ByteDance Image to Video

ByteDance Image to Video is the workhorse first-frame animator in ComfyUI's ByteDance family. You give it one image and a prompt, it gives you a moving shot that starts from that exact frame. It's the 1.x-generation API node - the default model is seedance-1-0-pro-fast - which makes it the "older sibling" to the 2.x nodes, but it's still very much alive and, for a lot of i2v work, the one people actually reach for because it's simple and predictable.

Like every partner node here, it's a cloud call: image and prompt go up to Comfy's API layer, ByteDance's Seedance renders, a VIDEO comes back. No local model, no VRAM, billed through your Comfy account credits. If you want the current flagship tier instead, that's ByteDance2FirstLastFrameNode - but this one is the straightforward, cheap-to-draft option.

The inputs

• image - the first frame of the video. Quality in, quality out: a clean, well-composed frame gives the model a stable base to animate.
• prompt - the text prompt used to generate the video.
• model - defaults to seedance-1-0-pro-fast; other Seedance 1.x options populate here.
• resolution and aspect_ratio - output size and shape.
• duration - 3 to 12 seconds.
• seed (optional) - for generation; 0 = random.
• camera_fixed (optional) - the honest little toggle. Checking it tells the platform to append a "fix the camera" instruction to your prompt. The tooltip is refreshingly upfront: it does not guarantee the actual effect. Treat it as a nudge, not a lock.
• watermark (optional) - add an "AI generated" watermark.
• generate_audio (optional) - and here's the trap: this parameter is ignored for every model except seedance-1-5-pro. If you tick it on the default model, nothing happens.

What comes out

One VIDEO output, animated from your first frame.

Gotchas

• generate_audio is a no-op unless you're on seedance-1-5-pro. The tooltip says it in plain words; people still tick it, see no sound, and file a bug report. Switch models first if you want audio.
• camera_fixed is a hint, not a guarantee. Design for motion, don't design around the camera staying put.
• Seed = 0 is random. If you set a seed, it's used; leave it and each run is a fresh roll.
• It's a paid call. Draft on lower resolution/duration, spend on the final - the standard credit budget advice for this whole family.

The workflow that clicks: make a frame you love locally, drop it in here with a motion-focused prompt, and let the API do the animating - the way most people use the ByteDance family day to day.