VRGDG_StringConcat

VRGDG_StringConcat is a prompt-builder with three named inputs and one output, and it's best understood as the front door of the pack's music-video workflow. The three fields map directly onto how you describe a video you want to make:

• instructions - your overall direction: what the video should be, how it should feel, any constraints.
• song_theme_style - the song's theme and visual style.
• pipe_separated_lyrics - the lyrics, fed as pipe-separated chunks (line one | line two | ...), ready to be split and synced.

The node stitches these into one concatenated_string, which then flows to the LLM node that turns it into the full scene plan. It's the difference between typing the same three things into a text box every run and having them wired in as named inputs you can swap independently - change the lyrics without touching the style, or the style without touching the lyrics.

Why a dedicated node for "glue strings together"

You could build the same thing from a couple of Text nodes and a generic concat node, and plenty of workflows do. What this node adds is a stable contract: because the inputs have fixed, meaningful names, the rest of the pack's workflows know what to expect. The LLM node downstream reads a prompt assembled this way and can reliably find the lyrics vs. the style vs. the instructions - a generic concat produces an anonymous blob. In a pipeline that runs for hours unattended, that named structure is what keeps every stage on the same page.

What to actually set

For a beginner, song_theme_style is where you get most of your creative mileage - it's the seed of the whole visual language. instructions is your directorial override ("keep her in frame", "no text overlays"). pipe_separated_lyrics is the mechanical part: the format matters, so if the workflow is mis-splitting lyrics, check whether your separators actually match what the node expects. The output is the one concatenated_string - wire it into the LLM node's user_input (or wherever the workflow routes the master prompt) and the pipeline takes it from there.

Install

Same pack, once:

cd ComfyUI/custom_nodes
git clone https://github.com/vrgamegirl19/comfyui-vrgamedevgirl

or ComfyUI Manager → "vrgamedev", restart. Requirements kornia, librosa, imageio install with it.

The honest take: this node is only interesting inside the music-video workflow - on its own it's string concatenation with extra steps. But that's the point. It's part of an opinionated assembly line where each node has one job and a name that says what it is, and this one says "this is where your video idea starts." If you're loading the pack's workflow from the repo or a tutorial, this is the node you edit first, because it's where your song becomes the prompt.