Classical 3D rendering is so cooked. In the future, every pixel will be generated, not rendered. But until then, you get generative AI nodes inside Nuke. The sky's the limit when you blend classical visual effects with generative ai. We’re on the brink of fully generative content—where creation becomes... so abstracted that you can literally turn your mind inside out onto any canvas, from image to video to 3D. For now though, you get predictable curves and animation in the world’s greatest VFX tool — Nuke.show more

Github for ComfyUI for Nuke by Francisco Contreras, now with node animation thanks to @ClausSteinmassl

I know actors are worried about this. I tell my actor friends, imagine gathering a small team around you and together you can produce full feature length movies. You might not be an actor in a movie, but you get to make whatever you want. "Oh. Really? Like, full creativity?"

It’s a crucial reframe, and super well said

justice for isthisa3Dmodel

If you’re not using generative AI (like in this example) you’d traditionally use a parametric 3D approach to pull off facial retargeting like this — so it *does* have to do with 3D rendering. Hope the connection is clearer now.

I want to see a talented human actor use their skills, guided by a talented director and captured by all the talents that make cinema magic. If I wanted to see fake stuff, I would... No. I dont want to see fake stuff.

I totally agree, but I’m not even sure we will be manipulate puppet actors. I think movies will be generated, VR immersive if desired.

This is very impressive, but I wonder what our answer is to the fact that after AI gives you the "90% of the image" it needs to be tweaked for "the remaining 90%", which is usually what all the art directors / directors / vfx supes demand 🤔

I think that’s where hybrid workflows like this become so critical for that last mile / last meter control