VNCCS Character Creator V2

Keeping one character looking like themselves across a hundred sprites is the problem the whole AI-art community has been chewing on for years - new pose, new outfit, and suddenly it's a different person. VNCCS Character Creator V2 is the node where you stop generating one-off images and start defining a character record that the rest of the suite will keep consistent for you. It's the first thing you touch in VNCCS 3.0's Step 1 workflow, and it's the node that actually creates the character before anyone puts clothes or a pose on them.

What it does, and why it's a "V2"

The original Character Creator exposed a wall of raw tag fields - eyes, hair, body, skin colour, all typed by hand (that node still exists as plain CharacterCreator, and it's worth reading separately if you want the API-friendly version). V2 replaces that with an actual widget UI: press NEW, name your character, then either fill tags in manually using the pencil-icon tag builders for sex, age and NSFW status, or hit CHARACTER WIZZARD and just describe who you want - it fills the fields for you. There's a GENERATE PREVIEW button so you can sanity-check the look before committing to a full run.

You also pick the generation model here: Illustrious, the veteran SDXL anime checkpoint with a huge LoRA library and fast iteration, or Anima, the newer Cosmos-Predict2-based model that's much better at understanding what you actually typed but runs three-ish times slower and has a thinner LoRA ecosystem so far. Neither is wrong - Illustrious if you want speed and an existing LoRA for the style you're after, Anima if you're describing something specific in plain language and want it followed.

Inputs and outputs

This is a widget-driven node, not a socket-driven one - the schema shows no required or optional inputs at all, because everything (name, tags, model choice, the wizard, the preview button) lives inside the node's own panel rather than on wires. What comes out the other side:

• character (IMAGE) - the generated character sheet/portrait.
• sheets_path (STRING) - where VNCCS saved the files, which downstream nodes like Pose Studio and the generator nodes read from.
• background (wildcard *) - carries the background/matte choice forward so later stages composite consistently.

From here the output normally flows into VNCCS Pose Studio and then one of the generator nodes (VNCCS_CharacterGenerator or VNCCS_ClothesGenerator) to actually render the sprite sheet.

Installing it

Easiest path: open ComfyUI Manager, search "VNCCS - Visual Novel Character Creation Suite", install the latest version, restart. Manually, it's cd ComfyUI/custom_nodes && git clone https://github.com/AHEKOT/ComfyUI_VNCCS.git, then relaunch and use Manager's Install missing custom nodes to grab the rest of the chain - Pose Studio in particular lives in the author's separate VNCCS-Utils repo, so don't skip that step. If you're setting up ComfyUI from scratch, the author also ships VNCCS Easy Install (github.com/AHEKOT/VNCCS_Easy-Install), a preconfigured ComfyUI with VNCCS already wired up.

Common issues

Character generation here depends on models actually being downloaded through the VNCCS Control Center node first - if the preview or the full run just errors with nothing useful in the console, check Control Center's download status before assuming this node is broken. If you're migrating characters made on an older VNCCS (2.x), the pack ships a dedicated VNCCS_MigrationAssistant workflow - run that before deleting your old character folder, and confirm the migrated character actually works before you clean up. And a small one that trips people up: the NSFW toggle in the tag builders controls whether the base character generates clothed or not - it's not a content filter, it's deciding what the naked base layer looks like that clothing gets generated on top of later.