kie-kling-video

Generate video with **Kling (Kuaishou)** through the Kie.ai jobs API. > **Setup:** needs `KIE_API_KEY` in `~/.claude/.env`. ## Models | Key | Model id | Mode | Notes | |---|---|---|---| | `2.6-t2v` (default) | `kling-2.6/text-to-video` | text-to-video | sound optional; 5/10s; 1:1\|16:9\|9:16 | | `2.6-i2v` | `kling-2.6/image-to-video` | image-to-video | 1 input image; 5/10s | | `3.0` | `kling-3.0/video` | t2v + i2v | 3–15s, multi-shot, native audio, std/pro/4K | | `2.1-master-t2v` | `kling/v2-1-master-text-to-video` | text-to-video | cfg_scale + negative prompt | | `2.1-pro-i2v` | `kling/v2-1-pro` | image-to-video | tail/last frame support | ## Flow 1. **Parse** `$ARGUMENTS`. A start image → image-to-video. 2. **Settings** (ask, or defaults): - **Model:** default `2.6-t2v`. Use `3.0` for the best quality / 4K / multi-shot; `2.1-pro-i2v` for first→last-frame control. - **Duration:** 5 or 10s (Kling 3.0: 3–15s). The script snaps to allowed values. - **Aspect ratio:** `16:9` (default), `9:16`, `1:1`. - **Audio:** `--audio` for native sound (2.6 / 3.0). - **3.0 quality mode:** `--mode std|pro|4K` (default `pro`). - **2.1 tuning:** `--cfg-scale 0.5` and `--negative-prompt "..."`. 3. **Craft the prompt** (≤1000 chars for 2.6, ≤5000 for 2.1). Show before generating. 4. **Generate:** ``` python3 ~/.claude/skills/kie-kling-video/generate.py "<prompt>" \ --model 2.6-t2v --duration 5 --aspect-ratio 16:9 [--audio] \ [--image <url-or-path>] [--last-frame <url-or-path>] \ [--mode pro] [--cfg-scale 0.5] [--negative-prompt "..."] \ --slug <slug> ``` 5. **Present:** output → `~/videos/kling/<date>-<slug>/`. Offer a longer clip, different model, or audio. ## Notes - Kling is a strong, cost-effective image-to-video workhorse — good default for animating a still. - Local input images are auto-uploaded. - Default `9:16` for Shorts/Reels/TikTok.