Text to Conditioning

Here's the problem this solves. All the WAS text nodes - Multiline, Concatenate, Find and Replace, Load Line From File - produce a STRING. But your KSampler doesn't eat strings; it eats CONDITIONING. The core CLIP Text Encode node has its prompt baked into a widget and won't take a text wire. Text to Conditioning is the bridge: give it a CLIP and a text string, and it hands back conditioning you can plug straight into the sampler.

That's the node that makes the whole WAS text toolkit actually usable for prompting. Without it, all that text plumbing has nowhere to go.

How it works

It's essentially a CLIP Text Encode that accepts its prompt as an input wire instead of a typed widget. Per the README: "Convert a text string to conditioning." Under the hood it runs your string through the CLIP encoder exactly the way the standard encode node does - same math, same result - but because the text arrives on a wire, anything upstream can generate or transform that text first.

This is what unlocks dynamic prompting. Load a prompt from a file, replace a token, concatenate a style, then encode - all in the graph, with the encode as the last step. The core encoder can't do that because you can't wire text into it.

Inputs and outputs that matter

Two inputs matter: the CLIP (from your checkpoint loader - same CLIP you'd feed a normal encode) and the text STRING (from any WAS text node or a Text Multiline). The output is CONDITIONING, which goes into your KSampler's positive or negative input like any other encoded prompt.

Wire it once for positive, once for negative if you're building both prompts dynamically. It behaves like the standard encode from the sampler's point of view - the sampler can't tell the difference.

How to install it

ComfyUI Manager: search was-node-suite-comfyui, install, restart. Manually:

cd ComfyUI/custom_nodes
git clone https://github.com/WASasquatch/was-node-suite-comfyui

then install the pack's requirements.txt in your ComfyUI venv and restart. It uses the CLIP model you already loaded - no extra downloads.

Common issues & troubleshooting

My output looks different from the core CLIP Text Encode. It shouldn't, materially - it's the same encoding. If results differ, check you fed the same CLIP (from the same checkpoint) and the exact same text. A subtle text difference (a trailing token from a concatenation, a stray space) is the usual culprit, not the node.

It wants a CLIP and I don't have one wired. You need to feed it the CLIP output from your checkpoint loader (WAS's or core's). Conditioning can't be made without the text encoder, so the CLIP input isn't optional.

Weighting/embedding syntax - does it respect it? It's a text encode, so standard prompt weighting works. But advanced features from other packs (BlenderNeko's advanced encode, special attention tricks) aren't part of this node - it's the plain path. For those, use the dedicated encoder that supports them.

Do I even need it? Only if you're generating or transforming prompt text elsewhere in the graph. If you just type a static prompt, the core CLIP Text Encode is simpler and one fewer dependency. Text to Conditioning is the payoff for building prompts out of the WAS text nodes - that's when it's essential.

Whole pack won't import after a ComfyUI update. The standard WAS Node Suite issue - package downgrades colliding with a ComfyUI bump. Re-run the pack's requirements against the activated venv, or install.bat. Maintenance-only since late 2023.