[John Martin]

yo bro! for a shape-agnostic solution you absolutely need to discover an effect called VEGAS.

here’s the best method i studied for you:

1 precomp your object matte.

2 add the Vegas effect to the precomp.

3 set Render to Transparent.

4 ncrease Segments until the outline looks smooth enough.

5 adjust Width and Softness untill it looks right.

6 animate Length (or Rotation if thats the effect your after) to make the outline move around the object.

7 vegas follows the alpha automaticly, so if the object changes shape the outline updates by itself.

I hope you like this method and i wish you the best 🔥

[John Martin]

Don’t worry! There is an open-source free equivalent of the tool i told you,
you simply need to install this: https://ollama.com/download
and this: https://github.com/jonigl/mcp-client-for-ollama

and you need to have a decent gpu in your local system, for example a 3090.

then you need to install all of this!

I already did it successfully in a couple of hours, and i control after effects in a crazy new way, it’s like having Jarvis, no paid Claude or other services, needed, just pure local gpu power!

in case you need help installing this, send me a private message and we schedule a 2 hour session, thank you! 🔥

[John Martin]

Yup, you got it! btw, if you have Claude Pro, you can give it a reference and make it do the animation for you while you sleep, by using a protocol called MCP, here’s how to install and use it, you’re welcome! 🔥

https://www.youtube.com/watch?v=YTOwQj-n7ME

[John Martin]

got it! to replicate that you need to involve shape morphing with the path along with the classic properties, lemme send you the project file replicating the perfect motion 1:1, let me know if this makes you slightly happier 🔥

https://we.tl/t-gE73GVDvRrSgKUJK

[John Martin]

hey Luke! I recorded a custom tutorial with the solution and i included the project file for you to study the technique, hope you like it! 🔥

https://youtu.be/y2Py1Yukw5w

[John Martin]

hey Jason i recorded a custom tutorial just for you with the solution, hope you like it! 🔥

https://youtu.be/CFGMeKDCktQ

[John Martin]

hey man! send me your footage and the mustace png with transparency you wanna track on his face and i’ll record a custom tutorial for you with a mindblowing method, thank you!

[John Martin]

Hey man! i’m gonna keep it short and simple for now, you can try to undistort the footage first, displace with depthmap in 3D (Trapcode Form/Mir or Plexus reads depth as displacement), then innsert Tao geometry in that same undistorted 3D space, redistort the whole comp at the end.

This should Keep your camera solve’s 3D space matching the depth extrusion without double-distorting Tao’s render. lemme know if that helps even a tiny bit, took me a littl ebit to figure this out!

[John Martin]

Hey bro!

Here’s a 100% guaranteed method:

– Physically print every frame on paper at 12fps

Wrap each sheet and unwrap it, then Wear dirty boots and step on each sheet without destroying it

Scan each sheet at 1200dpi

Merge the files back as image sequence

Remove background , post,

That’s it

The truth is that ai now is perfectly able to recreate that look completely in the book, but it requires training a custom look model trained in real worn film artifacts, almost nobody knows how to do this (except me? lol just kidding)

Hope that helps, Good luck!

[John Martin]

Hey Lucas! the fix is to animate in 3D space (e.g. C4D or Blender) mapped to the actual cube geometry, then bake/render the faces flat, so transitions between faces are geometrically correct with no timing gaps.

Lemme know how it goes and in case we escalate this further