Seedance Pro Text/Image-to-Video

Seedance Pro is where this pack stops being "cheap and cheerful" and starts being "the good stuff." This node runs the Pro 1.0 line - seedance-1-0-pro-250528 or the faster seedance-1-0-pro-fast-251015 - and it's a genuinely different proposition from the Lite nodes: it handles text-to-video and image-to-video in one node via a mode switch, defaults to 1080p, and the regular model supports an end frame, which turns an image-to-video clip into a first+last frame interpolation. Seedance is ByteDance's flagship video model and, per the KB's ByteDance panel, one of the closed API products the open-weights community constantly measures itself against - you're renting it per render, not downloading it.

It's the middle step between the Lite nodes and the newer Pro 1.5 node. The 1.5 node adds draft-then-render and audio, but this node's mode switch plus model choice makes it a tidy workhorse: text-only, or image-to-video with an optional end frame. The fast model skips end-frame support - the dropdown says so right in the name - and only the regular model does the first+last trick.

How it works

Prompts and images go into an async task on BytePlus; the node polls until the hosted video URL is ready (up to five minutes). In image-to-video mode the image becomes a first_frame; connect end_frame with the regular model and it's sent as a last_frame too. The 1080p default is the one to watch - that's a heavier, slower, pricier render than the Lite defaults.

Inputs and outputs that matter

• model - regular (end-frame capable) or fast (no end frame, quicker/cheaper).
• mode - text-to-video or image-to-video. Image mode requires the image input.
• prompt - motion and scene; the image carries the look in image mode.
• resolution (480p/720p/1080p), ratio, duration (3–12s).
• image and end_frame - optional images; end_frame only works in image mode with the regular model.
• seed, camerafixed, framepersecond (locked at 24).

Output: one STRING with the video URL. Preview/save it or pull frames.

Install

cd ComfyUI/custom_nodes
git clone https://github.com/FloyoAI/ComfyUI-Seed-API
cd ComfyUI-Seed-API
pip install -r requirements.txt

Restart, set your BytePlus API key in config.ini or BYTEPLUS_API_KEY. Manager: search "ComfyUI Seed API Integration".

Where people get burned

The classic mistake is asking for end-frame support with the fast model - it won't happen, and the node silently ignores the end frame there. Another one: switching mode to image-to-video and forgetting the image, which just errors into the console. And the 1080p default will quietly make your test renders slow and pricey; drop to 720p while you're iterating on the prompt. Empty string out = check the console, as always with this pack. If you mostly do text-to-video, the Lite node is cheaper for experiments and this Pro node is where you land when the clip actually matters.