Synthesis AI announced today that they have rolled out groundbreaking text-to-3D technology that produces realistic digital humans in a cinematic-quality. The company says their new technology is the first to render high quality text-based 3D digital humans which can be used for mounting applications like gaming, virtual reality, films and simulations.
Synthesis AI’s proprietary technology uses generative artificial intelligence and visual effects pipelines to create human-like models that come to life through a text prompt. All you need to do is simply input their description, such as their age, gender, ethnicity and style, and the AI will generate a 3D version that matches the parameters. Also, users can quickly edit the 3D model by changing the text prompts or adjusting facial expressions and lighting through sliders.
This special text-to-3D technology was introduced to broaden the scope of advanced artificial intelligence applications, by offering perfectly labeled synthetic data to cross train machine learning models. Synthetic data is artificially generated data that imitates human-generated data, but uses some sensitivity or personal data.
Synthesis AI CEO and founder, Yashar Behzadi said that their text-to-3D capability takes a programmatic API-driven approach as the starting point, and outputs a high resolution 3D model that can be used as a synthetic data across numerous of use cases such as VR/AR, gaming, VFX, smart cities and other industrial and manufacturing simulations.
The introduction of text-to-3D plays a significant role in reducing the overall cost of 3D applications that are growing in many industries. This complex process typically requires the collaboration of several elements such as creating geometry, meshes and texture layers. Moreover, the deployment of high-quality models usually requires expensive photogrammetry setups and tons of intensive work to combine these images in the best way.
However, Synthesis AI’s new text-to-3D technology smartly combines generative AI and VFX pipelines to generate labeled synthetic data to train machine learning models. This difficulty-free approach can produce satisfactory results within a short period and at a reasonable cost. All its users need to do is to type a definition of the required digital human, and the text-to-3D technology will do the rest.
Synthesis AI company provides three products: Synthesis Humans and Synthesis Scenarios that represent in-depth offerings of human-centric synthetic data, and Synthesis Labs that makes its no-code 3D generative AI capabilities more accessible to all experience levels.
Synthesis AI stands out for its focus on enabling enterprise, industrial and public sector customers to simulate reality by synthesizing in any given person, place or object. These applications include synthetic data that train computer vision models in movies, VR/AR and other content production areas.
Currently, the text-to-3D technology is available to a select group of beta testers starting in Q2 2020. The company is planning to continue to improve the performance of the underlying models by gathering feedback on its community.
