scenario-kling
Compare original and translation side by side
🇺🇸
Original
English🇨🇳
Translation
ChineseScenario Kling Video
Scenario Kling 视频
Overview
概述
Kling, Kuaishou's video family on Scenario, ships as specialists, eighteen at authoring time: V3 (Omni plus dedicated T2V and I2V tiers), O1, 2.6, motion control, lipsync, and a talking avatar. Pick the member whose conditioning matches the job, then treat as the contract: the same parameter changes shape, default, and legality between members. Kling Video to Audio is audio output, the skill's domain.
model_schema_getscenario-audioConnection and the core loop: see the skill in this repo; model-agnostic video work: the skill. If a sibling skill named here is missing from your available skills, ask the user to install it (); unattended, proceed from tool schemas and flag the gap.
scenarioscenario-videonpx skills add scenario-labs/skills --skill <name>Kling是快手在Scenario平台上推出的视频模型家族,在本文撰写时共有18个专用模型:V3(含Omni以及专用的T2V、I2V档位)、O1、2.6、动作控制、口型同步和数字人模型。请选择适配任务需求的模型,将作为接口约定:同一参数在不同模型中的格式、默认值和合法性可能不同。Kling视频转音频的输出功能属于技能的范畴。
model_schema_getscenario-audio连接方式与核心流程:参见本仓库中的技能;通用视频处理功能:参见技能。如果你的可用技能中缺少此处提到的关联技能,请提示用户安装(命令:);若处于无人值守模式,可基于工具Schema继续操作并标记缺失项。
scenarioscenario-videonpx skills add scenario-labs/skills --skill <name>Quick reference
快速参考
Pick by job; discover ids with (, , ):
searchtarget="models"query="kling"public=true| Job | Members |
|---|---|
Every mode in one model, tier via | V3 Omni ( |
| Text to video at a fixed tier | V3 T2V Standard / Pro / 4K, 2.6 T2V Pro |
| Animate a still, optional last frame | V3 I2V Standard / Pro / 4K, O1 I2V, 2.6 I2V Pro |
| Consistent characters from photos | O1 Reference Images, V3 I2V |
| Edit or restyle footage by prompt | O1 Video Editing / Reference Video, Omni ( |
| Motion from a video onto a character image | Motion Control (V3 Pro / Std, V2.6) |
| Talking head | AI Avatar (image plus audio), Lipsync (video plus audio or text) |
Schema traps (names from live schemas, caps at authoring time):
- V3 dedicated lines take or
prompt, never both;multiPromptis an array ofmultiPromptshots, and{prompt, duration}(shotTypeorcustomize) exists only on the V3 T2V members and I2V 4K. Omni keepsintelligentrequired; itspromptis a JSON string of up to 6 shots whose durations sum tomultiPrompt.duration - is a string enum on most members (
duration, not"5"); Omni alone takes a number, 3 to 15. The V3 dedicated lines reach 15 seconds; O1 and 2.6 stop at 10.5 - On O1 and V3 I2V an element is a plus up to 4 angle
frontalImage: both required on O1, each optional on V3 I2V, where areferenceImagescan define the element instead.video - Budgets shrink beside a video: O1 Reference Images takes 7 total (elements plus images), the O1 video members 4, and Omni's drops from 7 to 4 next to
referenceImages. Tags bind by order (referenceVideo,@Element1); Omni prompts use@Image1and<<<image_1>>>.<<<video_1>>> - A reference clip runs 3 to 10 seconds on Omni and the O1 video members, 2 to 10 on Lipsync (trim first, per ). Omni's
scenario-videodefaults toreferenceVideo, lending style and camera to a new clip;videoReferenceType: "feature"edits the clip and ignores"base";durationmode refuses a reference video either way.4k - (0 to 1, default 0.5): raise toward 0.8 for storyboard fidelity, drop toward 0.3 to let the model invent.
cfgScale - Prompt in director order: scene, subject, action, one camera move, then audio and style; natural sentences beat tag lists, and complex scenes hold together near 5 seconds, not 15.
根据任务选择模型;可通过接口查询模型ID(参数:、、):
searchtarget="models"query="kling"public=true| 任务类型 | 可用模型 |
|---|---|
单模型支持所有模式,通过 | V3 Omni( |
| 固定档位的文生视频 | V3 T2V 标准版 / 专业版 / 4K版、2.6 T2V 专业版 |
| 静态图片动效化,支持可选尾帧 | V3 I2V 标准版 / 专业版 / 4K版、O1 I2V、2.6 I2V 专业版 |
| 基于照片生成一致性角色 | O1 参考图模式、V3 I2V |
| 基于提示词编辑或重制视频风格 | O1 视频编辑 / 参考视频模式、Omni( |
| 将视频动作迁移到角色图片上 | 动作控制(V3 专业版 / 标准版、V2.6) |
| 数字人播报 | AI 数字人(图片+音频)、口型同步(视频+音频/文本) |
Schema注意事项(名称来自线上Schema,限制为撰写时的版本):
- V3专用系列模型只能传入或
prompt二者之一,不可同时使用;multiPrompt是由multiPrompt格式的镜头组成的数组,{prompt, duration}(可选shotType或customize)仅在V3 T2V系列和I2V 4K版中支持。Omni模型要求必须传入intelligent;其prompt为JSON字符串,最多支持6个镜头,所有镜头的时长总和需等于multiPrompt。duration - 大多数模型的为字符串枚举类型(例如
duration,而非数字"5");仅Omni模型接受数字类型,范围为3到15。V3专用系列最长支持15秒;O1和2.6版本最长支持10秒。5 - 在O1和V3 I2V模型中,元素(element)由1张(正面图)和最多4张不同角度的
frontalImage(参考图)组成:O1模型要求两者都必须提供,V3 I2V中两者均为可选,也可通过referenceImages来定义元素。video - 当使用视频输入时,参考资源的配额会减少:O1参考图模式总共支持7个(元素+图片),O1视频类模型支持4个;Omni模型的配额在搭配
referenceImages使用时会从7个降至4个。标签按顺序绑定(referenceVideo、@Element1);Omni的提示词使用@Image1和<<<image_1>>>格式。<<<video_1>>> - Omni和O1视频类模型的参考视频时长要求为3到10秒,口型同步模型为2到10秒(需先按照技能的说明裁剪)。Omni的
scenario-video默认referenceVideo,用于为新视频赋予风格和镜头运动;videoReferenceType: "feature"模式用于编辑原视频,此时会忽略"base"参数;duration模式下无论哪种方式都不支持参考视频。4k - (范围0到1,默认0.5):调高至0.8左右可提升分镜还原度,调低至0.3左右可让模型更具创造性。
cfgScale - 提示词建议按照导演逻辑排序:场景、主体、动作、一个镜头运动,最后是音频和风格;自然语句的效果优于标签列表,复杂场景在5秒左右时长下能保持连贯,不建议使用15秒。
Audio is a per-member switch
音频功能因模型而异
generateAudiokeepAudiolastFrameImagetextvoiceIdgenerateAudiokeepAudiolastFrameImagetextvoiceIdMotion control reads two inputs
动作控制需要两个输入
Motion members require a character and a driving ; names which input holds the character and, on V2.6, caps the driving clip at authoring time: 10 seconds for , 30 for . Defaults differ (V3 Std says , the others ), so set it explicitly. decides whether the source audio survives.
imagevideocharacterOrientationimagevideovideoimagekeepOriginalSound动作控制类模型需要角色(图片)和驱动(视频)两个输入;参数用于指定角色所在的输入源,且在V2.6版本中,该参数会限制驱动视频的时长(撰写时的规则):角色为时驱动视频最长10秒,角色为时最长30秒。不同模型的默认值不同(V3标准版默认为,其他模型默认为),因此请显式设置该参数。参数用于决定是否保留原音频。
imagevideocharacterOrientationimagevideovideoimagekeepOriginalSoundWorked example: two-shot character clip from a still
示例:从静态图片生成双镜头角色视频
- with
search,target="models",query="kling image to video". Prefer the newest non-deprecated hit, e.g.public=true(a live hit at authoring time: re-discover each session).model_kling-v3-i2v-pro - with that id: multiPrompt shape, duration values, element caps.
model_schema_get - the start frame and the character's frontal photo (see the
upload_assetskill).scenario - with that
model_run,model_id, anddry_run=truefor the cost estimate; re-estimate after changing durations, tier, or audio.parameters={"startImage": "asset_s", "multiPrompt": [{"prompt": "Wide shot, @Element1 crosses the plaza, slow dolly-in", "duration": "5"}, {"prompt": "Close-up, @Element1 smiles and says: 'we made it'", "duration": "4"}], "generateAudio": true, "elements": [{"frontalImage": "asset_f"}]} - Repeat with
model_run, thenwait=falsewith the returned job id, re-called withjobs_waiton timeout, never a secondpending_job_ids.model_run - the output and review both shots with sound.
asset_display
- 调用接口,参数为
search、target="models"、query="kling image to video"。优先选择最新的未弃用结果,例如public=true(撰写时的有效结果:每次会话都需重新查询)。model_kling-v3-i2v-pro - 传入该ID调用接口:确认multiPrompt格式、duration可选值、元素配额等信息。
model_schema_get - 调用上传起始帧和角色正面照(参见
upload_asset技能)。scenario - 调用接口,传入
model_run、model_id和以下dry_run=true来预估费用:parameters;修改时长、档位或音频设置后需重新预估。{"startImage": "asset_s", "multiPrompt": [{"prompt": "Wide shot, @Element1 crosses the plaza, slow dolly-in", "duration": "5"}, {"prompt": "Close-up, @Element1 smiles and says: 'we made it'", "duration": "4"}], "generateAudio": true, "elements": [{"frontalImage": "asset_f"}]} - 再次调用接口,设置
model_run,然后使用返回的任务ID调用wait=false;若超时,使用jobs_wait重新调用,切勿重复发起pending_job_ids请求。model_run - 调用展示输出结果,带音审核两个镜头的效果。
asset_display
Common mistakes
常见错误
- Passing both and
promptto a V3 dedicated line: exclusive there (Omni alone keepsmultiPromptrequired).prompt - Numeric durations: outside Omni, is the string
duration, not"5".5 - Expecting audio beside a last frame (2.6 I2V) or a reference video (Omni): exclusive pairs.
- Carrying one member's caps to another: 15 seconds, , and 4K each exist on one member, not the next.
elements - Compound camera moves ("dolly in while orbiting"): one dominant move per shot; split the rest across shots.
multiPrompt - Overloaded negative prompts: a short artifact list steers; a long one stiffens motion.
- Prompting Omni to extend a clip: it edits in place, and no Kling member extends at authoring time. Continue from the clip's
baseid (lastFrame) as the start frame of a new I2V run, then concatenate, perasset_get.scenario-video - Skipping on 4K or avatar runs: at authoring time 4K ran several times Standard and avatar cost spanned a 100x range. Iterate on a Standard tier; no V3, O1, or 2.6 schema carries a seed, so a 4K re-run of the keeper is a new take, while upscaling it (
dry_run) keeps the take.scenario-video
- 为V3专用系列模型同时传入和
prompt:二者是互斥的(仅Omni模型要求必须传入multiPrompt)。prompt - 使用数字类型的duration:除Omni外,其他模型的必须是字符串
duration,而非数字"5"。5 - 在使用尾帧(2.6 I2V)或参考视频(Omni)时期望开启音频:这些组合是互斥的。
- 套用其他模型的配额限制:15秒时长、参数、4K分辨率这些特性分别只在特定模型中支持,并非所有模型都有。
elements - 使用复合镜头运动(例如“边推进边环绕”):每个镜头建议只保留一个主要运动;其余运动可拆分到的多个镜头中。
multiPrompt - 负面提示词过多:简短的瑕疵列表可以引导模型,过长的负面提示词会导致动作僵硬。
- 提示Omni 模式延长视频:该模式是原地编辑,且撰写时所有Kling模型都不支持视频延长。可通过
base获取视频的asset_getID,作为新的I2V任务的起始帧生成后续内容,再按照lastFrame技能的说明拼接。scenario-video - 在4K或数字人任务中跳过:撰写时4K的费用是标准版的数倍,数字人费用差异可达100倍。建议先在标准版上迭代;V3、O1、2.6版本的Schema都不支持种子(seed)参数,因此对满意的结果重新跑4K会生成新的版本,而通过
dry_run技能超分辨率则可以保留原版本。scenario-video