3d Titles like Panic Room

[Brian Manning]

Hi guys,

anybody got links to After Effects tutorials that deal with intro titles that look like they are part of tha action?
They have lights and shadows etc?
On Videocopilot.net there is one, tutorial no 71, if I recall, but I couldnt get it to work properly.

In the excelent movie, Moon, by Duncan Jones the titles are similar.

[Cole McDonald]

Make a null, motion track it to a point in your shot. (two if there's any camera rotation - then I'm not sure what to attach the second track to)
Drop in your text, skew it to match the angles of the buildings in your shot
Parent the text position to the null.

The text should now move with the buildings.

They do these titles on "Fringe" and "Warehouse 13" as well.

[Brian Manning]

right you are, that makes sense.
What about the other finer details-the shadows and light source on the text? - from what I gather you have to make a solid layer for the ground and another for the wall for example for the shadow to hit.Would you also have to motion track the 2 solids to their positions? that would explain why i lost my shadow before in a moving shot.

[Cole McDonald]

You can add all that stuff together into a precomp, then parent the precomp,, that will make it all move in synch... or just parent them all separately.

[Brian Manning]

Ah right - u mean create the text withextrusion and lights and shadows cast onto a couple of solid layers then pre compse that and track it to the footage?

[Ian Holb]

Are these static shots you are putting the titles on?

If not, then you'll need a match moving program like PFTrack or boujou, and then export the data to After Effects or, in your case, a 3D app to integrate into your scenes. I remember reading one of the biggest challenges for the effects artists on Panic Room was giving the 3D elements a sense of weight and space so that it actually looks like it was in the scene rather than superimposed over the film image.

[Cole McDonald]

If the camera is hand held or what you're comping needs to be comped at multiple depths, then yes, you'll need a more complex solution. But a camera (even moving) that's only panning and tilting (even hand held) will comp just fine with a single track point... rotation adds a second point of reference and pushing/pulling requires a 3rd (or more) to calculate the spatial relationships.

I've put an airship semi-convincingly (it was a still) above a barn tracked with a single point to a camera move: http://www.yafiunderground.com/Video...s-Festival.mov (audio and CC issues have been fixed since)... this was for a 48 hour project and the airship was a single frame from a greenscreen model shot of a pirate ship with airplane wings glued on... and a large water balloon in a net suspended upside down to get the balloon to push through the net a bit and give a feeling of lift when turned right-side up.

The motion is applied using a single track after the animation is set as it would have been with a static camera (for the most part, this is my second go at green screen work and my first at any real animation).

[Brian Manning]

I have spent hours trying to find tuts that deal with this effect comprehensively. No wonder I had no luck, it sounds complicated. So , you need Boujou and a 3d app. like cinema4d and after effects? I dont think I have the time to research these various tuts for each app. and try combine them for the effect I'm after.

But Videocopilots tut 71 had the effect with a still image, so I guessed it wouldnt be so hard to put it to live action, guess I was wrong.

[Brian Manning]

I have spent hours trying to find tuts that deal with this effect comprehensively. No wonder I had no luck, it sounds complicated. So , you need Boujou and a 3d app. like cinema4d and after effects? I dont think I have the time to research these various tuts for each app. and try combine them for the effect I'm after.

But Videocopilots tut 71 had the effect with a still image, so I guessed it wouldnt be so hard to put it to live action, guess I was wrong.

Cheers for the info lads.

[Cole McDonald]

It actually should be no different with live footage, so long as the live footage is shot in such a way as to fit into the main footage... or animated that way... thinking either into the comp, or into the animation.

The first, you build/shoot a static shot with moving lighting to mimic the movement through space that will happen when integrated into the scene... just like #71, #77

The second, builds the comp, tracks the footage with 1 2 or 3 points (enough to solve any 3d move mathematically - so long as they all stay on screen) to stabilize, parent camera to it then exports the camera movement to the 3d package... where you build the 3d scene to match (#105), then render using the camera movement with alpha and drop into the composite. Once you remove the camera parent, the shot will go back to moving and the track sent to the 3d package should keep the 3d moving in synch with it.

If you have examples you've made that haven't worked, post them and I'll help you troubleshoot them... source footage and final please.

[Brian Manning]

Wow dude, thats really generous of you,
its just I ain't never used c4d etc before or boujou and unless I could sit beside u in front of a PC I wouldnt learn. If its somethingf straight forward enough like the lightsaber tutorials etc where no other apps are involved I might do it but.....
TBH I cant afford the time to invest in this.
much appreciated though, seriously

im just finally glad someone told me plainly whats involved, cos i been looking this up for a while with no result...

Brian.

[Cole McDonald]

I've been helped with stuff I didn't know, so I help with that I can ;) I've also never used Boujou, nor C4D (I'm a Shake and Blender guy on a mac). I know the techniques, the rest is fitting them to the particular footage one is given - or coming up with an alternative method... never give up though, there's always a solution, it's just not always obvious ;)

[Steve Kalle]

For future reference, signing up to FXPHD.com for a term gives you free access to Cinema 4D, PFTrack, Maya, Nuke... just to name a few. Their Cinema 4D classes commonly discuss using C4D and AE together.

However, the cheapest 3D tracking and best for the money is SythnEyes (it was used extensively on Avatar).

[Cole McDonald]

OK, it's a quick track, with the first point I tried from a compressed and blown up video... but the point should get across.

First I took the clip and found higher contrast points that stay uncovered throughout the extent of the clip... I tracked 3 things (although only 1 is used for the text), as I was unsure what I was going to do with the track at that point. I first applied a plugin I wrote that zaps all the blacks up to a level and pulls the white point down to just above that (I call it my threshold keyer... used for pulling difficult luma mattes) to increase the contrast in the clip... this would work just as well with a contrast adjustment. We'll be removing it after the tracking is done.
http://www.dvinfo.net/forum/attachme...1&d=1281928694

Next, I tracked the point I've isolated with the contrast adjustment... in Shake, it's just a tracker node, in AE, it's a tracker effect applied to a Null object with the same net result. Track on little tracker...
http://www.dvinfo.net/forum/attachme...2&d=1281928694

Repeat for the tracking points you want to use, I've named them so I can remember what the heck they are.
http://www.dvinfo.net/forum/attachme...3&d=1281928694

Next, I create a precomp (or a node branch in shake) of the text I want and drop that over the main footage. Corner pin this (move the corners as if they were on a piece of wood resting where you want it in the scene, flat angled, whatever, here's where you have to think in 3d rather than the 2d image you're working with). Twiddle them until it looks "right".
http://www.dvinfo.net/forum/attachme...4&d=1281928694

and I went ahead and tracked a roto matte onto the wheel chair so I can have the driver roll over the text (again, quick and dirty for demo purposes)...
http://www.dvinfo.net/forum/attachme...5&d=1281928694

The result:
http://www.dvinfo.net/forum/attachme...6&d=1281928694

The text is tracked by a single point in the comp... tracking multiple points and attaching the corners to them would get much better results, but this is quick and dirty with low resolution footage from youtube. If you were to mark points on set with green painters tape that you want to use to corner pin, you could get absolutely perfect tracks by controlling what you're shooting on set, what goes into the camera, what goes into the edit/fx and therefor, what comes out with much more precision.

There's no magical solution to some of this, at certain zoom settings, there's quite a bit of barrel distortion at the edges/corners of the frame that you have to account for manually here... there are perspective changes that need to be accounted for here (the wheel chair matte narrows horizontally as the camera sees more of the back of the chair than the side). I didn't spend any time key framing those adjustments here, but another half-hour to an hour with the original footage and this shot could look phenomenal.

Font size, style, face and color can also have a huge impact on your audience. Smooth curves and text that stands out and promotes readability will also have a huge impact on this effect. Using text from a 3d app with lighting that moves with the scene when comped in, will look even better as it'll have volume and look like it's part of the scene.

Hope that helps.