Launched this week

Silver Fox — Made with Astra

Launched this week

A VTuber model built with Astra, from art to facial rigging

2 followers

A VTuber model built with Astra, from art to facial rigging

2 followers

Can Astra build a VTuber model? Silver Fox documents the result: original character art, layered parts, facial rigging, expressions, and a model running in VTube Studio. Facial modeling is partly working; the body and full-motion polish are still in progress.

Silver Fox — Made with Astra gallery image

Launch tags:Art•Artificial Intelligence•OpenAI Day

Launch Team

LM-Kit OneThe Private AI Application Server

Promoted

OpenAI

GPT-6 Astra Challenge

What became possible in your product with Astra that was not practical before?

Maker

Astra made it practical to carry one original VTuber character through a long, iterative modeling workflow in a single workspace: character artwork, separated parts, eye and mouth variants, facial rigging, and a model running in VTube Studio. I could point out problems such as mismatched eyes, a detached-looking neck, stiff joints, or an awkward mouth, and have the agent inspect the assets, revise them, write supporting scripts, and check the result. Silver Fox is the evidence: the face and expression rig are partly working, while the body, arms, joints, and final camera polish are still in progress. Astra coordinated image generation, Live2D Cubism, VTube Studio, scripts, and focused Luna assistance. The achievement is making and refining the VTuber model itself; it does not depend on connecting the avatar to an AI chatbot.

Report

1d ago

Maker

📌

Hi Product Hunt! This is Silver Fox, a VTuber model we have been building with GPT-6 Astra in Codex. The experiment is simple: how far can an agent take VTuber modeling, from a character idea through artwork, separated parts, facial rigging, and a model that runs in VTube Studio? The result is a silver-haired fox character with a partially working face and expression rig. The showcase includes real VTube Studio captures, plus newer eye and mouth artwork that is still being refined. Body modeling, arms, joints, and full-motion naturalness are still works in progress. Astra directed the production work and iterative fixes, using image generation, scripts, Live2D Cubism, and VTube Studio. Luna subagents helped with focused implementation and verification. The point of this entry is the character-modeling result and what we learned while making it. This is a modeling project, not an AI chatbot running the character and not a one-click modeling service. The newest face and mouth artwork is not yet fully integrated into the native rig, and final camera acceptance remains pending. I'd love to hear from VTuber artists and riggers: which parts of the face look convincing, and what should we fix next?

Report

7d ago

OpenAI

GPT-6 Astra Challenge

What became possible in your product with Astra that was not practical before?

Maker

Report

1d ago

Maker

📌

Report

7d ago