EvoLink Kling V3 Turbo Image-to-Video (Official)

Kling V3 Turbo is the budget tier of Kuaishou's Kling line, and EvoLink_KlingTurboI2V is its image-to-video node. What you give up for the lower price is spelled out in the feature list: no last frame, no sound, no negative prompt, and a 1080p ceiling instead of 4K. What you get is a faster, cheaper Kling that's ideal for drafts, batch tests, and anything where you're iterating more than finalizing.

It's the "let me check if this idea works before spending real money" node. First frame in, clip out, done.

The inputs that matter

• first_frame (required) - the IMAGE input that starts the clip. Must be at least 300px per side, ratio between 1:2.5 and 2.5:1.
• prompt (required) - shot and motion description, up to 2500 characters, Chinese-friendly.
• duration - 3–15 seconds, default 5, billed by duration.
• quality - 720p (default) or 1080p. The tooltip is explicit: Turbo doesn't do 4K.

Then api_key and timeout_seconds (1200). And that's the whole node - deliberately stripped down, which is honestly part of the appeal.

How it works

The frame uploads through EvoLink's file channel (auto-deleted after 72h, no third-party host), the job posts to POST /v1/videos/generations as kling-v3-turbo-image-to-video, and the pack polls until the clip completes, downloading it as a VIDEO object. Wire video into a native Save Video node. Outputs: video, task_id, status_info, result_urls (24h expiry), response_json. Needs a recent ComfyUI for VIDEO output support.

Installing

No GPU, no model files, only requests:

cd ComfyUI/custom_nodes
git clone https://github.com/deeplearning-goethe/comfyui-evolink.git

then fully restart ComfyUI, or search EvoLink in ComfyUI Manager.

Troubleshooting

401 key, 402 balance, 403 access, 429 rate limit; VPN users whitelist api.evolink.ai and files-api.evolink.ai. The realistic Turbo-specific issues: don't hunt for knobs that don't exist - there's no last_frame, no sound, no negative_prompt here, so if you need those, you're on the wrong node (grab the standard Kling V3 or O3 I2V node). And keep the first frame within the size/ratio limits; the API rejects frames outside 1:2.5 to 2.5:1.