Guides
A-pose vs T-pose: which to use for a 3D character
Why characters are modelled in A-pose or T-pose, how each affects rigging and shoulders, and which pose to ask an AI generator for.
3 min read
Quick answer
Should a 3D character be modelled in A-pose or T-pose?
Both work for rigging. T-pose, with arms straight out, is simple to rig and is a common default for auto-riggers and humanoid systems. A-pose, with arms angled down about 30–45°, deforms the shoulders more naturally because it sits closer to the arm's resting position. For AI generation, either works as long as the arms are clearly separated.
Why a "rest pose" at all?
Characters are modelled and rigged in a neutral rest pose (also called a bind pose): standing straight, limbs apart, nothing touching. Then the rig moves them into every other pose. A good rest pose makes rigging easier, because each body part is clearly separated, and makes deformation better, because joints start from a sensible position.
The two standard choices are T-pose and A-pose.
T-pose
Arms straight out to the sides, horizontal, palms down. Legs straight, slightly apart.
Pros:
- Simple to rig: arm bones line up with world axes.
- Arms are fully separated from the torso.
- A common default for auto-riggers, humanoid retargeting and many animation libraries.
Cons:
- Shoulders are raised far from their natural resting position, so the shoulder and armpit area is modelled in an extreme state. That can cause odd deformation when arms come down.
- Takes up more horizontal space in a reference sheet.
A-pose
Arms angled down, typically about 30–45° from horizontal, slightly forward. Legs slightly apart.
Pros:
- Shoulders closer to a natural pose, so they deform better for most animations.
- Common in modern game and film character work.
- Clothing modelled in A-pose often looks more natural.
Cons:
- Arm bones aren't aligned to axes, which matters for some older tools.
- If the arms are too low, they can be close to the hips, causing weight-painting headaches.
Which should you choose?
- Using Mixamo or a humanoid retargeting system: either works; Mixamo asks for a neutral pose. T-pose is the safest if you're unsure.
- Character with lots of arm-down animation (most games): A-pose for better shoulders.
- Character in bulky clothing or armour: A-pose, so the clothing hangs naturally.
- Fast prototyping: whichever your rigging tool likes.
Hands and legs
- Hands: open and relaxed, fingers slightly apart. Fists and fused fingers are much harder to rig.
- Legs: a little apart, so the inner thighs are separate.
- Feet: flat, pointing forward.
- Head: facing forward, neutral expression, mouth closed (or slightly open if you'll rig the jaw).
For AI generation
Generators follow the pose in your input. To get a rig-friendly mesh:
- Draw or prompt an A- or T-pose explicitly: "full body, front view, A-pose, arms angled down away from the body, legs apart, open hands".
- Check the gaps: arms must not touch the torso; legs must not touch each other.
- In Vi3W, check the image step before generating 3D. If the arms are glued to the sides there, they'll be fused in the mesh.
- Use a front and side reference if possible. Character turnaround sheet to 3D.
Then rig it: using Mixamo with an AI-generated character.
Ready to generate your character in a clean rest pose? Try it in Vi3W.
Last reviewed