Vidu Q2 Turbo

Vidu Q2 Turbo is the speed tier of Shengshu's second-generation video line, and the Runware node makes the trade explicit from the start: positivePrompt is a required input. Where the Pro sibling lets you skate by on references alone, Turbo wants to be told what to do. That's not a bug - it's the model's personality. The Turbo tier is optimized for throughput, the kind of model where you fire off a prompt, get a clip back quickly, and iterate. When you need to test five ideas in an afternoon, this is the one that makes that affordable.

The "Turbo" suffix is doing real work here. Like the distilled fast models in the local ecosystem - the LCMs and turbo fine-tunes that trade a bit of fidelity for dramatically fewer denoising steps - Q2 Turbo is built to be quick on the cloud's side, which on a metered API mostly shows up as a smaller bill per run. You're giving up a little polish relative to Q2 Pro in exchange for speed and cheaper iteration. For most prototyping, that's the right trade.

How it works

Standard pack machinery: the node (vidu:3@2, videoInference) builds a request, uploads any wired images via the SDK, sends it over REST, and returns a native VIDEO. referenceImages and frameImages are both optional sockets, so you can do pure text-to-video or anchor it with images. duration is a float, default 5 seconds - the same generation-two fractional control as Pro. Total frames = duration × the model's fps, with no separate frame count.

The inputs that matter

• positivePrompt (required) - the one thing Turbo insists on. Lead with subject, then motion.
• referenceImages / frameImages - optional identity anchors and starting frames.
• duration - fractional seconds, default 5.
• providerSettings.vidu.bgm - background music toggle.
• providerSettings.vidu.movementAmplitude - auto default; override for explicit motion intensity.
• seed / numberResults - reproducibility and batch variations.

Then the pack tail: width/height, outputFormat (MP4), outputQuality, safety toggles, ttl. One video output.

Install and API key

Same one-time install for the whole pack:

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

Restart ComfyUI (or install Runware from ComfyUI Manager). API key from runware.ai/api-keys, set in ComfyUI Settings → Runware API key, exported as RUNWARE_API_KEY, or via runware auth login.

Gotchas

Don't buy Turbo expecting Pro quality at a discount - it's fast and a little rougher, and the rough edges show most in complex motion and fine details. If a Turbo result is close but not quite there, the fix isn't more Turbo runs, it's switching to Q2 Pro for the final version. The required prompt also means this node punishes lazy prompting harder than the reference-driven Vidu nodes: Turbo follows what you wrote, so vague prompts give you generic motion. And the standard cloud-video warning: even at Turbo prices, each clip is a paid call, so numberResults above 1-2 is you choosing to spend. For quick idea testing it's great - just know what you're spending.