By
Tushar Sharma

Synthetic to Real

Synthetic to Real

ZEG is pushing boundaries to create synthetic content on demand. Over last few days, we created 100s of bottles (see sample below of pre-rendered bottles). The synthetic ones are not your usual photos with information at pixel level - these are 3D with information down to shape, color, pose, material, and much more. In fact, you could re-create them using our simple json file and manipulate in real time.

Note: Synthetic images are pre-rendered. Once rendered, they look same as real

There are various applications of content on demand. We have been approached by multiple industry types from automotive (gasket identification using Computer Vision for productivity improvement) to utilities (Identify damage in water pipes automatically using a camera) to bikes (rust on bikes)! In all these cases, the problem is getting real world data to automate the workflow using AI. Manually collecting data of such products will require massive manual work and lot of time. Plus, don't forget the annotation. In our case, we provide data fully annotated with information not possible to capture using a general photo - imagine all parameters in real world are now available in virtual.

You could power the algorithms in your industry and automate workflows with Synthetic data that can be generated at click of a button - 'Load and Go'

We see other applications too such as enabling gamers bring their own assets into the game by just taking a picture of items around them. Seeing picture below felt like we are home.

Reach out to us at contact@zeg.ai to learn more about our synthetic content generation capabilities


Tushar Sharma