I think people will be puzzled by why I published this video. I've been producing stuff at a much higher level of quality for many years after all.
This is the very first actual output from a process I originally envisioned in 2011, Which I call the McJarkanizer. The very first research papers on very primitive style transfer were starting to come out in the academic community, With significant improvements by 2014. Code of this type couldn't really run effectively on consumer hardware for another decade.
There are a tremendous number of prerequisites required to do this process, But many of those were already created during development of the save point pipeline, And when a few other technologies and hardware converged at a point where I could see it was all immediately possible, I began orchestrating the process and coding it.
This is the very first thing to come out of the very first version of this pipeline that actually worked and completed the job. It's extremely rough and not suitable for production, not even close. But to the trained eye, The significance is already clear.
I'll show you all what one of the branches of the main save pipeline looked like in its first month and what it looked like two years later.
The point of this particular aspect of the pipeline is to allow one click conversion of entire films into other visual styles. The first job will be to get the quality on point which might take six months of development or more. During development it will likely expand to do other associated things such as replacing an actor through an entire film With no other changes, Turning a photorealistic Western movie Into a photorealistic sci-fi movie, And other tasks based around the concept of controlled mutation of existing media.
Last edited:
Upvote
3
