☁️BizyAir Conditioning (Combine)

This one gets confused with ConditioningAverage constantly, and it's worth being precise about the difference up front, because they solve different problems. Average blends two conditionings into one by weight. Combine doesn't blend anything - it just puts both conditionings into the same list and hands them to the sampler as two separate, simultaneously-active entries. Each keeps its own full-strength embedding intact; nothing gets diluted or averaged down.

Why that distinction matters in practice. If you want a smooth crossfade between two ideas - "70% this style, 30% that one" - you want Average, because it gives you a strength dial. If you want to add a concept on top of an existing prompt without touching the original's encoding at all - say, a base scene description plus a separate "style: watercolor" conditioning that you encoded independently - Combine is the right tool, because both pieces stay at full strength and the sampler just sees more conditioning to satisfy, rather than a mathematically diminished version of either one. It's also handy for going past a text encoder's practical prompt-length comfort zone: instead of cramming everything into one long string and hoping the encoder handles it gracefully, you encode two shorter, focused prompts separately and combine them here.

Inputs that matter: conditioning_1 and conditioning_2, both required, both BIZYAIR_CONDITIONING - meaning both need to trace back to a BizyAir CLIP/text-encode chain, not a stock ComfyUI one. There's no strength or weight parameter at all, which is the whole point - this node doesn't have an opinion about balance, it just merges the lists.

Output: one BIZYAIR_CONDITIONING, containing both inputs as separate entries. Wire it wherever you'd wire any other conditioning - into a sampler's positive or negative slot, or into another conditioning node further down the chain (nothing stops you from combining a combine).

One thing worth knowing before you lean on this a lot: stacking multiple simultaneous conditionings isn't free in the sense that "more prompts on the same canvas" doesn't always read as "more control." Depending on the base model, two unrelated conditionings competing for the same generation can produce a muddier result than one well-written prompt would - this node gives you the mechanism, but writing one good combined prompt is often still the simpler path unless you have a specific reason (length, modularity, wanting to swap one half without re-touching the other) to keep them separate.

Install: one pack install covers this node - search "BizyAir" in ComfyUI Manager and click install, or cd ComfyUI/custom_nodes && git clone https://github.com/siliconflow/BizyAir.git and restart, or comfy node install bizyair on the Comfy CLI. As with every node in this pack, it's a cloud call under the hood, so you need a registered BizyAir account with your API key set before it (or anything feeding it) will run.

Troubleshooting: if the combined result doesn't seem to reflect both prompts the way you expected, remember there's genuinely no weighting here - if one idea is dominating, that's the underlying model's own balance between the two conditionings, not something you can fix by adjusting this node, since it has nothing to adjust. If you want that kind of control, that's Average's job, not Combine's. And as always with this pack: a stalled or vaguely-failed run is worth checking your API key over before you go hunting through the graph for a logic error.