Pose Control
Generate 3D Characters in T-Pose, A-Pose, or Any Custom Pose
Pose control sets a character's stance before generation, so the model leaves Meshy standing the way your pipeline needs it — T-pose or A-pose for rigging, or any custom pose you set joint by joint — textured and ready to export as FBX, GLB or STL.

How to Generate a Character in the Pose You Need
Three steps in the Image to 3D workspace — no rigging software, no posing plugin, no re-posing afterwards.



- 1
Upload a Reference Image
One photo, sketch, or piece of concept art. The character in it can be standing however it happens to stand.
- 2
Pick the Stance
A-pose or T-pose for rigging, or a custom pose from six presets and per-joint control. Set before generation, not after.
- 3
Generate and Export
About a minute for the mesh, plus two to three more for texturing on Meshy 7 — then out as FBX, GLB, STL, OBJ, USDZ, BLEND, 3MF, or DXF.
Why Meshy's Pose Control Stands Out
Less for the Model to Invent
A reference image only shows one side. Limbs clear of the torso leave less of the back hidden, so there is less for the generator to invent.
The Photo's Pose Stops Being a Constraint
Your input can show the character standing however they happen to stand; the model still comes out in the stance you picked.
Three Stances, One Control
A-pose, T-pose, and custom sit in the same panel. Switching between them costs nothing — no other tool, no restart.
Six Presets, or Every Joint
Start from a preset — punch, guard, kneel, run, sit — then drag individual joints until the stance is the one you actually wanted.
A Textured Character, Not a Grey Reference
Posing tools hand back a mannequin to model from. This hands back the finished character, with PBR textures up to 8K.
Three Ways to Set a Stance

T-Pose — Arms Straight Out at Shoulder Height
The calibration stance rigging and retargeting tools expect. Every joint sits on a clean axis, which is exactly why automated skeleton detection prefers it.
A-pose and T-pose are available in Image to 3D, Text to 3D, and batch image generation.
A-Pose vs T-Pose vs Custom Pose: Which to Use for Auto-Rigging
A T-pose holds the arms straight out at shoulder height; an A-pose angles them down about 45 degrees. Both are neutral stances a rigger can read — a custom pose is the opposite choice, for models that will never be rigged.

T-Pose — When Something Automated Has to Read the Skeleton
Unity's manual calls a T-pose the required pose for making a humanoid Avatar, and Roblox's R15 rig expects it too. Trade-off: weights baked from an arm held this high can pinch when it comes down.

A-Pose — When a Human Finishes the Rig
Unreal's UE4 Mannequin ships in an A-pose, which is why Epic's retargeting docs cover fixing an A-to-T base-pose mismatch. Trade-off: the 45° angle is not standardised, so T-pose-tuned tools may map the upper arm imprecisely.

Custom — When the Model Is Not Headed for a Rig
Six presets and per-joint control, for output that is never going to be rigged — a printed figure, a scene, a portfolio render. Trade-off: auto-riggers want a neutral stance, so pick A or T if rigging comes next.
Poses People Are Publishing
Real models from the Meshy community — neutral stances built for rigging, and custom ones where the pose is the point.
















Browse every posed model and T-pose model the community has published.
What People Build With It



Loved by 12,000,000+ Creators Worldwide
Related Use Cases
Related Blog Articles

ComparisonsBest AI Auto-Rigging Tools in 2026: 7 Options Compared for Real Character Work

Comparisons10 Best 3D Character Creator & Design Tools (Free & Paid) [2026]

ComparisonsHow to Make Anime 3D Models: Essential Tools

ComparisonsCreate 3D Animation Characters in Minutes: Best 3 Tools Revealed
Related Tools
Pose Control FAQ
Humanoid characters — bipedal figures with a head, torso, arms, and legs. Pose control drives a humanoid skeleton, so creatures, vehicles, props, and architecture have nothing to pose; for those, generate without pose control. Two other combinations are not supported: a custom art style cannot be applied in the same generation as a pose, and an image flagged as NSFW fails rather than generating.
Because the stance is not a text instruction. Pose control passes the skeleton you set as a structural reference alongside your prompt or image, which is a far stronger signal than a phrase in a sentence. Typing T-pose leaves the stance open to interpretation; selecting T-pose does not.
Match the stance to whatever consumes the model next. Choose T-pose when an automated rigging or retargeting tool reads the skeleton; choose A-pose when a human finishes the rig, or when the character's arms will mostly be down and shoulder deformation matters. If the model is headed straight into a game engine and you are unsure, pick T-pose.
No. Pose control changes the stance the geometry is generated in; PBR texturing then runs on that geometry as usual, up to 8K. A posed character exports with the same texture maps a default-stance one would.
Explore More Features







Pick a Stance. Generate the Character.
One reference image, the stance your pipeline needs, and a textured character that is one click from a rig.
Last updated: September 9, 2026

