35MM Color Fringing Simulation?

[VideoMilk78]

I was wondering how to create this effect with Avisynth?


https://i.postimg.cc/hPB0fSVt/temp-Image84jz-W9.avif (https://postimg.cc/NLdXnvN3)

It's especially visible in the top right however if you need a better picture I can provide one

[VideoMilk78]

I'm looking for Chromatic Aberration I somehow completely forgot to mention this in the first post

[DTL]

Separate planes in RGB - resize red and/or blue slightly - crop (add borders) - combine planes.

Something like
r_plane=LanczosResize(r_plane.width+2, r_plane.height+2).Crop(1,1,-1,-1)
b_plane=LanczosResize(b_plane.width-2, b_plane.height-2).AddBorders(1,1,1,1, r=1)

[Selur]

I have seen something similar when chroma was shifted by a field before deinterlacing,...

[VideoMilk78]

Separate planes in RGB - resize red and/or blue slightly - crop (add borders) - combine planes.

Something like
r_plane=LanczosResize(r_plane.width+2, r_plane.height+2).Crop(1,1,-1,-1)
b_plane=LanczosResize(b_plane.width-2, b_plane.height-2).AddBorders(1,1,1,1, r=1)

This is what I came up with, nothing special but it worked for me. Thanks for the basic layout.

Function ChromaticAbberation(Clip c,int "AX")
{
AX = default(AX, 1)

r=c.ExtractR().crop(0,0,-AX,0).AddBorders(AX,0,0,0)

g=c.extractG()

b=c.extractB().crop(AX,0,0,0).AddBorders(0,0,AX,0)

rgb=MergeRGB(r,g,b)
return rgb
}

[DTL]

CA (of magnification) is scale-based distortion. The glass elements from shooting (and film scanner) lenses have different refractive index for different light waves and Red and Blue colours as most distant from medium Green start to make different size images. This causes CA of magnification (there is also axis-position CA and with typical 'secondary' green-magenta spectrum it causes green-magenta colouring of edges of near and far objects from the focus plane).

The magnification-based CA is easy to simulate with slightly different sizes of R and B planes relative to G. The sign of aberration may be different - depending on the lens correction type and residual CA left. You may select at the reference images if R needs to be smaller in comparison with B or reverse. In your example at the top-left corner it looks B image plane have bigger size in comparison with R (or R+G).

Shift-based may be from very bad film transfer where RGB channels become spatially misaligned by all-frame equal shift to some direction.