Superside Seedance Pro Image-to-Video

Seedance Pro is the higher-quality rung of ByteDance's image-to-video ladder, and this node is its ComfyUI front door over fal.ai. It takes a starting image plus a prompt and produces a video - up to 1080p, with one control that's easy to underrate until you've needed it: an end-frame image that tells the model exactly where the shot has to land.

What it does

The one-image-to-video case is the core: still frame in, motion out. Where Pro earns its keep over Lite is resolution (1080p vs 720p) and the end_image input. If you've ever watched a generated clip drift toward an ending you didn't want - the camera swings somewhere odd, the subject walks out of frame, the final pose is wrong - you know why end-frame control matters. Feed the last frame you actually want and the model has a destination to animate toward, which makes Pro the right pick for anything with a choreographed ending rather than a free-form drift.

The inputs that matter

• image (required) - the starting frame.
• end_image (optional) - the desired last frame. The feature that justifies the Pro tier for narrative work.
• prompt - the motion instruction. Default is a skier gliding over fresh snow; replace it with what your scene does.
• resolution - 480p / 720p / 1080p (default). This is the Pro upgrade, and it's where the per-call cost lives.
• duration - 3–12 seconds, default 5.
• aspect_ratio - auto or 21:9 / 16:9 / 4:3 / 1:1 / 3:4 / 9:16.
• camera_fixed - locks the camera so the scene moves within a fixed frame. On by default it's a product-shot staple; off it lets the camera roam.
• seed - reproducibility. Set it to iterate on the same take; leave random for variety.
• enable_safety_checker - on by default.

The output you have to handle

Same as the Lite node and worth repeating: the output is a STRING - a video URL, not a native VIDEO tensor. The node hands you a direct fal.ai link to the rendered clip. To save it, you'll need a URL-to-file download node or a manual download - don't expect a VIDEO wire you can plug straight into SaveVideo. (The pack's SeedVR Upscale Video node, by contrast, returns native VIDEO.)

Installing it

It ships in comfyui-superside-nodes - clone, install requirements, restart:

cd ComfyUI/custom_nodes
git clone https://github.com/Superside/comfyui-superside-nodes.git
cd comfyui-superside-nodes
pip install -r requirements.txt

Paste your fal api_key. No model files - Seedance Pro runs entirely on fal's side, and every render is metered.

The honest take

Pick Pro when the deliverable actually needs 1080p or the end-frame handoff - that's what you're paying for. Pick Lite when you're iterating on motion and prompts, because burning Pro credits on every draft is how a video project turns into a line item someone questions. The end-frame control is the sleeper feature: once you've used end_image to pin a product's final hero pose or a character's exit, you'll structure your shots around it. And remember, whatever tier you choose: your source image and prompt leave the machine and go to ByteDance's servers - that's the standing trade of every closed-model API node in this pack.