[Mark Woloschuk]

Oh my… that’s way fancier than I would have imagined. But the important part is that it works quite nicely.

Thanks Dan.

P.S. I’m still first in line for that book/video expression training I just know you’re going to spring on us some day.

[Mark Woloschuk]

Thanks Darby but that still doesn’t fix it. When I access the tD value it still doesn’t return a value that makes sense.

I’m aware of the falloff settings in Sound Keys (and indeed am using an exponential falloff) but in this case I need the fade to happen in the layer not in the sound key data since this piece of audio moves so fast.

Imagine someone playing the piano very fast and every key they hit lit up as soon as they touched it but faded out more slowly.

Unless I’m missing something fundamental about Sound Keys, I think the way to do it is within each layer’s opacity values.

But thanks for the effort!

– Mark

[Mark Woloschuk]

Thanks Mike,

For some reason I never got an email letting me know you responded. We were doing a mockup of an art installation for a grant proposal and time was short so I ended up handing this task off to a flash person but I like your solution better.

Thanks again – I know it will come in handy for many things.

[Mark Woloschuk]

well you ask a sort of non-specific question so i’ll give you a sort of non specific answer.

i find that using the orthographic views (ie. left, right, top etc.) helps because they constrain movement to only 2 axes and it’s easier to smooth out motion path beziers in only 2 axes at a time. also i find that it is often better to turn off the camera’s orient towards point of interest (layer>transform>auto orient) in some cases or to orient along the path in others. parenting to animated nulls is another technique that is often way easier than trying to animate a camera (especially for orbiting paths and shots where you want the camera to maintain a fixed distance from objects.

finally i’ll give you the very obvious advice to turn off lights and shadows and use low res proxies of footage (or solids, whatever) to avoid the rendering hit of 3D comps and make your system more lively and responsive while you try to get your moves right.

hope that helps.

mw

[Mark Woloschuk]

turn on the view masks toggle (next to the safe areas toggle on the comp window) so you can see the mask and then use the arrow tool to click on a mask vertice. this will reveal all mask vertices – selected ones are solid and unselected are hollow. if you don’t click carefully AFX will select multiple mask points in that case shift click to turn on or off the ones you need. double clicking on a mask edge will bring up transform tools that will allow you to move, scale and rotate the entire mask. hit enter to get out of transform mode.

i will point out that it is way, way* easier to rotoscope with multiple masks (ie. one for the head, one for the left arm, one for the upper torso etc. etc.) that have fewer points than one big complicated one. it also helps because you can apply different levels of feathering to the different masks. also you may find it helpful to switch your mask to rotobezier mode (select your mask and find the option in the LAYER>MASK menu) and if you’re footage is interlaced (eg. 29.97i) to change your comp settings to 59.94 so that you’re applying keyframes on fields instead of frames.

mw

* way, way, way…

[Mark Woloschuk]

try to pull a key of the hot spots and then fill them with color fills that match the skin tone at that spot. then overlay some combo of noise and/or fractal noise, ramps etc. so that the color isn’t totally flat.

it might be better or it might just look weird but its worth a shot.

mw

[Mark Woloschuk]

i’d reset your comp to 59.94 (assuming you’re working with 60i video) to set mask key frames on fields instead of frames and then render out at 29.97. also it would probably be easier to use several masks, one for each squeegee stroke although i guess that would depend on your source footage.

hope that helps.

mw

[Mark Woloschuk]

a friend of mine used a similar technique in an awards video. basically he exported his video as a series of still frames and then used a pict to ASCII text utility (you can probably find one on tucows) to convert the images. he then copied the resultant ASCII text into illustrator files so that he could zoom in and out of frames.

easy, neh? bear in mind that this will work best with subjects shot or keyed against a white background and that to do it for 30fps over several minutes would be a mighty tedious endeavour. perhaps you can figure out some clever way to batch the process.

mw

[Mark Woloschuk]

i would also point out (heh – very apt) that you can just turn off auto-orient towards point of interest or enable auto-orient along path to stamp out the damn point of interest altogether. i can’t believe i didn’t know you could do this until i imported a match move camera from boujou.

previously i used to tie the p.o.i. postion to the camera position with this expression:

[position[0], position[1], position[2]+X]

where x = the difference between camera and p.o.i. z position

which is a decidedly lame way to do it.

mw

[Mark Woloschuk]

i would also point out (heh – very apt) that you can just turn off auto-orient towards point of interest or enable auto-orient along path to stamp out the damn point of interest altogether. i can’t believe i didn’t know you could do this until i imported a match move camera from boujou.

previously i used to tie the p.o.i. postion to the camera position with this expression:

[position[0], position[1], position[2]+X]

where x = the difference between camera and p.o.i. z position

which is a decidedly lame way to do it.

mw