For more informations, see the tutorial on YouTube:  https://youtu.be/pA_tZCscOFg

This workflow uses the Screen Share node by the MixLab node suite ( https://github.com/shadowcz007/comfyui-mixlab-nodes?tab=readme-ov-file ) to generate "real time" (emphasis on the quote unquote) images sourced from the viewports of virtually any 3D software.

In order to be as lightweight as possible, it uses either Hyper SD (this version) or SDXL Lightning ( this other version  https://openart.ai/workflows/camel_milky_30/hyper-sdxl---generate-from-blender-any-3d-app-viewport-via-screenshare/3ywJV6S8PjeQAV9o9X1W ) and only a Depth ControlNet. A open pose ControlNet is provided in the workflow, but can be disabled for better performance.

It can be expanded with more ControlNets, depending on the subject and composition wanted, as well as a Popup Preview node ( https://github.com/NimaNzrii/comfyui-popup_preview ) to lock an image generation popup into place inside of our 3D software environment.