Combine CLIP Text Encode

This is the discrete-fields sibling of Sage_CombineCLIPMultilineTextEncode. Where that node splits one multiline text box by newline, this one gives you an expandable list of separate text inputs - click to add another box, each one gets its own field, each one gets encoded through CLIP individually and merged into a single conditioning. Same underlying idea (encode ideas separately rather than concatenating them into one long string and hoping CLIP handles the mix gracefully), different UI for building the list.

Which of the two you reach for is mostly a workflow preference. If you're pasting in a block of text where each line is naturally its own idea, the multiline version is less friction. If you're building prompt pieces from separate sources - say, a style snippet from one place and a subject snippet from another - having them as genuinely separate fields you can individually clear, reorder, or swap out is nicer here.

Inputs and outputs

• clip - required, the CLIP model to encode with.
• texts - the expandable list itself. Add as many text fields as you need; each becomes its own encoded unit before merging.
• operation - a dropdown, described by its tooltip as choosing "whether to scale the combined text conditioning." Same mechanism as the operation dropdown on the multiline version - worth testing on an actual prompt rather than guessing at the effect from the name alone.

Output is a single conditioning, the merged result of every text field, ready to wire straight into a sampler.

Installing it

Through ComfyUI Manager: search Sage Utils, install, restart. Manually:

cd ComfyUI/custom_nodes
git clone https://github.com/arcum42/ComfyUI_SageUtils
cd ComfyUI_SageUtils
pip install -r requirements.txt

Restart ComfyUI afterward. No model downloads specific to this node.

Common issues

An empty text field in the list behaves differently depending on how you've built your graph - if you added a field and never filled it in, don't assume it's silently ignored the same way a blank line is on the multiline version; check your actual output if a merge feels off with unfilled fields left in the list.

If the combined conditioning feels stronger or weaker than expected, that's the operation setting - try flipping it before assuming a wiring mistake. And since this is a small, actively maintained but single-developer pack, an unusually large number of text fields in the list is more likely to be genuinely untested than to be a documented ceiling - if you're stacking a dozen-plus fields and something looks off, that's worth a GitHub issue rather than assuming user error.