling 2.6 对照时，双方都接收这些图片吗？
比较 Hailuo 2.3 时，双方都提供目标输出规格吗？
同题测试需要记录哪些设置、等待和结果？

模型横向对比表

按同一组维度比较当前模型和候选模型的输入、设置与使用边界。

输入选择
Seedance 1.5 Pro
可只填提示词，也可加入 1–2 张图片；音频生成是输出请求，不是上传音频参考。
Seedance 2
先查看其输入场景能否接收手头素材，图片、视频或音频的规则不能直接套用本页。
输出时长
Seedance 1.5 Pro
4–12 秒之间的每个整数，默认 8 秒
Seedance 2
按其当前时长要求设置；只有两边都接受的输出时长，才适合用来对照预算和结果。
设置与请求选项
Seedance 1.5 Pro
设置分辨率、时长和文字任务的宽高比，再选择是否请求音频、固定镜头或 NSFW 检查。
Seedance 2
先核对所选场景实际开放哪些设置；同名开关也要分别确认请求含义和结果表现。
分辨率选择
Seedance 1.5 Pro
480p／720p／1080p
Seedance 2
查看目标分辨率是否在其当前选项内，并确认该规格是否与所用素材兼容。
Seedance 2.0 Fast
先检查当前规格与素材要求；名称中的 Fast 不保证排队或生成更快，预算也须在共同支持的条件下比较。

适用情况

依据公开资料与当前站内设置，说明模型适用的任务及需要注意的限制。

默认预计预扣为 28 积分

默认是 720p、8 秒，音频、固定镜头和 NSFW 检查请求均关闭；无图片时宽高比为 1:1。该组合预计预扣 28 积分，调整分辨率、时长或音频后要重新计算，成功后也仍需确认实际用量。

参数名称不能代替结果判断

Pro、分辨率和固定镜头名称都不能证明画质、稳定性或速度。比较时先确认双方支持相同输入与目标规格，再记录画面、声音、等待、失败和重试情况。

能力详解

文字与图片两条输入路径

可以不上传图片，或使用 1–2 张图片配合提示词。图片支持 JPEG、PNG、WebP，每张不超过 10 MiB；需要可正常解码，并在生成前通过尺寸、大小与签名元数据核验。此工作流不接收视频或音频参考素材。

54 种设置组合的预算

按分辨率、4–12 秒的整数时长和音频开关，可得到 54 种设置组合。关闭音频请求时，预计预扣为 7–90 积分；全部合法组合的预计预扣范围为 7–180 积分。每次先用相应的每秒费率乘以时长，再将总额向上取整。

画幅与镜头请求分开处理

文字任务可选择宽高比；图片任务会隐藏该项，发送时不带画幅选项。固定镜头开关只是请求意图，主体、背景与运镜是否符合要求，都要在结果中检查；关闭它也不保证镜头一定运动。

适合做什么

检验文字里的动作要求

不上传图片，先固定画幅、分辨率、时长和音频请求，用一段明确的描述检查主体动作、镜头变化、文字和构图。若结果含声音，再听它是否符合对白或环境声要求。

比较单图与双图的影响

先用 1 张图片测试，再在其他设置不变时加入第 2 张。记录人物、构图与动作的变化，判断额外图片是否有帮助；同一组图片也不能保证在不同模型中得到相同结果。

对照音频请求的成本与结果

在同一组视频设置下分别关闭和开启音频请求，重新核对每次预计积分，再检查实际声音内容与画面是否合用。记录可用的部分和仍需修改的部分，不用固定倍数估计预算。

提示词技巧

先写能逐项检查的镜头要求

依次交代景别、主体、动作和环境变化，再写希望怎样运镜。若请求音频，可另写对白、环境声或节奏要求；少用“电影级”“完美一致”代替具体动作，方便与结果逐项对照。

上传前整理图片构图

先裁切并整理图片，明确希望保留的主体和画面关系。加入图片后，宽高比选项会隐藏，本次图片任务不按该选项指定画幅；回到文字输入时，再核对原来选择的比例。图片构图和最终画幅仍要用结果确认。

请求固定镜头时仍要说明动作

说明哪些人物或物体可以移动，背景应怎样变化，以及希望保留的画面关系。固定镜头开关只提交相关请求，不能代替这些描述，也不能证明镜头会按预期保持不动。

为什么选它

同一工作流可从文字或最多 2 张图片开始，方便围绕现有素材设计小样。
可选 480p／720p／1080p，并在 4–12 秒之间设置整数输出时长。
预算由分辨率、时长和音频请求共同决定，调整其中一项后应重新核对预计积分。
可以用相同输入对照固定镜头请求的两种选择，检查真实结果中的运镜变化。

使用前要知道

提示词去除首尾空白后至少 3 字符，原文最多 2,500 字符。写得过短难以说明目标，填写前先检查关键动作与素材关系。
最多使用 2 张图片，每张不超过 10 MiB，格式为 JPEG、PNG 或 WebP。上传与生成前会分别核验素材，格式、容量合格不代表任务的其他条件都已通过。
开启音频请求后需要重新计算预计积分，不能把取整后的金额直接翻倍。成功状态也不等于实际用量已经确认；查看费用明细时，还要区分预扣、待核对、结算和退还状态。
模型名、分辨率和开关都不保证质量、速度或稳定性。长期使用时，应保留测试设置并定期复核结果与费用；重要文件请自行保存。

可调参数

根据这些信息，判断模型是否适合你的任务。

分辨率
可选
参数类型: 选择项
默认值：720p
480p
720p
1080p
输出时长
可选
参数类型: 数值
默认 8 秒
4–12 秒，仅限整数
请求生成音频
可选
参数类型: 选择项
默认关闭
不请求音频
请求音频
宽高比
可选
参数类型: 选择项
默认值：1:1
1:1
4:3
3:4
16:9
9:16
21:9
请求固定镜头
可选
参数类型: 选择项
默认关闭
不锁定镜头
请求固定镜头
NSFW 内容检查
可选
参数类型: 选择项
默认关闭
关闭
开启

积分消耗

预计预扣 7–180 积分／次

关闭音频请求时，480p、720p、1080p 每输出秒分别按 1.75、3.5、7.5 积分估算；开启时分别按 3.5、7、15 积分估算。将对应费率乘以输出时长后，合计向上取整，不能把取整后的无音频金额直接翻倍。Kyeo 受理任务后先预扣预计积分；任务成功后，还须确认实际用量有效且可用于结算。实际用量信息尚未确认、出现异常或超过预扣额时，会保留待核对状态。失败或结算差额的退还情况，请查看明细状态和账户可用余额；余额还受积分有效期和欠额抵扣影响。

使用建议

准备多次生成时，建议先用少量素材测试，核对输入设置和结果是否适合任务，再决定是否继续。每次生成均按工作台显示的计费规则扣点。

常见问题

相关模型

按你的任务比较这些模型的输入要求、功能和当前状态，再查看对应详情。

Seedance 2.0

从文字描述、首帧或参考素材出发制作视频。标准版提供文字、首尾帧和多模态参考场景；先选与素材相符的路径，再核对分辨率、预计积分和结果中需要检查的内容。

AI 视频模型
预计预扣 164–3,840 积分／次
查看详情立即使用

Kling 2.6

用文字或 1 张起始图制作单镜头短视频，可选 5 秒或 10 秒，并选择是否生成声音。预计预扣由时长和声音组合确定；生成后仍需检查对白、音画同步与实际结算状态。

AI 视频模型
预计预扣 55～220 积分／次
查看详情立即使用

Hailuo 2.3

已有一张产品图或人物图，可以配非空提示词尝试短视频动作。先选择 Standard/Pro 及有效的分辨率、时长组合查看预计预扣，再用实际结果检查动作、身份和镜头表现；选项本身不保证成片质量。

AI 视频模型
预计预扣25–80积分；最终费用需结算
查看详情立即使用

信息来源

页面内容依据公开模型资料与 Kyeo AI 的模型信息整理。能否在线使用，请以当前模型状态为准。

来源说明

本页按 Kyeo 的接入配置说明文字和图片生成视频：可只填提示词，也可加入最多 2 张图片；输出时长为 4–12 秒的整数。音频和固定镜头都是请求选项，效果需查看结果。预计积分按分辨率、时长和音频请求计算，提交前请确认工作台当前选项。

内容复核日期: 2026-08-27
Seedance 1.5 Pro 本站接入说明
平台

用于说明本站的输入路径、参数和素材核验；当前是否开放及有哪些可选项，仍需在工作台确认。

ByteDance Seedance 1.5 Pro
平台

说明本页预计积分所用的分辨率、时长与音频请求条件；取整和结算应分别理解，最终费用仍需确认实际用量。

提示词与图片输入要求
平台

用于核对本站提示词长度、图片数量、格式和单文件容量；素材上传后，提交生成前仍会再次核验。