
3D Human from Photo
Make a 3D model out of any image with full-body humanoid on it
Perfect for characters concepts
Quick 3D preview
Recognizes standard poses

Input
Image

Output
3D mesh

Main task
Create a 3D character

Neural network
ICON

Release date
05.03.2022
Parameters
Resolution
Choice from 4 options It defines the overall quality of mesh We recommend to use x256 if you want to quickly generate the best possible 3D model
Pose method
Choice from pymaf and pare It makes a slight difference: pare works slightly better and gives a better mesh. However, pare doesn't work with resolution higher than x128 We recommend not to change unless you're an experienced user
Advanced parameters
Body iterations
Range: from 1 to 200 It defines overall quality of a model shape. The higher the number is, the more detailed edges there will be and more humanoid features the final 3D object will have. It will also fill space between joints, so we recommend to keep it at 100
Cloth iterations
Range: from 1 to 200 It defines the level of detailization of model's structure. If the number is low, the overall mesh will look more rounded and have less sharp details. However, more random bumps will be seen We recommend to keep it at 100-150