VRGDG_PromptCreatorUI_V2

There are two nodes in this pack named almost identically - VRGDG_PromptCreatorUI and VRGDG_PromptCreatorUI_V2 - and this is the one current workflows actually use. The V1 still exists because the pack ships both the older and newer workflow variants, and nobody wants to break saved graphs. But if you're starting fresh, this is the one you want.

Like its sibling, it's a UI-only node: no inputs, no outputs, no data flowing through. Its entire job is to be the control surface where you define style, mood, and scenes for the music video workflow, with all that state living in the workflow JSON for the rest of the graph to read. The V2 is the same idea with the current layout and presets, matched to what the newer versions of the workflow expect.

The mechanism

Same pattern as the pack's other "dashboard" nodes. The music video pipeline works like this: your audio gets transcribed into lyric segments, those segments become scene slots, and the creative direction you type into the Prompt Creator steers how the LLM node turns each lyric into a full scene prompt. This node is where that creative direction enters the graph. Since it carries no data ports, you never wire it - you fill it in.

What you'll actually find on it: fields for the subject/character, style and mood, and scene direction - the same shape of brief you'd write for any image-to-video job, but structured so the workflow can consume it programmatically. The more concrete your subject description, the better the scene prompts downstream, because the workflow's prompt-splitting nodes only have what you gave them to work with.

Which one do you pick?

• Loading a shipped workflow → use whichever version it references. If it calls for V2, don't swap in V1 and expect identical results; the node set around them differs.
• Building your own → V2. It's the current one and the one the author iterates on.
• Bypassing either → pointless. Zero-port nodes don't affect rendering; they just hold state.

Installing it

It comes with the pack:

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

restart ComfyUI, or use ComfyUI Manager → search vrgamedev. No models to download for this node itself. The workflow it belongs to does expect the pack's dependencies (kornia, librosa, imageio) installed.

The honest take

The V2 is a "which version should I click" answer more than a feature story. There's nothing to configure on the node itself that's exotic - the real payoff is upstream, in how carefully you brief it. Fill in the subject and style once, and the rest of the pipeline stops asking.