Virtual Try-On Room with SMPL Anthropometry

vTry-On Room uses SMPL (Skinned Multi-Person Linear) model that represents various body shapes in natural human poses.

ML enabled solution

3D models are based on skinning and shape blending

Partner's Profile

Background

vTry-On Room is a machine learning case study that enables virtual try-on experiences by fitting virtual clothing onto SMPL models. The SMPL (Skinned Multi-Person Linear) model is used as a representation of the human body. It provides a parametric model that captures the shape and pose variations of a human body.

Objective

By adjusting the body shape parameters of the SMPL model, users can simulate how different garments will fit on individuals with various body shapes.

Visit website

Solution

With the SMPL Anthropometry library, we fit virtual clothing onto SMPL models to create virtual try-on experiences.

The SMPL Anthropometry library provides functions to manipulate the SMPL model's shape and pose parameters, apply clothing simulations, and render the virtual try-on experience. By adjusting the body shape parameters, it allows for realistic representation of individuals with different body shapes, enabling a personalized virtual try-on experience. This application can be applied in e-commerce, fashion design, and virtual dressing rooms.

Virtual Try-On

With the simulated clothing draped onto the SMPL model, the virtual try-on experience can be created. By rendering the virtual model with the clothing, users can visualize how different garments would fit and look on their body shape. They can explore different clothing styles, colors, and sizes without physically trying them on. This can help in making informed decisions before purchasing clothes online or aid in the design process for fashion designers.

Body Reconstruction

The first step is to obtain the SMPL model parameters that best match the body shape of the person for whom the virtual try-on is being performed. This can be achieved by utilizing body scanning technologies, such as 3D body scanners, or through image-based techniques that estimate body shape from 2D images or video.

Clothing Simulation

Once the body shape parameters are obtained, virtual garments can be simulated and draped onto the SMPL model. Various simulation techniques can be employed, such as physics-based cloth simulation or geometric-based methods, to simulate the behavior and appearance of the clothing on the virtual body. These techniques take into account the body shape, pose, and the physical properties of the fabric to realistically simulate the clothing's behavior.