PixVerse V5

PixVerse V5 is the current quality leader of the PixVerse text-to-video nodes in the Runware pack - and notably, the node that dropped the camera-movement knob V4 had in favor of raw model quality. You get the same prompt-in, VIDEO-out shape, the same style and effect dropdowns, and a model that simply renders motion, light, and consistency better than its predecessors. If V4 is where PixVerse learned to direct, V5 is where it learned to look good doing it.

The one thing worth flagging up front: this is the node you pick when the footage itself has to be good - hero shots, anything that'll be watched full-size - because you're paying a premium per run for it. It's the flagship, and it's priced like one.

What you actually set

• positivePrompt - required. V5 responds well to detailed scene description; put the mood and light in there.
• frameImages - IMAGE socket for image-to-video. V5 is a strong image-to-video model, and starting from a frame you generated locally is the workflow that gets the best results.
• size - 360p to 1080p, all the standard aspects, default 720p (16:9).
• duration - 5 or 8 seconds.
• providerSettings.pixverse.style - anime, 3d_animation, clay, comic, cyberpunk, or default.
• providerSettings.pixverse.effect - the viral templates from the earlier versions.
• negativePrompt, seed, numberResults - the usuals.

Output: one video socket, type VIDEO, into a video preview or straight to your output folder.

Install

One-time pack install: ComfyUI Manager → search Runware → install → restart, or

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

Runware API key into ComfyUI Settings → Runware API key or RUNWARE_API_KEY. No models to download - cloud-side weights, that's the pack's whole deal.

Where people get burned

• The price is the feature. V5's per-run cost is the highest in the line, and numberResults: 4 plus 8-second duration stacks it fast. Watch the title-bar readout; this node can chew through a small credit balance quickly.
• No camera control. If you need a specific dolly or crane move, that knob lived on V4 and it isn't here. V5 bets on the prompt and the model doing it better unprompted.
• safety.mode defaults to none. Opt into content checking (fast/full) if you want a guardrail.
• It's not a different UX. Same dropdowns, same output - the only real difference from V3.5/V4 is the model behind it and the bill. If you can't tell the quality apart on your content, drop back and save the money.

The verdict: for a polished, single-take text-to-video clip, V5 is the one to reach for. Just go in with your eyes open about the cost and you'll be fine.