根据用户的视频需求(场景、时长、音频、语言、平台限制、预算、是否需要本地部署/角色一致性等),从 16 个主流 AI 视频模型(12 商业 + 4 开源)中推荐最匹配的 1-3 个,并解释为什么。覆盖商业:Seedance 2.0、HappyHorse 1.0、Kling 3.0、Sora 2、Veo 3.1、**Gemini Omni**(2026-05 新)、Runway Gen-4.5/Aleph、Pika 2.5、Hailuo 02、Hunyuan Video 1.5、Wan 2.7、即梦 AI;开源:LTX-Video 0.9.7、Mochi 1、CogVideoX 5B、Higgsfield Soul。用于"用哪个 AI 视频模型好"、"Sora 还是 Kling"、"国内有什么模型"、"哪个免费"、"哪个支持中文"、"哪个能编辑已有视频"、"哪个能本地部署"、"哪个角色一致性最强"等触发场景。
npx skills add https://github.com/cclank/lanshu-awesome-ai-video-kit --skill model-selector
15 个主流 AI 视频模型(11 商业 + 4 开源)的"购物顾问"。问用户 3-4 个关键问题,给出推荐 + 理由。
<model>-prompter skillseedance-prompter / kling-prompter / happyhorse-prompterseedance-debugger| 模型 | 中文 | 音频 | 物理 | 编辑 | 时长 | 开源 | 国内可用 | 主战场 |
|---|---|---|---|---|---|---|---|---|
| Seedance 2.0 | ★★★ | — | ★★★ | — | 15s | — | ✓ | 电影叙事 |
| HappyHorse 1.0 | ★★★ | ★★ | ★★ | — | 15s(默认 5s) | — | ✓ | ASMR / 短片 |
| Kling 3.0 | ★★★★★ | ★★★★ | ★★★★ | ✓ | 2m | — | ✓ | 中文剧情 / 图生视频 |
| Sora 2 | ★★ | ★★★★ | ★★★★★ | — | 25s (Pro) | — | △ | 电影艺术片 |
| Veo 3.1 | ★★ | ★★★★★ | ★★★ | — | 148s chained | — | △ | 多人对话 |
| Gemini Omni ⭐NEW | ★★ | ★★★ | ★★★ | ★★★★★ | 待定 | — | △ | 迭代编辑 / 文字渲染 / 跨模态 |
| Runway Gen-4 | ★★★ | — | ★★★ | ★★★★★ | 30s | — | ✓ | 视频编辑 |
| Pika 2.5 | ★★ | ★ | ★★ | — | 25s | — | ✓ | 创意特效 |
| Hailuo 02 | ★★★ | — | ★★★★★ | — | 10s | — | ✓ | 物理动作 |
| Hunyuan 1.5 | ★★★★ | — | ★★★ | ✓ | 10s | ★★★★★ | ✓ | 开源 / LoRA |
| Wan 2.7 | ★★★ | ★★★★★ | ★★★ | ✓ | 15s | — | ✓ | 数字人 lip-sync |
| 即梦 3.0 | ★★★★★ | ★★ | ★★★ | — | 15s | — | ✓ | 中文 + 剪映集成 |
不要全部问,挑最影响结果的 3-4 个:
| 维度 | 问题 |
|---|---|
| 时长 | "需要多长?5 秒 / 10 秒 / 15 秒以上?" |
| 音频 | "要原生音频吗?只要环境音 / 要对白 / 要 BGM / 不要" |
| 语言 | "是中文场景还是英文场景?" |
| 编辑性质 | "全新生成还是编辑已有视频?" |
| 平台限制 | "国内可用 / 国际可用 / 本地部署?" |
| 用途 | "短视频爆款 / 影视级 / 商业广告 / 个人创作?" |
是编辑已有视频?
├── Yes → Runway Aleph(独家)
└── No → 继续
├── 要数字人/真人对口型?
│ └── Yes → Wan 2.7(业界最准)
├── 要多人对话+精确配乐?
│ └── Yes → Veo 3.1
├── 中文剧情/中式美学?
│ └── Yes → 即梦 3.0(创作者) / Kling 3.0(开发者 API)
├── 极致物理感(水/火/动作)?
│ └── Yes → Hailuo 02 > Sora 2
├── 创意爆款特效(变身/融化)?
│ └── Yes → Pika 2.5(Pikaffects)
├── 本地部署/开源/LoRA 训练?
│ └── Yes → Hunyuan 1.5
├── 极致电影质感/艺术片?
│ └── Yes → Sora 2 > Runway Gen-4
├── 5/10/15 秒紧凑短片 + 环境音?
│ └── Yes → HappyHorse 1.0
└── 复杂多镜头叙事?
└── Yes → Seedance 2.0 / Kling 智能分镜
## 推荐模型
**最佳选择**:[模型名]
**理由**:[2-3 句话说明为什么]
**对应 skill**:[skills/xxx-prompter] (如果有)
**对应方法论**:[methodology/NN-xxx-公式.md]
**官方平台**:[链接]
**备选**:[模型名]
**为什么备选**:[一句话]
## 你的需求 → 我的判断
| 你的需求 | 对应能力 | 评级 |
|---|---|---|
| [需求 1] | [模型支持情况] | ★★★★★ |
| ... | ... | ... |
澄清问题:
用户答:单个片段 10 秒,要有对话,现代都市。
输出:
> 最佳选择:Kling 3.0
> 理由:中文理解业界并列第一(与即梦),原生音频支持角色定向发声,最长 15 秒单段,5 层进阶公式天然适合剧情。
>
> 对应 skill:skills/kling-prompter
> 对应方法论:methodology/09-kling-公式.md
> 官方平台:klingai.com
>
> 备选:Wan 2.7(如果对话很多且需要数字人感强)/ 即梦 3.0(如果在剪映里直接做后期)
判断:编辑已有视频 → 直接推荐 Runway Aleph。
输出:
> 最佳选择:Runway Aleph
> 理由:编辑已有视频是 Aleph 的独家能力(其他 10 个模型都做不到)。re-style 动词专门用于风格转换,能保持原始运动和时序不变。
>
> 示例 prompt:
> `
> Re-style this clip as a hand-painted watercolor animation. Soft brush textures, flowing edges, warm palette. Preserve all motion and timing exactly.
> `
>
> 对应方法论:methodology/13-六大模型公式速查.md#1-runway-gen-4--aleph
> 官方平台:Runway
>
> 备选:没有真正可替代的。如果非要,Hunyuan 视频也能改一些(开源),但不是 native 编辑能力。
澄清问题:
用户答:RTX 4090 24GB,想自训 LoRA。
输出:
> 最佳选择:Hunyuan Video 1.5(8.3B 版本)
> 理由:开源最强、24GB 显存可跑、原生支持 LoRA 训练、社区活跃(ComfyUI 完整工作流支持)。
>
> 推荐参数:Q5_K_M GGUF + 30 steps + dpmpp_2m_sde + sgm_uniform + cfg 6.0
> 官方资源:
> - GitHub: Tencent-Hunyuan/HunyuanVideo-1.5
> - HuggingFace: tencent/HunyuanVideo-1.5
> - ComfyUI 教程: docs.comfy.org
>
> 备选:CogVideoX(智谱开源)/ LTX Video(Lightricks 实时)— 这两个未纳入本仓库主目录,但同属开源阵营。
Lightweight WSI tile extraction and preprocessing. Use for basic slide processing tissue detection, tile extraction, stain normalization for H&E images. Best for simple pipelines, dataset preparation, quick tile-based analysis. For advanced spatial proteomics, multiplexed imaging, or deep learning pipelines use pathml.
Microscopy data management platform. Access images via Python, retrieve datasets, analyze pixels, manage ROIs/annotations, batch processing, for high-content screening and microscopy workflows.
This skill should be used when working with pre-trained transformer models for natural language processing, computer vision, audio, or multimodal tasks. Use for text generation, classification, question answering, translation, summarization, image classification, object detection, speech recognition, and fine-tuning models on custom datasets.
This skill should be used when working with pre-trained transformer models for natural language processing, computer vision, audio, or multimodal tasks. Use for text generation, classification, question answering, translation, summarization, image classification, object detection, speech recognition, and fine-tuning models on custom datasets.
Digital pathology image processing toolkit for whole slide images (WSI). Use this skill when working with histopathology slides, processing H&E or IHC stained tissue images, extracting tiles from gigapixel pathology images, detecting tissue regions, segmenting tissue masks, or preparing datasets for computational pathology deep learning pipelines. Applies to WSI formats (SVS, TIFF, NDPI), tile-based analysis, and histological image preprocessing workflows.
Microscopy data management platform. Access images via Python, retrieve datasets, analyze pixels, manage ROIs/annotations, batch processing, for high-content screening and microscopy workflows.
Computational pathology toolkit for analyzing whole-slide images (WSI) and multiparametric imaging data. Use this skill when working with histopathology slides, H&E stained images, multiplex immunofluorescence (CODEX, Vectra), spatial proteomics, nucleus detection/segmentation, tissue graph construction, or training ML models on pathology data. Supports 160+ slide formats including Aperio SVS, NDPI, DICOM, OME-TIFF for digital pathology workflows.
Lightweight WSI tile extraction and preprocessing. Use for basic slide processing, tissue detection, tile extraction, and stain normalization for H&E images. Best for simple pipelines, dataset preparation, and quick tile-based analysis. For advanced spatial proteomics, multiplexed imaging, or deep learning pipelines use pathml.
Take cclank/model-selector from the repository into ~/.claude/skills for personal
use, or into .claude/skills inside a project.
The agent identifies a skill by the name field in its header. Two skills with the
same name cannot sit side by side — one of them will be ignored.