CLIP Text Encode - Advanced

You know how the basic workflow needs two separate CLIP Text Encode nodes - one for the positive, one for the negative - and then you have to add a CLIP Set Last Layer node if you want clip skip? This node collapses all of that into one box: both prompts, a clip skip dial, and a choice of prompt-weighting syntax, all in a single node that outputs both conditionings.

Why you'd reach for it

If you're coming from A1111, this is the most familiar feeling text encoder in this pack. It takes your positive and negative side by side like the old webui did, it understands A1111's (word:1.2) weighting if you pick that interpretation, and the clip skip field is right there instead of being a separate node. For SD 1.5 anime models - the Pony/Illustrious line where clip skip 2 is basically the house style - that's a real convenience. See the clip-skip conventions in this KB before you fiddle: on SDXL the dial mostly does nothing because the encoder already runs at the penultimate layer.

How it works

This is where the pack does something clever - it bundles in a copy of the old Efficiency Nodes pack as a git submodule and delegates the encoding to it. So you're getting the battle-tested encode_prompts machinery: proper handling of A1111 vs compel weight syntax, clip skip, and both prompts in one call. That also means it does its own prompting-heavy lifting, which is why it can offer the weight_interpretation dropdown that plain ComfyUI encoders don't have.

Inputs and outputs

The few you'll actually touch:

• clip - a CLIP model (forced input, from your checkpoint loader).
• POSITIVE_PROMPT / NEGATIVE_PROMPT - the two prompts, multiline text boxes.
• clip_skip - INT, default -1 (no skip). ComfyUI counts negative: A1111's clip skip 2 = -2 here.
• weight_interpretation - comfy, A1111, compel, comfy++, or down_weight. A1111 if you're porting prompts with (weight:1.3) syntax, comfy if you only use (word:1.3) native syntax.

Outputs: POSITIVE and NEGATIVE conditioning, straight into your sampler.

Installing it

Part of "quadmoon's ComfyUI nodes". Install via ComfyUI Manager (search the pack title or Install via GIT URL with https://github.com/traugdor/ComfyUI-quadMoons-nodes.git), or:

cd ComfyUI/custom_nodes
git clone https://github.com/traugdor/ComfyUI-quadMoons-nodes.git

Then restart ComfyUI. No models to download - but note the dependency: the encoding is borrowed from a bundled copy of Efficiency Nodes, pulled in as a git submodule. If you cloned by hand and this node errors out with an import failure, the submodule didn't come along. Fix it inside the pack folder:

git submodule update --init --recursive

Common issues

Two things bite people. First, the submodule: a plain git clone doesn't always fetch it, and then this node dies on load-time import errors while the rest of the pack works fine. Run the submodule update command above. Second, the clip skip: don't blindly set -2 on everything. It matters on SD 1.5 anime checkpoints; on SDXL it's already effectively at 2, and on modern LLM-encoder models (Flux-era and beyond) the parameter does nothing at all - there's no CLIP layer to skip.