The Ingredients
Before we get started, we need to take inventory. You will need:
The first thing you need to do before you start shooting is to make a storyboard. Using paper and pencil, draw a rough sketch of the various scenes in the movie. This will help you better understand the final output, not to mention help you plan the way you will shoot. Our video, for example, has two scenes-the first one with a shot of the car revving up and jumping off the table; the second is the fall itself. It is this fall that we freeze in time, pan around, and then continue.
The Basics Of Bullet Time
So how do these bullet time videos work, anyway? The effect is simple enough (well, on paper, at least) to create:
Ready… Action!
Our setup follows these guidelines, if a little liberally. The location: an abandoned meeting room in our building. As you can see in the picture, we’ve got eight cameras arranged in an arc on the floor to capture the car in the middle of its fall. You also need to make sure that all the cameras are focused on the same spot. To do this, place an object at the point where you expect the action to happen-where the car hit the ground, in our case-and focus each camera on this point.
Ideally, you should have identical cameras between the first and the last. This way, it becomes easier for you to take a still picture in the middle of the action and have a chance at capturing the same moment on all cameras, thanks to their identical response times. Realistically, though, you’ll be dealing with a bunch of different cameras with different response times between clicking the button and the actual taking of the photo, so video mode for all of them would be advisable-we’ll just extract the frame we want from the videos. Make sure all of them are recording video at the same resolution-320 x 240 is a likely common denominator.
As with many things, more is better-at 25 frames per second, a one-second bullet time effect requires 25 cameras (since we’re showing one frame from each camera). High-budget movies use at least a hundred. So don’t be too picky when soliciting partners in this crime!
Once you’ve got your cameras and scene set up, there’s nothing left to do but start filming! Since we’ve chosen video mode for all cameras, we don’t need to worry about recording at the exact same moment. The trade-off is that we’ll have considerable motion blur to deal with, depending on how fast your subject is moving. Canon cameras come with a “Fast Frame Rate” setting that counters this a bit. The cameras otherwise used for this kind of effect are extremely quick, recording more than a hundred frames per second as opposed to the basic camera’s thirty.
Turn on all the cameras (extra hands are a big help here) and enact your scene-this completes the easiest part of this exercise.
The Preliminaries
Once you’ve finished recording and copied all the video feeds to your PC (a good idea is to name the feeds camera01, camera02 and so on), the first thing you need to do is extract all the frames from the movies. You’ll require plenty of hard drive space for this. Start VirtualDub and open the first movie. Go to File > Save Image Sequence and select your options from here. To save time, check the box at the bottom to add this to VirtualDub’s job queue. Repeat this for all the video feeds, and when you’re done, hit [F4] to open the Job Control dialog, click Start to begin extracting all the frames and go get yourself a snack-the process consumes some time.
Once this is done, we begin the tedious process of identifying the perfect frame for the effect-ideally one with as little motion blur as possible-from each of the cameras. Once you’ve isolated these frames, rename them for easy reference.
The canon cameras captured our bullet-time moment quite well thanks to the Fast Frame Rate Mode
Bring It All Together
Start Windows Movie Maker (WMM-it’s bundled with Windows XP under Start > Programs) and click on Import Video from the column on the left. For now, import just the feeds from the cameras for the bullet time effect-we’ll deal with introductory/concluding footage later.
Once done with that, click Import Pictures and import the frames you selected.
Now drag the selected frame from camera 1 on to the timeline. To herald the beginning of the effect, we displayed the frame for a good while. We’ve also applied a Sepia Tone effect (right-click on the frame in the timeline > Video Effects > Sepia Tone > Add). This gives us a cool Max Payne 2-like effect, but more importantly, it masks the (very visible) colour differences in the frames from each camera. Do the same for the frames from the rest of the cameras (except the last one), but reduce their presence on the timeline a lot more, so that they are displayed for less time. A good way to judge is play back the video-if it looks like a slideshow, maybe you should display frames for a shorter time. We’ve stretched the last frame out too, to mark the end of the effect.
After you’re done placing frames, drag the feed from the last camera on to the timeline. You don’t need any of the video before the frozen frame, so navigate to that frame, hit [Ctrl] [L] to split the clip, and delete the first part. There… you’ve done it! Save your project and your video. Be sure to save in the AVI format; VirtualDub doesn’t open files in the WMV format.
Post Production
Now, we’ll mask the imperfections in the video and also add the remaining footage. Open the video in VirtualDub. Go to Video > Filters and click Add. From the filter list, choose Temporal Smoother. This will add some intermediate frames in the bullet time video to make it look smoother. From Video > Compression choose the DivX encoder and save the video.
Go back to WMM and import this smoothed clip and the rest of your storyboard-related footage. Unless you need to fine-tune your videos, you can get the job done with the simple Storyboard mode in WMM. Overlay a title, add a soundtrack-anything goes as long as you aren’t bored. And you’re done!
Pack Up
You can find the result of our endeavours on this month’s DVD. It won’t win any amateur video awards, and here’s the biggest reason why: falling cars are unpredictable-as you can see, our car didn’t fall where we wanted it to, resulting in a jittered effect rather than a smooth pan-around. If you have more control over your subject’s position, you’ll wind up with a better effect.
So the next time you’re at home in the rain, don’t waste your time on the idiot box-make a few calls and start your own directorial venture! And if you send it to us, we’ll even feature it on the CD/DVD for all to see!