Impressive! Fun to watch, packed with information with a chill music in the background! Thanks for this one.. curious to see what's inside the plinko group😂
So then in theory, you should be able to animate paint strokes, or a roto shape with say a heat distortion effect and actually have some moving and pulsing clouds. No?
You can give animation a try, but the Inference node has no clue about the prev/next frame. I expect some flicker to be there... Kinda like a stop motion effect. Give it a try!
Very nice video though, if it takes 8 hours to calculating, isn't it better to create on Houdini? It might takes less 8 hours include modeling and rendering time.
It take hours to train the model *once*. 3D software needs to render every shot. CopyCat makes more sense when you have lots of similar shots to accomplish. Training a machine learning model for a single cloud makes no sense, of course. :-)
this is sooooooooo useful, finally a video explaining everything to perfection!!
this is a great tutorial, great visualization of what copycat is doing, and a great way to use the tool - thank you!
thanks, now I get an understanding of how copycat does its calculations . very well explained
Pretty amazing video! This explanation of how copycat works is the best I've seen so far. Thanks once again!
Glad it was helpful!
Super tutoriel!! Je comprends beaucoup mieux maintenant. J'essaierai de l'utiliser au travail pour le cleanup
Impressive! Fun to watch, packed with information with a chill music in the background! Thanks for this one.. curious to see what's inside the plinko group😂
A bunch of cylinders, 5 cubes and a ParticleGenerator for the balls with ParticleGravity+ParticleBounce
thaks so much, great video!.
Excellent!! well explained the process through Plinko Board demonstration, thanks for your amazing efforts👏👏
Cool explanation, impressive
Hey, thanks!
This is great! Well, it moves the weights along the error gradient, not quite randomly. But I love the analogy, this is great!
So then in theory, you should be able to animate paint strokes, or a roto shape with say a heat distortion effect and actually have some moving and pulsing clouds. No?
You can give animation a try, but the Inference node has no clue about the prev/next frame. I expect some flicker to be there... Kinda like a stop motion effect. Give it a try!
Very nice video though, if it takes 8 hours to calculating, isn't it better to create on Houdini? It might takes less 8 hours include modeling and rendering time.
It take hours to train the model *once*. 3D software needs to render every shot. CopyCat makes more sense when you have lots of similar shots to accomplish. Training a machine learning model for a single cloud makes no sense, of course. :-)
@@pixelfudger I understand when I may need this method, thanks!
Man you just keep fudging those pixels haha really great video. How did you make that plinko board though?
The balls are particles with a ParticleGravity and a ParticleBounce.