EvoLink Seedance 2.0 Fast Image-to-Video (Official)

You've got a still you love and you want to see it move. EvoLink Seedance 2.0 Fast Image-to-Video (Official) is the budget way to ask: Seedance 2.0's Fast tier, capped at 720p, driven by a required first frame and an optional last frame. If you've never used a first/last-frame workflow, this is the cheapest place to learn it - the Fast lane keeps the tuition down while you figure out what a video prompt even sounds like.

What it is

• first_frame (required) - the starting image, wired from any image output. Your local file uploads through EvoLink's channel and is auto-deleted after 72 hours. The clip's aspect ratio follows this frame.
• last_frame (optional) - connect one and the video has to move from the first frame to this one. Different aspect ratio? It auto-aligns to the first frame.
• prompt (required) - the motion description, Chinese welcome.
• duration - 4–15 seconds, default 5, billed by duration.
• quality - 480p or 720p; Fast tops out at 720p.
• generate_audio - on by default, no extra charge.

Outputs: video (into Save Video - nothing shows otherwise), task_id, status_info, result_urls (24-hour expiry), response_json. Default timeout is 30 minutes; Fast jobs usually finish quicker, but "quicker than 30 minutes" still isn't "while I watch."

Install and key

Same EvoLink install as the whole pack. ComfyUI Manager → search EvoLink → install → full restart. Manual:

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

Dependency is just requests; no models, no GPU. Key from evolink.ai/dashboard/keys (sk-), pasted once - saved locally, blank after, never in a shared workflow JSON.

How to get real value out of it

Fast I2V is the "does this concept have legs" node. Test your motion prompt and your frame here at 480p; if the idea survives, re-run it on the full Seedance 2.0 I2V at higher resolution. The first/last-frame trick is the one worth keeping in your pocket: use your first frame as the last frame too and the clip becomes a loop by construction - the model has to return to where it started. It's the cheapest "infinite loop" you can buy from an API. Billing note, same as every Seedance call: per-render, and the community's pricing math has been done publicly - drafts at the Fast tier are how you keep that math on your side.