☁️BizyAir CLIP Text Encode (Prompt)

What it is

This is the plainest, most-used node in the whole pack once you're actually building workflows: BizyAir's cloud version of the standard CLIPTextEncode, which turns a text prompt into the conditioning a sampler actually uses. If you've built even a single ComfyUI workflow before, you already know this node - positive prompt in one instance, negative prompt in another, both feeding a sampler. The only difference here is that the CLIP encoding runs on BizyAir's servers, not your own machine.

Use this one for any model that expects a single-string prompt through a single text encoder - most non-Flux checkpoints you'd load through BizyAir. Flux specifically wants its own node, BizyAir_CLIPTextEncodeFlux, because Flux splits conditioning across two separate encoders (CLIP-L and T5-XXL) rather than accepting one string.

How it works

Text goes in, CLIP encodes it into an embedding, that embedding comes out as conditioning. There's no other logic here - no guidance parameter, no dual-encoder split, just the encoder doing what CLIP text encoders have always done: turning your words into a representation the diffusion model can attend to during sampling.

Inputs and outputs that matter

Two required inputs, and that's the entire node: text, a multiline string for your prompt, and clip, typed BIZYAIR_CLIP - the output of a BizyAir CLIP or model loader upstream, not a stock ComfyUI CLIP socket. There's nothing else to configure; this node does one job.

The single output is CONDITIONING, typed BIZYAIR_CONDITIONING, which feeds into a BizyAir sampler's positive or negative conditioning input - same role as stock ComfyUI's conditioning output, just wearing a BizyAir-specific type so it only connects to other BizyAir nodes.

Installing it

Search "BizyAir" in ComfyUI Manager and install, or clone manually: cd ComfyUI/custom_nodes && git clone https://github.com/siliconflow/BizyAir.git, then restart ComfyUI. Every node in this pack, including this one, needs a BizyAir API key set up first - the login prompt on first launch walks you through it.

Common issues

The most common trip-up is architecture mismatch in the other direction from what you'd expect - if you loaded a Flux checkpoint through a BizyAir loader and try to encode your prompt with this plain node instead of BizyAir_CLIPTextEncodeFlux, you'll get conditioning that doesn't match what Flux was trained to expect, since Flux wants its guidance and dual-encoder split handled explicitly. For every other architecture loaded through this pack, this is the correct, default text-encode node, and if it's not connecting to your sampler, the usual cause is a stray non-BizyAir node somewhere in the chain - BIZYAIR_CLIP and BIZYAIR_CONDITIONING don't accept or produce stock ComfyUI types, so the whole chain from loader to sampler needs to stay inside this pack's node family.