Amazing stuff! It's incredible how fast AI video has progressed. The consistent physical characteristics of the slime/gunge is very convincing and the victim/audience reactions are getting better!
Exciting to think where we'll be in a year and beyond. I can only imagine the potential for massive WAM scenarios in gameshow type settings that would be beyond the capabilities of IRL producers. I'm a fan of those 3D-rendered scenes with enormous mess quantities, so the thought of something similar with photorealism is a joy!
Jimmy20 said: Amazing stuff! It's incredible how fast AI video has progressed. The consistent physical characteristics of the slime/gunge is very convincing and the victim/audience reactions are getting better!
Exciting to think where we'll be in a year and beyond. I can only imagine the potential for massive WAM scenarios in gameshow type settings that would be beyond the capabilities of IRL producers. I'm a fan of those 3D-rendered scenes with enormous mess quantities, so the thought of something similar with photorealism is a joy!
Some of these clips were surprisingly convincing.
At the moment, continuity is still a problem with longer scenes - this is why all the clips are short. Prompting only works with very simple direction (e.g. "the young woman is getting pink gunge dropped on her head"), and longer scenes will have the same problems that more complex still images have. The way to make longer clips is to chain together short clips that start where the last one finished, and it's very easy for these to become disjointed.
My clips also start mid-action. The look of the gunge can be impressive, but it kind of needs the gunge to be in the starting image to come out looking good. It's possible to start with a clean image, but these don't work anywhere near as good.
Both of these problems can be overcome in time.
I think we need to be clear on this forum that this is something that will supplement the work of 'real' producers, and not replace or compete with it.
Love you, too
