ScarlotSoft Combine Conditioning
Two conditioning inputs, one combined output — ScarlotSoft Combine Conditioning
- conditioning_1
- conditioning_2
- CONDITIONING
Sometimes the cleanest way to prompt is to not prompt once. Split a scene into two encoded conditions - one for the subject, one for the background or style - then glue them together before the sampler. That glue is ScarlotSoft Combine Conditioning: two CONDITIONING sockets in, one CONDITIONING out. Under the hood it does the same thing ComfyUI's core ConditioningCombine does: a Python list addition of the two conditioning objects. Zero mystery, zero cleverness, and that's fine - this is a convenience wrapper with a friendlier name and the suite's theming.
Why would you bother? Conditioning in ComfyUI is a list of [tokens, metadata] pairs, and adding two lists concatenates them so the sampler attends to both sets of text. That's the whole trick behind "main subject from this encoder, everything else from that one." It matters when you have two CLIP encodes you want to keep separate - say a CLIPTextEncodeModular for the subject and a plain one for lighting/style - because joining them at the end is cheaper to read and easier to toggle than cramming everything into one box.
The schema is two inputs, one output, nothing else: conditioning_1, conditioning_2, and the combined CONDITIONING. Wire the result into a KSampler's positive or negative socket. One thing to keep in mind from the prompt-engineering playbook: combining two conditionings does not insert a separator between them - it's a flat concatenation, so the token streams run together. If your two halves are long, remember the SDXL CLIP models have a hard token ceiling and things past it get dropped; combining doesn't raise that ceiling, it just adds lists.
Install with the whole pack - ComfyUI Manager, search ScarlotSoft, install, restart:
cd ComfyUI/custom_nodes
git clone https://github.com/scarlotsoft/ComfyUI_ScarlotSoft
No dependencies. Dark theme included; hard-refresh after a UI-mode switch if the canvas looks off.
Be honest about the trade here: if you're not otherwise using this pack, there's no reason to install it for this node - the core ConditioningCombine is identical in behavior. This one exists so ScarlotSoft workflows stay self-contained and consistently named, and if you're following along with a workflow from the pack it's what you'll see between your two encoders and the sampler. When it's wired up and the sampler runs, it's doing exactly what it looks like: turning two prompts' worth of conditioning into one. That's all a combine node should ever be.
Inputs (2)
| Name | Type | Default | Description |
|---|---|---|---|
| conditioning_1 | CONDITIONING | — | |
| conditioning_2 | CONDITIONING | — |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| CONDITIONING | CONDITIONING | — |