[Shin Kurokawa]

” Speaking of Twixtor, does anyone else find it’s stability a bit flaky with Premiere? ”

Have you tried “Plenty of Memory” unchecked? (i.e. OFF).
Also, I’ve seen it freeze up if I define the color
source / secondary source in T4.5, then move tracks
around, but it doesn’t happen all the time.

” Any chance of the tutorials getting broken up to more manageable file sizes Pete ?”

As for *these* tutorials, the main reason why
they’re big is that they have a lot of footage
(incl. full-NTSC/PAL sized clips,
even though they’re very compressed and shared
by the html guide as well as the project
files for various host apps). They also have
a lot of illustrations as opposed to formulas
and such, to add some Physics-For-Poets feel.:)
These were originally done for the recent
NAB swag CD.

Also, not a step-by-step tutorial, but smaller
sample projects are available at
https://www.revisionfx.com/rstwixtortutorials.htm
All the way at the bottom of that page.

-Shin

[Shin Kurokawa]

This is on which host app?
(Shake? AE? C*? flame?)
Because you set up shapes differently
in each app.

Basically, in the Flex suite when
you draw the masks you always do it in pairs:
* A: “From” shape. followed by…
* B: “To” shape.
Perhaps the easiest way to do this is
to copy-paste the 1st mask you draw
(which is a “From” shape). The pasted
shape is a “To” shape. You could then
animate the “To” shape.

Note that it maybe easier and more
flexible (harhar) to split up
a large shape into smaller shapes.
It may get very labor-intensive to
define all the corresponding features
(esp. in motion morph). A simple motion
morph might need at least 10-20 pairs,
more or less depending on the features
involved.

There is an exception to this, where
you draw a shape which you might need
to use as a ‘boundary’ which tells
FLex “don’t warp anything beyond this”.

So, you might end up with a collection
of shapes like this:
Shape1from
Shape1to
Shape2from
Shape2to
Shape3from
Shape3to
ShapeB==== boundary (assuming defined)
Shape4from
Shape4to
Shape5from

Also note in this example that the last
shape “Shape5from” is unpaired with its
corresponding “to” shape. In this case
the shape has no effect whatsoever.

-Shin

[Shin Kurokawa]

This is on which host app?
(Shake? AE? C*? flame?)
Because you set up shapes differently
in each app.

Basically, in the Flex suite when
you draw the masks you always do it in pairs:
* A: “From” shape. followed by…
* B: “To” shape.
Perhaps the easiest way to do this is
to copy-paste the 1st mask you draw
(which is a “From” shape). The pasted
shape is a “To” shape. You could then
animate the “To” shape.

Note that it maybe easier and more
flexible (harhar) to split up
a large shape into smaller shapes.
It may get very labor-intensive to
define all the corresponding features
(esp. in motion morph). A simple motion
morph might need at least 10-20 pairs,
more or less depending on the features
involved.

There is an exception to this, where
you draw a shape which you might need
to use as a ‘boundary’ which tells
FLex “don’t warp anything beyond this”.

So, you might end up with a collection
of shapes like this:
Shape1from
Shape1to
Shape2from
Shape2to
Shape3from
Shape3to
ShapeB==== boundary (assuming defined)
Shape4from
Shape4to
Shape5from

Also note in this example that the last
shape “Shape5from” is unpaired with its
corresponding “to” shape. In this case
the shape has no effect whatsoever.

-Shin

[Shin Kurokawa]

Hmm. Not sure what’s going on
but for me VideoGogh renders
in about 30~50 times real-time, meaning
60sec of footage would take something
like 1800-3000 seconds. This is on
AE/C* w/ PC/Mac, NTSC D1 frame size/
rate.

Or, did you mean there’s an instance
of VG for every layer? Have you tried
rendering intermediates, then apply
VG?
Are you using the latest version: 2.7

BTW: if “Render Current Over Last” is
checked, you’d get unpredictable results
if you submit the job to multiple nodes.

-Shin

[Shin Kurokawa]

Hmm. Not sure what’s going on
but for me VideoGogh renders
in about 30~50 times real-time, meaning
60sec of footage would take something
like 1800-3000 seconds. This is on
AE/C* w/ PC/Mac, NTSC D1 frame size/
rate.

Or, did you mean there’s an instance
of VG for every layer? Have you tried
rendering intermediates, then apply
VG?
Are you using the latest version: 2.7

BTW: if “Render Current Over Last” is
checked, you’d get unpredictable results
if you submit the job to multiple nodes.

-Shin