EvoLink Kling O3 Image-to-Video (Official)

Kling O3 is Kuaishou's newest generation of video model, and EvoLink_KlingO3I2V is its image-to-video entry in this pack. It's the full-featured one: up to 4K output, optional last frame, per-model aspect ratio, and sound generation - a capability the cheaper Kling tiers drop. If you've got a strong reference image and want a high-end animated result rather than a draft, this is the Kling node to reach for.

The inputs that matter

• first_frame (required) - the IMAGE input that starts the clip. It must be at least 300px on each side with a width:height ratio between 1:2.5 and 2.5:1, per the tooltip. Square or near-square frames sit safest in that range.
• prompt (required) - shot and motion description, up to 2500 characters, Chinese-friendly.
• last_frame (optional) - a second IMAGE input. Connect it and the model animates from the first frame to the last, which is the classic first/last-frame trick for controlled endings.
• duration - 3–15 seconds, default 5, billed by duration.
• quality - 720p (default), 1080p, or 4k.
• aspect_ratio - 16:9 (default), 9:16, or 1:1.
• sound - off (default) or on. Generated sound effects; on the standard O3 node this is available (unlike Turbo).

Then api_key and timeout_seconds (1200).

How it works

Your frames upload through EvoLink's file channel (auto-deleted after 72h, no third-party host), the job posts to POST /v1/videos/generations as kling-o3-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. The other four outputs - task_id, status_info, result_urls (24h expiry), response_json - are the pack-wide standard.

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 install via ComfyUI Manager by searching EvoLink. If nodes don't appear, check the pack isn't nested two folders deep.

Troubleshooting

The pack-standard error set (401/402/403/429) and the VPN note - whitelist api.evolink.ai and files-api.evolink.ai. The Kling-specific gotchas: respect the first-frame minimum size and ratio - the API rejects frames outside 1:2.5 to 2.5:1, and the error you get back may not say "resize your image" in so many words. And remember 4K costs meaningfully more per run than 1080p; for drafts, 720p is the sane default. Also be aware the video output needs a recent ComfyUI with comfy_api VIDEO support.