Text Combiner 6

This is Text Combiner 2 with room to breathe: six text inputs joined into one string. Where the two-input version is for a quick merge, this one is built to hold an entire structured prompt, one concern per slot, so you can rewire any part without touching the rest.

That structure is the actual value. When you prompt an anime model, a good prompt is really several independent groups - quality tags, scene, pose, character description, LoRA triggers, extras - and you want to iterate on them separately. Six labelled slots let you keep each in its own upstream node and swap them like Lego. The author even suggests a naming convention so a workflow reads clearly at a glance.

How it works

You feed text1 through text6, and the node concatenates all six into a single text output. Any slot can be a typed literal or wired from another node, and empty slots just contribute nothing. So you set up your graph so each input carries one category of the prompt, and the combined output goes to your CLIP Text Encode.

The inputs that matter

All six are strings; the skill is in how you assign them. The author's suggested convention, which is a genuinely sensible way to keep a big prompt legible:

• text1 - common/model-required tags, e.g. your quality prefix.
• text2 - base scene info: camera angle, style, location, character count.
• text3 - the character's pose and expression.
• text4 - the character's appearance/species details.
• text5 - LoRA trigger words.
• text6 - anything extra.

Output is a single text. Nothing forces this mapping - the node just concatenates in order - but adopting it means six months later you still know which box holds what.

Where it fits

This is the assembly point at the end of a modular prompt graph. Upstream: a Text Box or selector per concern. Here: they all merge in a fixed, readable order. Downstream: the encoder. Because order is stable and each slot is independent, you can A/B a single aspect (swap the pose input, leave everything else) - which is exactly the kind of controlled iteration that's painful with one monolithic text field.

Install

ComfyUI Manager → search ComfyUI_Mira → Install → restart, or cd ComfyUI/custom_nodes && git clone https://github.com/mirabarukaso/ComfyUI_Mira.git and restart. If the pack errors on load, pip install -r requirements.txt in the pack folder. No downloads.

Common issues

Same trap as every text combiner: it doesn't insert separators for you. Six tag groups joined without trailing commas can weld into malformed tags at the seams, so end each fragment with , . The other thing is discipline - the six-slot layout only stays useful if you keep to a convention; if you dump random tags into whichever box is nearest, you've just rebuilt one big text field with extra steps. Pick a mapping (the author's is fine) and stick to it.