PixVerse Image to Video

PixVerse is one of the many hosted video model vendors that ComfyUI now fronts with first-party API nodes, and the image-to-video variant is the one most people land on first: still image in, animated clip out. It ships with ComfyUI core, needs no install, and does all the real work on PixVerse's servers through Comfy's proxy, billed per render from your Comfy account credits.

The schema is small and honest. image is your start frame, prompt describes the motion (the tooltip is blunt: "Prompt for the video generation"), and quality defaults to 540p with a climb to 720p and 1080p. duration_seconds is 5 or 8, motion_mode is normal or fast, and seed is the standard non-determinism-tolerant re-run switch. Optional extras: negative_prompt for undesired elements, and pixverse_template - a special PIXVERSE_TEMPLATE input produced by the sibling PixVerse Template node, which injects a style preset into the generation. That template input is the vendor's personality; it's the closest this family comes to a one-click style, and it's worth a look if you keep landing on the same look.

The gotchas are baked into the code, not the marketing, so here's the honest version. Pick 1080p and PixVerse forces you back to a 5-second, normal-motion generation - the node silently overrides your choices, because the higher resolution doesn't support the fast tier or the longer cut. Likewise, choosing the 8-second duration forces normal motion. If you want fast motion, you're at 5 seconds and ≤720p. That's a real constraint on what this node can produce, and it's the kind of thing that eats an hour if you're not told.

Also worth knowing: generation is async and takes its time - a video job can sit in the queue for tens of seconds to a couple of minutes while the node polls. The output is a single VIDEO, same as every partner/video node, so it drops into Save/Preview or downstream without ceremony.

The motion_mode choice is the one creative decision that matters most. normal is the default, grounded, natural-motion look; fast is PixVerse's higher-energy mode, better for stylized or exaggerated animation. If your clip comes back sedate, that's usually fast mode waiting to be switched on - not a prompt problem.

Where it fits in a workflow: it's the quick-and-dirty "bring this still to life" node. You're not going to get the control of a first/last-frame setup, and 5–8 seconds at ≤1080p is a ceiling you should plan around. Use it for drafts and motion tests, then graduate to a bookended node when the shot matters. And remember the partner-node rules: the image leaves your machine, the vendor's content filter can reject a generation (that failure mode is the filter, not your graph), and the price badge is there to remind you to iterate cheap before spending on 1080p.

PixVerse's API nodes have been in core since mid-2025, and this one's been the stable, low-drama on-ramp to the service ever since.