Wan 3.0 Pro Prime · Text to Video

Let me give you the take first, because it's the useful part: for prompt-only work this is rarely the node you want. It has the thinnest input set in the pack - a prompt, a resolution, an aspect ratio, a duration, audio on and expansion on - and the steepest per-second price in the Wan family. There's nothing here to amortise that cost. No reference to hold an identity, no frame to anchor the composition, no document to carry the detail.

Which isn't to say it's pointless. When you need the best-looking thirty seconds you can get from a sentence and nothing else, and you need it back this afternoon, it's the top rung. Base Wan 3.0 and Prime exist for the ninety-nine percent of the time when you don't need that.

Open weights? No. Wan 3.0 is API-only, like everything after 2.2 - the community made its peace with that in December 2025.

How it works

Generated from the model's published schema, so the widgets and their option lists are the endpoint's. Queue and the pack quotes the run before credits are held, submits the task under an idempotency key, prints progress and the estimate on the node, then downloads the finished MP4 as a ComfyUI VIDEO.

The two defaults that shape the output before you touch anything: generate_audio on, and enable_prompt_expansion on. That means an on-by-default rewrite pass between your words and the model, plus a generated soundtrack you might not want. Both are worth a conscious decision rather than inheriting.

The inputs that matter

prompt - subject, action, camera, light, pacing, boundaries. With no other input to lean on, the prompt has to do all the work, and the "boundaries" part matters most at long durations: thirty seconds is enough time for a model to invent a second scene you didn't ask for.

duration_seconds - 2 to 30, or -1 for smart duration, where the model reads the prompt and picks the length. On a node this expensive -1 is genuinely attractive: it stops you paying for seconds of drift after the action is over.

resolution - 1080p, 2k or 4k. Native rendering tops out at 1080p and the higher tiers are enlarged - the tooltip says so plainly. aspect_ratio - 16:9, 4:3, 1:1, 3:4, 9:16, or adaptive; the default is adaptive, which for a text-only request means the model infers framing from your prompt.

There's no negative prompt field on this node, so anything you want excluded has to be written in as a positive statement.

Outputs: video and task id.

Installing it

Manager > Custom Nodes Manager > search PopNow > Install > restart, or:

cd ComfyUI/custom_nodes
git clone https://github.com/popnow-ai/popnow-comfyui

Create a key at popnow.ai/api/console/keys, paste it into Settings > PopNow or export POPNOW_API_KEY, then restart once more so the node list is rebuilt from your account's catalog. No model downloads, no dependencies beyond ComfyUI, no GPU.

Delete the legacy custom_nodes/popnow_api.py if it's still there.

Where people get burned

Both defaults. Prompt expansion rewrites your brief - sometimes better, often further from what you meant, and on an expensive node you feel it. Audio is included whether you planned to use it or not.

Then the bill, which on this node is the whole story: thirty seconds at 35.1 credits per second is about 1,050 credits, a little over a dollar for one click, and nothing about a text-only workflow stops you from queueing twenty of those. Set Settings > PopNow > Max credits per run and let the quote gate refuse anything above it. Draft on a cheaper tier; come here for the final.

Failure codes: 40201 no credits, 40004 unsupported settings combination with the field named, 400 a rejected value, 40310 verify your email. Refusals come back as content_rejected and are refunded. And interrupting a run doesn't cancel it - it keeps generating server-side and charges if it succeeds.