VRGDG Multi Cycling Text Picker

VRGDG_MultiCyclingTextPicker is up to twenty independent "choose one item from this list" pickers packed into a single node, each with its own preset, list, selection mode, and formatting - all joined into one output. It's the node that keeps a long music video from repeating the same camera move, lighting setup, and interaction every single scene.

Here's the boredom problem it solves. If every scene prompt says "slow push-in, warm lighting, walking through tall grass," every scene looks the same, and a 16-scene video becomes one scene rendered sixteen times. The fix is variety from a controlled pool: scene 1 gets a push-in, scene 2 a tracking shot, scene 3 an orbit, and so on - but only from the pool you approved. This node is that pool, and it cycles through it.

Inputs worth knowing

Each picker (preset_1 … preset_20) has the same set of controls. The ones you'll actually touch:

• picker_count - how many pickers are active (0–20). Set 0 and the whole thing is disabled, which the tooltip notes is the way to turn off "advanced prompt details."
• preset_N - a canned label/list. Picker 1 ships with a Camera Motion preset preloaded, and it's a good list: push-ins, tracking, dolly, orbit, crane. Others default to Custom for your own list.
• items_N - the actual list of choices. It parses lines, blank-line chunks, commas, pipes, JSON/Python lists, even objects with items/values/motions keys. That flexibility means you can paste in a JSON array from another tool and it just works.
• selection_mode_N - index (choose by the index_N step number, deterministic), or seeded random modes (with/without repeats).
• pick_count_N - how many items to grab at once (default 1). Pick two and two_item_template_N ("start with {item1} then follow with {item2}") formats them into a sentence.
• label_N - an optional label ("Camera Motion:") prefixed to that picker's output.
• joiner - how active picker outputs combine into combined_formatted_text: newline, blank line, comma, or pipe.

Outputs: combined_formatted_text (everything joined, ready for the prompt creator), results_json (the selections as structured data), and one formatted_text_1…formatted_text_20 per picker if you want to route them individually.

The workflow payoff

In this pack, the picker sits between the LLM and the scene prompts: the LLM writes the narrative, and the picker injects per-scene variation that stays inside your guardrails - a slow pan here, a crane rising move there, never a wildcard mid-scene meltdown. The seeded random modes plus per-picker seeds mean you can make a run reproducible when you need it and chaotic when you don't.

Install

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

Restart ComfyUI; it's under VRGDG/General. ComfyUI Manager: search vrgamedev. No extra downloads.

Honest note

This node is a lot of surface area for what it does - twenty pickers of widgets will sprawl across your canvas, and you'll likely only use 3–6. Don't feel obligated to fill them. And remember the picker only knows the lists you give it: a boring list produces boring variety, so curate the pools (the shipped Camera Motion preset is genuinely decent, and a good template for how to write your own). It's a variety machine, not a creativity machine.