Riverflow 2.0 Pro

Riverflow 2.0 Pro is Sourceful's text-to-image model, offered here as a cloud node - and its hook is the size presets. Where most image nodes hand you a raw width/height box, this one gives you a size dropdown of 34 options covering 1K/2K/4K across every aspect ratio, including "from input" modes that inherit resolution from a reference image. If your workflow is "generate at 1K, then decide later," this skips the resolution planning entirely.

It sits in the Runware/Image category, which makes it one of the plain, get-a-picture nodes in this pack: you prompt, it renders, you get an IMAGE. The "Pro" positioning plus the 4K options suggests the premium tier of the model, but honestly the settings surface is simple enough that you don't need to care about the architecture - you care that the size dropdown does the work.

What you set

positivePrompt is required. The rest that matter:

• size - the star input. 1K (16:9), 2K (from input), 4K (4:3)... choose a target; only one resolution/dimension gets sent to the API.
• referenceImages - IMAGE input for style/content reference if the model supports it.
• superResolutionReferenceImages - a second image socket that hints the super-resolution stage, which pairs naturally with the 2K/4K size options. This is the input that makes "upscale-friendly generation" concrete.
• numberResults - up to 20 per run, each a different seed. Twenty is also twenty billed generations, so be deliberate.
• watermark - optional RUNWARE_WATERMARK builder socket if you need a branded overlay.

Output is image (IMAGE) → straight into Preview Image or Save Image. outputFormat (JPG/PNG/WEBP) and outputQuality control the file side.

Installing

Part of the Runware/ComfyUI-Runware pack - one install gets the whole catalog.

ComfyUI Manager: search Runware, install, restart.

Manual:

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

No model files - the deps are just runware-sdk, pillow, soundfile. You need a Runware API key from runware.ai/api-keys (Settings panel or RUNWARE_API_KEY).

Where people get burned

The honest caveat is that Riverflow is not a household name like the Qwen or Wan models in the same pack, and "2.0 Pro" branding on a lesser-known model is exactly the kind of thing that can overpromise. Treat the first run as a test: check that the quality is actually better than your free/local baseline before you spend real money on 4K batches. And remember the super-resolution inputs only do their thing if you actually wire an image in - an empty superResolutionReferenceImages socket is just an unused feature.