Frequently Asked Questions About Virmii

A clear character image and a short written description. A reference image is required because Virmii does not currently offer text-only generation.

For best results, use a clear, well-lit full-body or half-body character image with the character facing mostly forward. Portrait images around 1024–2048 pixels on the long side generally provide enough detail. Keep the complete head, hair, clothing, hands, and feet visible where possible, with some space around the character. Avoid extreme poses, hidden limbs, overlapping props, text, and watermarks.

Virmii accepts PNG, JPG, and WebP reference images up to 10 MB. Images must be at least 256 × 256 pixels and must not exceed 40 megapixels. For better generation quality, we recommend a clear portrait image with a long side of approximately 1024–2048 pixels rather than relying on the minimum resolution.

Use a clean solid-color background or a subtle gradient with clear contrast between the character and the background. Soft, even lighting works best. Avoid complex environments, heavy textures, dramatic lighting, strong cast shadows, motion blur, particles, and colors that blend into the character's hair or clothing.

Time varies with provider capacity and the model's complexity. Virmii shows live stage-by-stage progress so you can leave the job page and return later.

No. Full facial tracking requires suitable facial topology and validated expression shapes. Results that cannot deform reliably are labelled as head-only rather than presented as full-tracking avatars.

A typical 3D VTuber workflow uses a model maker, a VRM-compatible avatar tool, face-tracking software, and streaming software such as OBS. Virmii connects those stages in the browser while still providing a VRM for compatible third-party tools.

Completed jobs can provide a VRM download and, when available, the generated GLB. Downloads use short-lived access links, so reopen the job if an older link expires.

Virmii creates a private live-stage URL for the avatar. Add that address to an OBS Browser Source, then use the Virmii Live Studio in a supported browser as the camera-tracking controller.

Virmii supports current desktop versions of Chrome and Edge for camera tracking and live transport. Static avatar preview may work in other modern browsers, but the complete live workflow is not guaranteed there.

VRM is designed for interoperability, but each third-party application supports a different subset of the specification. Check that application's VRM version and feature requirements before relying on it.

Raw camera video is processed locally in the browser by default. Virmii sends movement data to the live stage as needed, but does not use Convex to relay raw camera frames.

Only upload material you own or are authorized to use. Do not use Virmii to impersonate a person, violate publicity rights, or reproduce protected characters without permission.

You keep your rights in your input. As between you and Virmii, you may use the generated output subject to applicable law, third-party provider terms, and any rights that already exist in the underlying character or reference.

Virmii does not charge a recurring subscription. AI avatar generation uses credits purchased in one-time packs; the current price, included credits, and applicable taxes appear before checkout.

Use the delete control for an individual generation job. For an account-level access, export, correction, or deletion request, contact contact@virmii.com from the email associated with your account.