Making a shrinking grid of video clips without going crazy?
December 14, 2010 9:12 AM Subscribe
I would like to make a 40 second long outro to a promo video, in which the clip gets smaller, and as it does so, other clips around it come into frame. So, it starts with one clip, shrinks to a grid of 9 clips (3x3 grid), then to 25 clips (5x5 grid), and then the whole thing shrinks to reveal the image behind it. I have an idea that doesn't hurt my head to think about, but what's the easiest method? I'm thinking of doing this. Better explanation inside.
The only thing I can think of that doesn't totally hurt my head, is this image, which is trying to explain this:
VIDEO A
Take 9 videos that are 1280px x 720px each. Make an fcp sequence that's (1280 x 3)px by (720 x 3)px. Put all nine in frame in a 3x3 grid. Export movie.
VIDEO B
Make an fcp sequence that's 1280px x 720px. Import Video A at 100% (i.e. so only the middle clip of video A is showing). Over 20 seconds, reduce Video A to 33% (i.e. so all three clips are showing). Export.
Each clip in the grid in Video B is now (1280/3)px x (720/3)px = 427px x 240 px.
I would take the remaining 16 videos (25 - 9 = 16) and resize them using compressor to 427px x 240px.
VIDEO C
Make an fcp sequence that's (427 x 5)px by (240 x 5)px = 2133px x 1200px. Import Video B at 100%, and add the other 16 videos around it to make a 5x5 grid. Export.
Video D
Make an fcp sequence that's 1280px x 720px. Import Video C at 100% (ie, so only the middle nine videos are showing). Over 20 seconds, reduce video C to 60% (i.e. so all five videos are showing). Export.
Phew!
I would then have a video where one clip became 9 clips over 20 seconds, and then those 9 clips became 25 clips in the next 20 seconds. Would this be smooth? And seriously, is there another way?
The outro I'm trying to improve is the one at the end of this video (starts at 3:25). Thanks!
The only thing I can think of that doesn't totally hurt my head, is this image, which is trying to explain this:
VIDEO A
Take 9 videos that are 1280px x 720px each. Make an fcp sequence that's (1280 x 3)px by (720 x 3)px. Put all nine in frame in a 3x3 grid. Export movie.
VIDEO B
Make an fcp sequence that's 1280px x 720px. Import Video A at 100% (i.e. so only the middle clip of video A is showing). Over 20 seconds, reduce Video A to 33% (i.e. so all three clips are showing). Export.
Each clip in the grid in Video B is now (1280/3)px x (720/3)px = 427px x 240 px.
I would take the remaining 16 videos (25 - 9 = 16) and resize them using compressor to 427px x 240px.
VIDEO C
Make an fcp sequence that's (427 x 5)px by (240 x 5)px = 2133px x 1200px. Import Video B at 100%, and add the other 16 videos around it to make a 5x5 grid. Export.
Video D
Make an fcp sequence that's 1280px x 720px. Import Video C at 100% (ie, so only the middle nine videos are showing). Over 20 seconds, reduce video C to 60% (i.e. so all five videos are showing). Export.
Phew!
I would then have a video where one clip became 9 clips over 20 seconds, and then those 9 clips became 25 clips in the next 20 seconds. Would this be smooth? And seriously, is there another way?
The outro I'm trying to improve is the one at the end of this video (starts at 3:25). Thanks!
Response by poster: The only other solution I can think of is to create one fcp sequence of 1280px x 720px, and work out where to place all 25 videos, and how to make them shrink smoothly - making sure they drift inwards at the right rate. Is that the answer? My head hurts!
posted by omnigut at 9:23 AM on December 14, 2010
posted by omnigut at 9:23 AM on December 14, 2010
This is a natural for Adobe After Effects and some clever precomping. If you're new to this sort of thing, I'll hazard a guess that you would find this somewhat overwhelming to accomplish on your own.
posted by dbiedny at 9:44 AM on December 14, 2010
posted by dbiedny at 9:44 AM on December 14, 2010
Response by poster: So, if I go for the math version, do you see any problem with this?
Lets call the width (1280) "X" and the length (720) "Y"
Make all videos X by Y pixels
Make an fcp sequence that is X by Y pixels.
Add the first video to the top-left corner. So, it will start off-screen, at -2X and -2Y, @100%, and end up at -1/5X and -1/5Y @ 20% (in other words, the top left corner of the visible frame). I've just gotta work that out for all five videos, and put them all into the same clip.
You think fcp will be able to handle that? (Mac 10.5.8, 2GHz intel Core Duo, 2GB memory...)
posted by omnigut at 10:17 AM on December 14, 2010
Lets call the width (1280) "X" and the length (720) "Y"
Make all videos X by Y pixels
Make an fcp sequence that is X by Y pixels.
Add the first video to the top-left corner. So, it will start off-screen, at -2X and -2Y, @100%, and end up at -1/5X and -1/5Y @ 20% (in other words, the top left corner of the visible frame). I've just gotta work that out for all five videos, and put them all into the same clip.
You think fcp will be able to handle that? (Mac 10.5.8, 2GHz intel Core Duo, 2GB memory...)
posted by omnigut at 10:17 AM on December 14, 2010
Do you have After Effects?
Were it me doing this I'd use AE and do it all in one 3D move like so:
Create two nulls with one as the child of the other. This will allow you to use cartesian coordinates rather than absolute coordinates. The parent null will have absolute coordinates but the child null will be at 0,0,0 at center screen. You won't touch the parent null. The beauty of cartesian coordinates is that to position an element precisely to the right you merely offset by the width of the video. To the left is the same number except negative. Up is negative and down is positive as well. Not much thinking required.
Make each of all the 25 video elements a child of the child null. The heirarchy is now three levels deep:
Parent Null (absolute position) --> Child Null (linked to Parent Null): --> each of the 25 1280x720 video elements (linked to the Child Null)
To position, leave the first piece, the piece you are transitioning/zooming away from at center screen (0,0,0). The piece to the right of center gets coordinates (1280,0,0), the piece to the left gets the reciprocal (-1280,0,0), the piece above gets (0,-720,0), the element below (0,720,0).
You now have the center five elements positioned. Add the 20 others using the pattern. For example, for the element that lies 2 widths to the right of the center piece the coordinates will be (2560,0,0) and to the left (-2560,0,0). The upper righmost element in the grid of 25 will be positioned at (2560,-1440,0) and the upper leftmost at (-2560,-1440,).
You'll never need to make a move on or with the individual elements as that will all be handled by a single master move done with the child null.
You build the move on the Child null starting at position (0,0,0) and translates/moves in the Z-axis (zooms away from the viewer) to about (0,0,7108).
You'll have your entire grid of 25 elements moving in perfect unison.
posted by bz at 12:42 PM on December 14, 2010
Were it me doing this I'd use AE and do it all in one 3D move like so:
Create two nulls with one as the child of the other. This will allow you to use cartesian coordinates rather than absolute coordinates. The parent null will have absolute coordinates but the child null will be at 0,0,0 at center screen. You won't touch the parent null. The beauty of cartesian coordinates is that to position an element precisely to the right you merely offset by the width of the video. To the left is the same number except negative. Up is negative and down is positive as well. Not much thinking required.
Make each of all the 25 video elements a child of the child null. The heirarchy is now three levels deep:
Parent Null (absolute position) --> Child Null (linked to Parent Null): --> each of the 25 1280x720 video elements (linked to the Child Null)
To position, leave the first piece, the piece you are transitioning/zooming away from at center screen (0,0,0). The piece to the right of center gets coordinates (1280,0,0), the piece to the left gets the reciprocal (-1280,0,0), the piece above gets (0,-720,0), the element below (0,720,0).
You now have the center five elements positioned. Add the 20 others using the pattern. For example, for the element that lies 2 widths to the right of the center piece the coordinates will be (2560,0,0) and to the left (-2560,0,0). The upper righmost element in the grid of 25 will be positioned at (2560,-1440,0) and the upper leftmost at (-2560,-1440,).
You'll never need to make a move on or with the individual elements as that will all be handled by a single master move done with the child null.
You build the move on the Child null starting at position (0,0,0) and translates/moves in the Z-axis (zooms away from the viewer) to about (0,0,7108).
You'll have your entire grid of 25 elements moving in perfect unison.
posted by bz at 12:42 PM on December 14, 2010
Oh and make sure that the 3D engine is turned on for all layers/elements including the nulls.
posted by bz at 12:46 PM on December 14, 2010
posted by bz at 12:46 PM on December 14, 2010
Response by poster: Okay, so for your enjoyment, I've made a spreadsheet of the math I worked out for this problem. I've made the spreadsheet for a 5 x 5 grid of videos, but it could easily be changed for a 3 x 3 or 7 x 7 grid.
It lets you plug in the width and height of one of the images, the starting x and y coordinates, the end x and y coordinates (in case you don't want it to zoom in or out to the middle of the screen), and the percentage you want the image to reduce to (you'll have to work out the percentage yourself).
The graph automatically puts that information into a table. Hopefully it'll help you. If you have any questions about it, mefi mail me.
posted by omnigut at 3:28 PM on December 15, 2010
It lets you plug in the width and height of one of the images, the starting x and y coordinates, the end x and y coordinates (in case you don't want it to zoom in or out to the middle of the screen), and the percentage you want the image to reduce to (you'll have to work out the percentage yourself).
The graph automatically puts that information into a table. Hopefully it'll help you. If you have any questions about it, mefi mail me.
posted by omnigut at 3:28 PM on December 15, 2010
Response by poster: By the way - that helped make this test video
posted by omnigut at 3:40 PM on December 15, 2010
posted by omnigut at 3:40 PM on December 15, 2010
This thread is closed to new comments.
posted by omnigut at 9:19 AM on December 14, 2010