300 battle sequence

Gabriel McArtney · May 31, 2007

There is a shot in 300 that I would like to try and replicate:

http://www.youtube.com/watch?v=kmW-Tnl6nbk

A short explanation of how they achieved this shot can be found in the April issue of American Cinematographer. Basically they mounted 3 high speed Arri film cameras onto one head, each camera with a different lens (wide, medium, and close). One camera was mounted normally, i.e. horizontally, and the other two were mounted vertically and pointed directly down into a prism or beam splitter, so all the cameras appeared to have the exact same camera angle. In post, they switched between the cameras at certain points using a synthetic morph/zoom.

My question is: how did they use the prism to align the cameras, and where can I learn more about this technology?

And what morph effect/plug-in was used in post to switch between the shots?

Thanks in advance

Gabriel

David Mullen ASC · May 31, 2007

I believe they wanted to use Clairmont's over-under rig but ended up just putting the three cameras as close as possible next to each other.

Will Earl · May 31, 2007

Have a look for Cinefex 109, it explains the shot in great detail. I'll paraphrase what's in the article...

The Clairmont 'Crazy Horse' rig was in South Africa and unavailable so they tried using another beam splitter rig which didn't work out as the high-speed motors vibrated the glass causing everything to go out of focus. As David just pointed out, in the end they just stuck three cameras as cross as possible to each other.

Animal Logic then nested the three angles into After Effects and digitally zoomed from one angle to the next using a subtle morph in-between.

Most compositing applications can do some form of morphing technique, you'll have to try them out for yourself, last time I had to do any morphing was with Elastic Reality - but I don't think it's sold any more.

Edited May 31, 2007 by Will Earl

Stuart Brereton · May 31, 2007

in the end they just stuck three cameras as close as possible to each other.

Are you sure? The April edition of AC has a couple of pictures of their beamsplitter rig in action, and Larry Fong himself explains how it was done.

Gabriel McArtney · May 31, 2007

Thanks everyone for such fast replies. I will check out the cinefex article.

Actually, I tried shooting this with the camera's as close together as possible today, and you can definitely tell the angles are different.

Also: why go to the trouble when you could just use a zoom lens to punch in at set moments.

And: how did they maintain resolution when they digitally zoomed?

Thanks for all the info guys,

Gabriel

Phil Rhodes · May 31, 2007

Hi,

They maintained resolution because they're morphing into a close-up, not just blowing in on an existing frame. I'm sure there's moments, if you frame through it, where things probably look a bit iffy, but it's how it plays at full speed that counts.

Sure, you could use a zoom to punch in. If you think you're that good. I don't think I'm that good.

Strikes me that this is the sort of thing that's an awful lot of work - both in choreography, camerawork and postproduction. Much trickier than it looks I don't doubt.

I thought it was fantastic, though.

Phil

Scott Fritzshall · May 31, 2007

AC and Cinefex seem to contradict each other on this topic, but I'm inclined to go with what Cinefex says. The effects guys explain how, because of the camera's offsets, they had to use projection mapping to re-photograph the sequence from two of the cameras in order to make them look like they were all from the same angle. It is indeed a whole lot more complicated than it looks.

Stuart Brereton · May 31, 2007

There is a later picture in the AC article with Larry Fong leaning on a rig with all 3 cameras side by side, so perhaps they did it both ways. It seems strange though, that Larry doesn't mention any problems with the beamsplitter in the article.

Larry, perhaps you can clear this one up for us, if you're reading.... :-)

Will Earl · June 2, 2007

Actually, I tried shooting this with the camera's as close together as possible today, and you can definitely tell the angles are different.

That's the reason for using a morph instead of a straight transition - a morph is a distortion plus a transition.

Have a read of this, it'll give you an a rough idea of how a morph works.

because of the camera's offsets, they had to use projection mapping to re-photograph the sequence from two of the cameras in order to make them look like they were all from the same angle.

I can't see any mention of projection mapping being used to create the shot, it does say they had to figure out the camera offsets to render the backgrounds for the MS and CU shots.