GALIAIS-Nodes Prompt Weight

This is the thinnest node in the whole GALIAIS pack, and it knows it. The Prompt Weight takes a chunk of text and wraps every tag in it in (tag:weight) syntax. One input in, one string out, no surprises. It's a convenience wrapper for something you could type yourself - and the honest review is that's exactly what it's for: not getting you to a feature you couldn't otherwise reach, but saving you from hand-editing a twenty-tag string when you want to nudge the whole block up or down.

What you set

• 内容 - the tags or text to weight.
• 权重 - a float from 0 to 5. At exactly 1 it outputs the text untouched. Move it to 1.3 and every tag becomes (tag:1.3). Drop it to 0 and the output empties out entirely - which is technically "weighting" the block to nothing.

The single output is 加权提示词, the joined result.

The caveat that matters more than the node

Whether this node does anything useful is entirely up to your text encoder. The (word:1.3) attention-weight syntax is a CLIP-era convention - it works on SD 1.5, SDXL, and the whole SDXL anime lineage (Illustrious, NoobAI, Pony). On LLM-encoded models like Anima or Flux, the weight syntax is silently discarded, and this pack's sibling nodes are explicitly tuned for Anima-style output ordering. So on an Anima target, this node is a no-op dressed up as a transformer: the raw tags still pass through fine, the weights just evaporate.

Use it when you're on an SDXL-lineage checkpoint and want to emphasize a whole section without retyping it. Don't use it as your reason to buy into the pack - the interesting stuff is the composer and the diagnostics.

Installing

cd ComfyUI/custom_nodes
git clone https://github.com/GALIAIS/GALIAIS-Nodes.git

Restart ComfyUI. No dependencies, no DB, no config. One thing to know going in: 内容 is the Chinese label for "content", and the only other field is 权重 ("weight"). There's genuinely not much more to this node - which is the point. When the alternative is typing (curly_hair:1.2), (long_hair:1.1), ... for the tenth time, a small node that does it for you is a small win. Just keep it out of your Anima workflows.