VNCCS 3.2.2 adds Qwen Image 2.1 and MiniMax H3 to its ComfyUI sprite pipeline

AHEKOT · reddit · 2026-10-05

Developer V-chan released VNCCS 3.2.2, a ComfyUI pipeline that creates characters and turns them into sprite sets (poses, outfits, expressions) for visual novels and games. Key updates:

Qwen Image 2.1 is the new main model: base characters, poses, outfits, expressions — plus native transparent background (Alpha) generation that survives clothing edits, emotion editing, and SeedVR upscaling, eliminating green-screen cleanup;
MiniMax H3, a video model, is drafted for sprite duty: pose, outfit generation, and clothes cloning with the first frame locked as the character image;
Character Creator revamp: editable tag chips, 40 visual style presets, 61 species presets (mixable for hybrids), full-body/cowboy-shot framing;
Resolution scale slider (1–4 MP) remembered per model family; Qwen's Character Overhaul LoRA stabilizes characters at a small cost of unique style.

Klein9b remains available; a Control Center consolidates model families, assets, and Turbo downloads.

Original post →

VNCCS 3.2.2 adds Qwen Image 2.1 and MiniMax H3 to its ComfyUI sprite pipeline

More from Multimodal