Nourivex Image Upscale With Model

Let's be direct about what this is: a thin wrapper around ComfyUI's built-in ImageUpscaleWithModel, given a Nourivex name so it sits in the same namespace as the rest of the pack. It takes an upscale model and an image, and it returns the image, bigger. That's the whole contract, and the wrapper doesn't add any logic - under the hood it resolves the core node and calls it.

Why a wrapper at all?

Because the pack's author made a deliberate call: every I/O node in a Nour_Comfy workflow is a "Nourivex" node, so a workflow that ships with the pack references nothing outside it. That's a genuinely defensible philosophy. The most common reason a downloaded workflow fails to run is a missing node you never heard of, and keeping the I/O self-contained shrinks that problem to "install this one pack." The cost is that you now have two nodes doing the same job, and you only need this one if you're already running the pack for its other nodes.

What you feed it

• upscale_model - a loaded upscale model, from the pack's own Nourivex Upscale Model Loader or from ComfyUI's core UpscaleModelLoader. They both emit UPSCALE_MODEL, so they're interchangeable at this socket.
• image - the IMAGE you want enlarged.

One IMAGE comes out, ready for a preview, a save, or the pack's Image Route Toggle.

The honest expectations

This is a more pixels upscaler, not a more detail upscaler - the first rung of the upscaling ladder, not the last. An ESRGAN-family model like 4x-UltraSharp (which the pack's default workflow uses) enlarges the image cleanly in a few hundred milliseconds and adds nothing that wasn't there. It will not invent pores, fix a soft-focus face, or rescue a blurry eye - that's the job of a detailer pass or a generative upscaler, and this node isn't one. Where it shines is turning a 1024×1024 render into a 4x image that keeps its sharpness, which is exactly the "upscale after you've settled on the composition" step the pack's toggle workflow is built around. Do your detailing at native resolution first, then upscale - the community's standing advice, and it applies here unchanged.

One thing to know: the node doesn't download anything. The upscale model has to already live in models/upscale_models, and the model_name dropdown on the loader only lists what's there. If your dropdown is empty, that's a missing model file, not a broken node.

Install

Same pack, same one-liner:

cd ComfyUI/custom_nodes
git clone https://github.com/Nourivex/Nour_Comfy

Restart ComfyUI (or install Nour_Comfy via ComfyUI Manager). No Python dependencies, no extra models - just bring your own upscale weights.