NVIDIA has recently presented an impressive tech demo. With simple brush strokes, you can roughly outline how you imagine a landscape, and the software generates - at first glance - photorealistic results that you would dominate otherwise advanced masking techniques and use many different Photos. The future of compositing? Look for yourself:
Basis of the very impressive indeed "magic" is once again the so-called machine learning. In this case came Generative Adversarial Networks (GAN, "producing generic networks") are used.
The video title promises "masterpieces" (Masterpieces) and is once again in the comments on Youtube and in the social networks of the demise of skill and creativity, indeed all professions, conjured up.
But such fatalism was also already on the agenda, as in the last century, photography was on the rise and the painters saw them swim their skins. 😉
In detail, the generative assembled photocells are here far from high enough to actually find use in commercial life. But that will surely change in our lifetime.
But even this impressive technology is only a tool for creative people to realize their ideas more easily and to do their jobs faster. So instead of having to own together looking for suitable stock photos for easy background elements, the basis of further work can be created with a few brush strokes. The future is almost there, yes.
Happy Computer Arts and a great weekend!