• Wondering which camera, gear, computer, or software to buy? Ask in our Gear Guide.

Photoshop or After Effects or both

I know this would take a long time, but is the following possible in either photoshop or after effects. If it's both, which would be easier?

If you took a 30 second clip from final cut (roughly 900 frames?) and wanted to blur the background, would you have to do each frame separately, or is there a way to capture the person in the first frame and copy it into the others? Is this impossible because each frame is different?

I'd like to try to add some depth to a few scenes.
 
You could do it frame by frame in photoshop, but since you'll
have to cut out the subject in each frame, you may as well
use AE.

The way to do it in AE would be to duplicate the layer, then
mask out and rotoscope the subject on the top layer. You
may not need to go frame by frame since AE extrapolates
the in-between. Try masking the major movements and then
go back through frame by frame to tighten up where it didn't
do it well enough. (Also, CS5 has an auto rotoscope tool that
looks pretty cool, although I can't speak to how well it works
as I haven't used it)

After you've isolated the subject on a new layer, you can create
the blur one of two ways.
1. You can simply add a blur effect to the bottom layer
or
2. You can create a camera, make them 3d layers, move the
background back on the z-axis and then adjust the DOF and
aperture etc on the camera.

You might also want to create a mask on the bottom that is
the opposite of your 'subject' layer so that the subject doesn't
exist in the background. That way you won't have any weird
blur from your the edges of your subject that won't look right.

Hope that makes sense, and helps. :)
 
Basically, you take your source footage layer and CNTRL-D duplicate it. Then on the top layer, you will use the PEN TOOL to create a mask around the person or things you want to keep in focus. Then select the MASK layer and put about a 4-6 pixel FEATHERING on it, so it blends nicely into the background. Then you'll create a KEY FRAME for the MASK SHAPE. Select the lower layer and select CAMERA BLUR or GAUSSIAN BLUR. Then proceed one frame forward and adjust the pen tool mask to match the movement of the foreground person/object.

Then repeat for 898 more frames...
 
Basically, you take your source footage layer and CNTRL-D duplicate it. Then on the top layer, you will use the PEN TOOL to create a mask around the person or things you want to keep in focus. Then select the MASK layer and put about a 4-6 pixel FEATHERING on it, so it blends nicely into the background. Then you'll create a KEY FRAME for the MASK SHAPE. Select the lower layer and select CAMERA BLUR or GAUSSIAN BLUR. Then proceed one frame forward and adjust the pen tool mask to match the movement of the foreground person/object.

Then repeat for 898 more frames...

Well, that just sounds like HELL! Thanks. I have to move on and on and on to something completely new.
 
Then composting isnt for you :) Its a detail, obsessive thing.. I can work on 4 seconds of a clip for 8 hours!

I love detailed work, but this is just too much.

The distribution company already loves the movie, and they just want to edit it for TV (fingers crossed). But, I get big ideas in my head, and it's just worthless. If they really want to do something like that, they can have fun with it.
 
Yes, rotoscoping is always a pain in the butt...

It's actually MUCH easier if you break things up into several pieces, rather than just using a single mask. Create a mask for the head, one for the upper torso, one for each upper arm, one for each lower arm, one for each hand, etc... you can move their rotation axis with the pan behind tool to get it onto the pivot joint of each piece (base of the head for the neck joint, one end of the upper arm for the shoulder, etc) then most of the work is getting it roughed in by moving and rotating.. then you can go back and tighten it up a lot.

CS5 does have a new rotobrush that is supposed to make rotoscoping easier, but I doubt it would ever work perfectly unless the footage was shot specific for its use (strong contrast separation between subject & BG)

This tutorial might help make things easier as well.. (haven't watched it in a while I forget exactly how he did it in this one)..
http://www.videocopilot.net/tutorials/2d_depth_of_field/
 
Yes, the adobe demonstration videos are always mindblowing.. but like I said, I seriously doubt it works that well even MOST of the time, unless the footage was shot specifically for use with the tool.
 
Yes, the adobe demonstration videos are always mindblowing.. but like I said, I seriously doubt it works that well even MOST of the time, unless the footage was shot specifically for use with the tool.

That goes without saying. For demonstrations they always use something that's shot with those features in mind. It's like blue or green screen, demos look great, but if that screen isn't lit properly you'll have some ghostly shadow in the back ground.
 
Back
Top