SF WaveSpeed Sora 2 Text to Video Pro

The plain Sora 2 node gives you 720p and two aspect ratios. SF WaveSpeed Sora 2 Text to Video Pro is the same model with the training wheels off: you get a 720p and a higher-resolution tier (1024×1792 portrait, 1792×1024 landscape), plus the freedom to pick a standard or pro size on every clip. When the deliverable is a client mockup, a YouTube thumbnail story, or anything that's going to be watched big, this is the node you reach for.

It's not a different model - it's the resolution-and-pricing tier up. The prompt guidance is identical to the base node (scene, style, camera, audio cues), the durations are the same 4/8/12 seconds, and the cost varies by resolution and length. That last bit is the thing to keep an eye on, because pro-resolution clips are where the bill starts to feel real.

It's part of the comfyui-stillfront pack, under Stillfront/WaveSpeed.

How it works

Same mechanics as its sibling: your prompt, size, and duration go to WaveSpeed's openai/sora-2/text-to-video-pro endpoint, and the node returns the first output video URL. All generation happens on the hosted side.

The inputs that matter

• size - four options now: the standard 720*1280 and 1280*720, plus the pro tier 1024*1792 (portrait) and 1792*1024 (landscape). The tooltip frames it as Standard vs Pro resolution, so the choice is basically "cheap draft" vs "presentable output."
• prompt - same structured brief as the base node: scene, style, camera movements, audio cues.
• duration - 4, 8, or 12 seconds; pricing "varies by resolution and duration," which is the author being polite about the fact that pro-res 12-second clips are the expensive end.
• enable_sync_mode - default off, same polling behavior as the other Sora node.

Output is a single video_url STRING - wire it to a URL-fetching video loader/display, or capture it with a text node.

Installing it

Pack install plus WaveSpeed key:

cd ComfyUI/custom_nodes
git clone https://github.com/razvanmatei-sf/comfyui-stillfront
cd comfyui-stillfront
pip install -r requirements.txt

Restart, then wire in an SF WaveSpeed Client node with your key (or config.ini / WAVESPEED_API_KEY).

Troubleshooting

• Sticker shock on the bill. Pro resolution at 12 seconds is the premium tier for a reason. Work out your motion and composition at standard 4-second runs, then render the final at pro.
• Pro clips slower than expected. More pixels, more inference, more waiting. The async default hides this behind a poll; check the console for the task id if it seems stuck.
• Prompt handling identical to base Sora. Same strengths, same failure modes - over-ambitious prompts get partially followed. Rank your asks, keep camera directions explicit.
• URL output, not a VIDEO object. Same as the sibling - you'll want a URL-based preview/loader in the workflow.

If you're comparing this to the base node: use the base one for drafts and iteration, this one for anything that leaves your machine. Same model family, one tier of resolution and cost above - which, for video that's going to be judged, is usually worth it.