Live2D Artist & Rigger Needed – Full Body Model
Budget: ₹12,500 – ₹37,500 INR
Job Title: Live2D Artist & Rigger Needed – Full Body Model for AI Web Application
Project Overview
We are Anyboli Labs, an EdTech startup developing an interactive AI school tutor named "Alfie." We are looking for a skilled Live2D artist and rigger to recreate an existing flat character design into a fully rigged, web-ready Live2D model.
Alfie will be deployed in a custom WebGL web application (using PixiJS). The avatar is driven by an AI backend that generates text, which the browser converts into real-time audio and dynamic movement. Because of this, the final deliverables must include specific runtime configurations alongside the master project files for our internal development team.
What We Will Provide:
A flat PNG reference image of the character design.
A reference zip folder and screenshot showing the exact final export file structure our codebase requires.
Stage 1: Character Art & Separation (PSD)
You will use the provided flat PNG reference to redraw and separate the character into a high-resolution, Live2D-ready PSD.
Layering: Fully separate all body parts, facial features, and clothing for full-body physics rigging.
Advanced Mouth Chart (Crucial): Our platform uses an audio-analysis library (wawa-lipsync) to automate speech. The mouth cutting must include the separated layers necessary to support ARKit / Perfect Sync mouth forms (specifically the A, I, U, E, O vowel shapes) so the AI can lip-sync cleanly.
Stage 2: Expressions & Rigging
Standard Rigging: Full-body VTuber-style rigging including breathing, hair/clothing physics, XYZ head and body angles, and eye tracking.
3 Custom Expressions: Must be built so they can be toggled dynamically via code.
Excitement: Sparkle/star eyes or widened bright eyes.
Amusement: Happy, closed-smiling eyes.
Sadness/Frustration: Subtle pooling tears overlay.
Stage 3: Body Gestures (Motions)
We require 5 specific body-language animations created and exported as standard .motion3.json files (set up like VTube Studio hotkey animations). Our AI will trigger these when it detects certain emotions in the text.
Idle: A seamless, continuous breathing/subtle swaying loop.
Excitement: A "happy bounce" (e.g., bouncing slightly upward, raising shoulders, and a quick, happy double-nod).
Curiosity: A slight head tilt with an inquisitive posture.
Laugh: A quick shoulder-bounce or belly laugh.
Explain: A pointing or teaching hand gesture.
Required Deliverables
The final handover must include the master source files alongside the compiled runtime assets. Proposals that do not include the master source files (.cmo3 and PSD) will not be considered.
1. Source Project Files:
The Master Live2D Project File (.cmo3): The complete, uncompiled working project file with all parameters and physics setups intact.
High-Resolution Layered PSD File: The finalized character artwork with all original, cut layers organized cleanly.
2. Compiled Runtime Export Folder:
A standard VTube Studio / Cubism SDK 4.x export folder, strictly containing:
.moc3 (Compiled binary)
.model3.json (Master metadata configuration file)
.physics3.json (Physics data file)
.cdi3.json (Parameter dictionary file)
Texture Folder: The generated subfolder containing the layout texture atlas sheets (.png files).
motions/ folder: Containing the 5 individual .motion3.json animation files listed above.
expressions/ folder: Containing the 3 individual .exp3.json expression toggle files
Project Overview
We are Anyboli Labs, an EdTech startup developing an interactive AI school tutor named "Alfie." We are looking for a skilled Live2D artist and rigger to recreate an existing flat character design into a fully rigged, web-ready Live2D model.
Alfie will be deployed in a custom WebGL web application (using PixiJS). The avatar is driven by an AI backend that generates text, which the browser converts into real-time audio and dynamic movement. Because of this, the final deliverables must include specific runtime configurations alongside the master project files for our internal development team.
What We Will Provide:
A flat PNG reference image of the character design.
A reference zip folder and screenshot showing the exact final export file structure our codebase requires.
Stage 1: Character Art & Separation (PSD)
You will use the provided flat PNG reference to redraw and separate the character into a high-resolution, Live2D-ready PSD.
Layering: Fully separate all body parts, facial features, and clothing for full-body physics rigging.
Advanced Mouth Chart (Crucial): Our platform uses an audio-analysis library (wawa-lipsync) to automate speech. The mouth cutting must include the separated layers necessary to support ARKit / Perfect Sync mouth forms (specifically the A, I, U, E, O vowel shapes) so the AI can lip-sync cleanly.
Stage 2: Expressions & Rigging
Standard Rigging: Full-body VTuber-style rigging including breathing, hair/clothing physics, XYZ head and body angles, and eye tracking.
3 Custom Expressions: Must be built so they can be toggled dynamically via code.
Excitement: Sparkle/star eyes or widened bright eyes.
Amusement: Happy, closed-smiling eyes.
Sadness/Frustration: Subtle pooling tears overlay.
Stage 3: Body Gestures (Motions)
We require 5 specific body-language animations created and exported as standard .motion3.json files (set up like VTube Studio hotkey animations). Our AI will trigger these when it detects certain emotions in the text.
Idle: A seamless, continuous breathing/subtle swaying loop.
Excitement: A "happy bounce" (e.g., bouncing slightly upward, raising shoulders, and a quick, happy double-nod).
Curiosity: A slight head tilt with an inquisitive posture.
Laugh: A quick shoulder-bounce or belly laugh.
Explain: A pointing or teaching hand gesture.
Required Deliverables
The final handover must include the master source files alongside the compiled runtime assets. Proposals that do not include the master source files (.cmo3 and PSD) will not be considered.
1. Source Project Files:
The Master Live2D Project File (.cmo3): The complete, uncompiled working project file with all parameters and physics setups intact.
High-Resolution Layered PSD File: The finalized character artwork with all original, cut layers organized cleanly.
2. Compiled Runtime Export Folder:
A standard VTube Studio / Cubism SDK 4.x export folder, strictly containing:
.moc3 (Compiled binary)
.model3.json (Master metadata configuration file)
.physics3.json (Physics data file)
.cdi3.json (Parameter dictionary file)
Texture Folder: The generated subfolder containing the layout texture atlas sheets (.png files).
motions/ folder: Containing the 5 individual .motion3.json animation files listed above.
expressions/ folder: Containing the 3 individual .exp3.json expression toggle files