zhenzhen-vidu-q3-video-lowprice

If the Vidu Q3 short-play node is the specialist, Comfly_vidu_q3_video_lowprice is the generalist: one node, thirteen vidu-q3-* models in the dropdown, covering every direction Vidu Q3 generates - text-to-video, image-to-video, first-and-last-frame, reference-to-video, mix, and ad variants, each in pro, turbo, and pro-fast tiers. Whatever clip you're trying to make, there's probably a model here for it.

It runs on Zhenzhen's domestic storefront (api.seedance.nz), so it's part of the cheap "lowprice" family and shares their API-key plumbing. Everything is a cloud call - no weights on your disk.

How it works

Choose your model, write a prompt, pick duration/ratio/resolution, connect up to nine reference images if the model takes them, and the node submits the job and polls until the video's ready. You get video, video_url, task_id, and the response JSON out. The model choice does the real work:

• t2v models (vidu-q3-pro/turbo/fast-t2v) - pure text-to-video, no references needed.
• i2v models - animate one or more reference images; connect them to image1–image9.
• start-end models - first-and-last-frame video, your two keyframes.
• r2v / mix / ad - reference-driven and advertising flavors.

The settings that matter: seconds runs 4–15, and longer clips cost more. ratio has adaptive plus the usual 16:9 / 4:3 / 1:1 / 3:4 / 9:16 / 21:9 - adaptive follows whatever your reference images imply. resolution defaults to default (the model's own pick) with 720p and 1080p available. seed defaults to -1, which means random; set a fixed value to reproduce a run.

Installing and setting up

Install through ComfyUI Manager (search Comfyui-zhenzhen) or:

cd ComfyUI/custom_nodes
git clone https://github.com/T8mars/Comfyui-zhenzhen

Restart ComfyUI. No model files. This node takes no direct API key - connect the Comfly Seedance2 Low Price Settings node's api_config output, or set SEEDANCE_API_KEY in the environment. Same config node feeds every node in this low-price family, so set it once.

Troubleshooting

The config wiring is the number-one stumble: run this without an api_config connection and it dies with an "empty api_key" error - the settings node isn't optional here. If a specific model errors out, try a different tier: the fast models are cheaper and quicker but the pro line holds up better on complex motion. And remember seed of -1 means "surprise me" - if you're iterating on a prompt and want to see it again, pin the seed. skip_error is available for batch runs where one failure shouldn't kill everything.

One honest tip: with thirteen models in a dropdown, people tend to overthink the choice. Start with vidu-q3-turbo-i2v for a single reference image, get a feel for the output, then branch into start-end and r2v once you know what you're missing.