☁👑 Add Lora (API ONLY - CivitAI) 👑☁

This is the companion to the Civitai API image generator, and it does exactly one thing: it adds a LoRA to a cloud generation. The "API ONLY" in the name is the important part - this doesn't touch a local LoRA file. It builds a reference to a Civitai-hosted LoRA (by its AIR identifier) and hands that to the API node, which applies it server-side when it generates. If you're generating locally, this node isn't for you; it only makes sense wired into the Civitai API generator.

Chain several of these together and you build up a LoRA stack for the API call, the same way you'd stack local LoRA loaders - except the whole stack lives in the cloud request.

How it works

You give it a LoRA's URN and a strength, and it emits an add_LORA payload. That payload plugs into the Civitai API generator's add_LORA input. To stack more than one, feed this node's output into the next Add Lora node's optional add_LORA input and so on - each link appends another LoRA to the chain before it reaches the generator. The README's own demo runs the same seed with and without the LoRA to show the effect cleanly, which is the sane way to verify a LoRA is doing what you think.

The inputs and outputs that matter

• lora_urn (STRING, required) - the AIR identifier of the Civitai-hosted LoRA, e.g. the default urn:air:flux1:lora:civitai:790034@883473. Copy this from the LoRA's version page on Civitai.
• strength (FLOAT, default 1.0, range 0–2) - how hard the LoRA is applied. Same intuition as a local LoRA strength: 1.0 is full, dial down if it's overwhelming the base.
• add_LORA (optional input) - connect a previous Add Lora node here to chain multiple LoRAs into one stack.

Output: add_LORA - wire it into the Civitai API generator's add_LORA input (or into the next Add Lora node to keep stacking).

How to install it

ComfyUI Manager: search Bjornulf_custom_nodes, install, restart. Manual:

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

then restart. It travels with the Civitai API stack, which needs the civitai-py dependency from the requirements file.

Common issues & troubleshooting

Nothing happens / the LoRA has no effect. This node produces a reference, not an image. Its output must reach the Civitai API generator's add_LORA input, or it does nothing. Confirm the wire actually lands there.

"LoRA not found" from the API. The lora_urn is wrong or points at a version the API can't serve. Copy the full AIR string from the LoRA's version page - the number after @ is the version ID and has to be right.

Base/LoRA mismatch. A Flux LoRA won't work on an SDXL checkpoint and vice versa. Make sure the LoRA's URN base type matches the model_urn you set on the generator - the default LoRA URN here is a Flux one, so pair it with a Flux model.

It costs Buzz and can be flaky. Same caveats as the generator: API generations spend Yellow Buzz and Civitai's API is not the most reliable rail. Verify a LoRA cheaply (small size, few steps, fixed seed with/without) before committing to a big batch.

I want to stack three LoRAs. Chain three Add Lora nodes, each feeding the next via the add_LORA input, then the last one into the generator.