ImagineArt 1.5 Pro

If ImagineArt 1.5 is the economy seat, 1.5 Pro is premium economy - same route, nicer ride, slightly higher fare. This node is the higher-fidelity tier of Runware's in-house image model, and the schema is deliberately boring: prompt in, IMAGE out. The interesting part isn't what you configure; it's what you don't have to.

Pro exists for the jobs where the base model's speed leaves quality on the table. You'd reach for it when a prompt matters - a client-facing concept, a hero image, something you're going to actually show people - and you want more resolve than the budget lane gives without moving up to the reasoning-heavy 2.0.

How it works

The only required input is positivePrompt. The optional set is nearly identical to the base 1.5: width/height (1024² default), numberResults (up to 20 seeded variations), safety.checkContent (server-side moderation, surfaced as NSFW: yes/no in the title bar), and outputFormat/outputQuality for the file encoding.

There's no seed, no negative prompt, no CFG - Runware's model nodes expose the model's real parameters, and these models simply don't take them. That's not a limitation to work around; it's the product. You prompt, you pick from variations, you move on.

The mechanism is the same as every node in this pack: your prompt goes over REST through the runware-sdk, the generation happens on Runware's cloud, and the result comes back as a native IMAGE tensor you can drop into Save Image or pipe into an upscaler.

Installing it

Official pack, standard install:

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

Restart, set your API key in ComfyUI Settings → Runware API key or via RUNWARE_API_KEY. Dependencies are just runware-sdk, pillow, and soundfile - no checkpoints, no downloads.

Where people get burned

The main trap is paying Pro prices for economy work - you don't need the better tier for a 100-image brainstorming dump, and this pack makes it so easy to switch nodes that people forget they did. Compare a couple of outputs head-to-head before committing a batch. The other gotchas are pack-wide: no API key means instant errors, and it's a metered API on a top-up account, so keep numberResults sensible. And if you keep wanting more control - reference images, reasoning, a real seed - that's the 2.0 node's territory.