This brief explains what SMPL shape betas are, why Meshcapade Me currently limits them to 10 and will soon expand to 50, and how that choice affects file size, performance, and your day-to-day workflow.

1. What are “betas”?

In the SMPL family of body models each β (beta) coefficient weights a principal‑component axis of body‑shape variation learned from 3D body scans*. Changing a beta shifts the body representation toward a new shape while preserving articulation.

** Beta parameters represent variations in body shape which are learned from 3D body scans. You can think of them like sliders that modify how tall, short, heavy, etc the 3D body is. The more sliders you have, the more detailed a body you can create*

2. Current Support — 10 Betas (May 2025)

• Ten betas are enough to reproduce > 99 % of the shape variance in the original SMPL dataset.
• Each coefficient affects a major region (e.g., torso length, shoulder width, limb thickness).

3. Planned Upgrade — 50 Betas (target: Q3 2025)

• Adds fine‑grained control for local features such as ankle girth, finger thickness, subtle muscular definition.
• Option to recover better facial features as 10 betas recovers most of the body shape and 50 betas allows for more detailed face shape.
• Beyond 50, additional components contribute < 0.5 mm average per‑vertex change—imperceptible in most use‑cases but costly in file size.

4. Why not expose all 300 Betas?

5. Workflow Guidelines

1. Unused betas = 0

   Leave any betas you do not need at 0. This preserves the legacy behaviour and avoids unexpected shape drift.

2. Backward compatibility

   Assets authored today (10 betas) continue to load after the upgrade; the extra slots default to 0.

API Documentation

You can find more about how to extract and create from betas in our API documentation.