Described are systems and methods for generating synthetic image data including synthetic images, depth information, and optical flow data. Embodiments of the invention assemble image scenes from virtual objects and capture realistic perspectives of images scenes as synthetic images. Realistic perspectives captured in synthetic images are defined by camera views created from camera settings files. To simulate capture performance of smartphone cameras, stereo cameras, and other actual camera devices, capture, calibration, and camera intrinsic parameters included in camera settings files are identical to parameters included in actual cameras. Synthetic image datasets generated by the systems and methods described herein are used to train machine learning systems generating machine learning models for performing computer vision tasks.