VNCCS Character Creator

Before Character Creator got a proper widget UI with a wizard and a preview button, this is what building a character in VNCCS looked like: a node with one field per attribute, all exposed as plain sockets you can wire into or override programmatically. It's still in the pack, and it's worth knowing about even though CharacterCreatorV2 is what the README walks you through - this is the node you want if you're scripting character generation, batch-creating variations, or just prefer typing tags to clicking through a wizard.

What it does, and how it differs from V2

Where V2 hides its configuration inside a custom widget panel with no exposed sockets, plain CharacterCreator puts every attribute on its own input: sex, age, race, eyes, hair, face, body, skin colour, aesthetics, an NSFW toggle, a seed, even a lora_prompt field for LoRA trigger tokens. It's built for tag-based prompting - the aesthetics default is the classic Illustrious/SDXL quality prefix (masterpiece, best quality, amazing quality), and the negative default (bad quality, worst quality) follows the same convention. If you're running an Illustrious-family checkpoint through VNCCS and you already think in Danbooru-style tags, this node maps to that far more directly than the wizard does.

One detail worth noting: unlike CharacterCreatorV2, this node doesn't output an image at all - it outputs prompt text and paths. It's a prompt-assembly step, not a generation step; you're expected to feed its outputs into your own sampler or into the rest of the VNCCS pipeline yourself.

Inputs and outputs that matter

• existing_character (enum) - load an existing character to edit, or start fresh.
• The bulk of the fields are optional attribute strings with sensible defaults you'll want to override: sex, age (0–120, default 18), race, eyes, hair, face, body, skin_color, aesthetics, additional_details, background_color (default "green", matching the chroma-key convention used everywhere else in VNCCS), and nsfw (BOOLEAN) - same as elsewhere in the suite, this controls whether the base character generates clothed or not, it's not a content filter.
• negative_prompt, lora_prompt, seed, and new_character_name round out the rest.
• Outputs: positive_prompt, negative_prompt, seed, sheets_path, faces_path, face_details - and age_lora_strength (FLOAT), which scales an age-conditioning LoRA in proportion to the age value you set, keeping generated characters visually consistent with the stated age.

Installing it

Ships with the base pack, no separate install. ComfyUI Manager: search "VNCCS - Visual Novel Character Creation Suite", install latest, restart. Manual: cd ComfyUI/custom_nodes && git clone https://github.com/AHEKOT/ComfyUI_VNCCS.git, restart.

Common issues & troubleshooting

Nothing gets generated, only text comes out. That's correct behaviour, not a bug - this node builds prompts, it doesn't render images. Wire positive_prompt and negative_prompt into your own KSampler/checkpoint setup, or use CharacterCreatorV2 if you want an all-in-one widget that handles generation itself.

Results don't match V2's output for the "same" character. The two nodes aren't guaranteed to produce identical results even with equivalent settings - V2 layers a wizard and preview step on top that can adjust tags in ways this raw version doesn't replicate automatically. Treat them as two different interfaces to the same underlying idea, not interchangeable outputs.

Tag fields feel disconnected from natural-language prompting. That's by design - this node is built for a tag-based Illustrious-style workflow. If you're on VNCCS 3.0's Qwen/Anima pipeline and want natural-language descriptions instead, use CharacterCreatorV2's wizard or CharacterAssetSelectorQWEN's multiline fields.