% Laurence Boissieux Home Page

Creation under After Effect

First install the watercolorization plug-ins. Be sure to have the following dlls installed in your system:

msvcr80.dll
msvcp80.dll
msvcm80.dll
msvcr80d.dll
msvcp80d.dll
msvcm80d.dll 

The advection consists in the computation of 8 passes: fast regular, slow regular, fast reverse, slow reverse with the first texture and the same with a delay and the second texure. Then they will be combined through 2 processes into one advection layer. Fast and slow allow to take into account the different movements in the video. Thus we specify a period of regeneration for the advection texture. If the motion is quick then we have to regenerate more often, so small values (10 15 frames) are used, for medium motion the best is 30 and for low motions, use 60 80. That's for the theory. In practice, there are some slight problems using some of these values on the boundaries (especially when we apply the process with a delay....) So the best is to use fast=10 with slow=30 or 60 for sequence that clearly contain different types of velocities, and 30 for both slow and fast for medium motions (it then decreases the number of passes form 8 to 4) .

A tip is to create the compositions for all the passes in one After Effect project, add each composition to the rendering queue and then start the batch rendering for the 8 passes at once (then you can go and take a coffee meanwhile).

First pass: fast regular (aka norev)

Second pass: slow regular (aka norev)

In our example, we use 10 for fast and 30 for slow and get Cube_610_norev.mov and Cube_630_norev.mov



Cube_610_norev.mov (240 frames)
Cube_630_norev.mov (240 frames)

Third pass: fast reverse (aka rev)

Fourth pass: slow reverse (aka rev)

In our example, we get Cube_610_rev.mov and Cube_630_rev.mov



Cube_610_rev.mov (240 frames)
Cube_630_rev.mov (240 frames)

Fifth pass: fast regular delayed (aka norev_d)

Sixth pass: slow regular delayed (aka norev_d)

In our example, we get Cube_610_norev_d.mov and Cube_630_norev_d.mov



Cube_610_norev_d.mov (240 frames)
Cube_630_norev_d.mov (240 frames)

Seventh pass: fast reverse delayed (aka rev_d)

Eighth pass: slow reverse (aka rev_d)

In our example, we get Cube_610_rev_d.mov and Cube_630_rev_d.mov



Cube_610_rev_d.mov (240 frames)
Cube_630_rev_d.mov (240 frames)