Retired 30 year veteran camera guy/editor/motion graphics-infatuated with AI

Well the title almost says it all. I've worked on everything from HBO promos to Hollywood movie and all that exist in between. In my retirement my favorite thing to do is to go into my home edit bay and cut music videos and adaptations of poetry that benefit from the imagery of AI dramatically. In six months I've cut about 12 different short pieces as well as experimented with a lot of techniques with just single clips.

I wanted to share my first for a into doing a more narrative piece based on the epic poem "the rhyme of the ancient mariner". I wanted to see how far I could push AI into a more narrative form. And because of the predictive nature of the technology is certainly was a challenging experience. I swore to myself I would never try another narrative until something happened in AI that would give me more repeatable control.

And four weeks ago I stumbled across it in my hopeful online researching. It's a hybrid technology that in the past has been used for pre-viz. But it is way beyond that now. The company and software is called Iclone 8 - AI Studio. And the workflow is basically this similar to pre-viz you use a traditional 3-D animation stage set up where you can place your characters and frame and position your cameras and set up your lights and record all the different kind of marks you want to be hit by both the talent in the camera. And at the end of the session you output a color version of what you recorded at low resolution. No weeklong renders here. You also output the same video but as a depth map on everything in that video which tells the AI many things but the most important is how to wrap the images around all the guide stuff in the animation. So suddenly characters can hit their marks lines can be delivered when you want not when the model wants and it won't go off and hallucinate in the corner too readily anymore. The final stage is done in software called controlnet which integrates both the low Rez animation as well as the depth map animation and integrates that with whatever control you want direct through prompting. It's an interesting twist on the old or 90s way of doing character and logo animation. But it's not as demanding and it's quite rewarding to see the AI doing exactly what you want not what it wants.

As a CMX editor in a grass Valley suite with an avid down the road I had to read tons of manuals and I kind of like the fact that this is professional caliber software. It's very user-friendly but it's a bit different than the old… Type on the computer what you want to do and press the button approach that many people feel make them a director. Anyway I hope I shared it in the right forum. I look forward to communicating with other folks going through this interesting transition that really represents a new paradigm in the industry.