RedNode Rig Out

Some people run Krea 2 locally and NovelAI somewhere else, and hate keeping two sets of settings in sync. RedNode Rig Out is the bridge that ends the double-entry bookkeeping: it pulls everything an external engine needs straight out of your RedNode Studio Workspace - prompt, negative, seed, denoise, steps, CFG, sampler, scheduler, and the canvas size - and hands it over as plain values you wire into your NovelAI (or any) chain. You keep editing in one place, the workspace, and the outside engine just obeys.

The neat part is that the workspace doesn't have to be wired to this node at all. The node finds the Studio Workspace in the graph on its own - the same discovery trick the pack's Paint nodes use - and reads the rig you name. So you can have a big tabbed panel driving things and a tiny "Rig Out → external engine → Rig In" detour off to the side. The graph still reads left to right; the workspace just sits there as the source of truth.

Which rig, and the inputs that matter

Rigs live on the workspace's Models tab - named sets of model, seed and sampler settings, with an Active choice. The rig input defaults to (active rig), which follows that choice; type a rig's name and it becomes a named cockpit. There's a deliberate loudness here: if you typo a rig name, the node prints a warning and falls back to the active rig rather than silently using a different prompt. That's the failure mode the author decided had to be unmissable.

The image input is the mode switch, and it's the one thing to understand before you use this:

• Nothing wired - this is a fresh render. Denoise comes out 1.0, because an image-to-image with no image has to be, and the canvas size comes from the workspace's Latent tab.
• An image wired in - this is i2i. The denoise output carries the rig's own strength dial, and the picture rides through to the image output so the chain stays connected.

The outputs

Eleven of them, and they're all exactly what the label says: prompt and negative (with wildcards already rolled against the rig's seed), seed, denoise, steps, cfg, sampler and scheduler as text, width and height snapped to multiples of 8, and image. The sampler and scheduler come out as strings precisely so an external engine can read them - an External rig on the Models tab is the natural partner, since it has a free-text sampler and its own denoise dial, but any rig's numbers work.

Install

One pack, one install - and unlike the camera nodes, nothing here needs Krea 2:

git clone https://github.com/RedNodeAI/ComfyUI-RedNodeStudio.git ComfyUI/custom_nodes/ComfyUI-RedNodeStudio

Restart ComfyUI, or search RedNode Studio in ComfyUI Manager. No pip dependencies.

Honest caveats

This node outputs numbers and text - it doesn't call any API and needs no key. The actual render still happens in the external engine, so "out" here means "out to whatever you paste these into." That's the point, but it's worth saying plainly so nobody expects a one-click NovelAI render. The other thing to know: your external engine has to accept these fields as typed values, which is why the outputs are split per-value rather than one fat JSON blob. Set up the bridge once, wire the sampler/scheduler strings in, and the workspace stays the single place you ever touch.