Think of motion control as an uber remote control camera that can perform any camera move and record every aspect of that move so it can be exactly performed over and over again, without any human error. This repeatability is the key to enabling many special effects shots you see in film, high-end commercials and music videos. When you shoot something utilizing this technique over and over again, it is known as a repeat pass shot. Moco (the cool shorter term for motion control) also has many other uses besides performing repeat pass shots, and those uses could also be incorporated into this idea, but for the sake of simplicity, I will be focusing on the repeat pass concept.

A very extensive video tutorial on motion control shooting can be found at www.CameraControl.com
The video walks through several real world examples and goes into nearly every aspect of moco.





To view one of the best motion control pieces online check out Maximum Vegas:MGM Grand

So what do can we do with this?
I’m glad you asked ☺
By shooting a subject NUMEROUS times in SEVERAL set ups, locations or scenarios we develop a library of video clips that all have exact matching camera motion. The subject could be a vehicle driving on snow, sand, mud, etc or really anything you can imagine. For the explanation of this idea I am going to present the subject as a female model for a fashion brand.

For our demonstration, let’s assume we will shoot the same model over and over again wearing every possible combination of clothes from this season’s catalog. She will perform a closely choreographed series of moves as she walks across the stage modeling the clothes. Our motion control camera will track her movements in a very dramatic and complex camera move. The model will get very close to the camera at times, with the camera zooming in to various close-ups of the clothes. Other times, the camera might be in extreme positions, on the floor or high overhead. The key here would be to create a constantly moving scene that would keep the user engaged.

The most important elements of these repeat shots will be the start and end frames of the moco camera and the start and end frames of the model. All of these shots will need to be able to seamlessly loop, ideally so that the camera and the model will start and end in the same position.

What to do with all these shots?
Once we have shot all of our repeat passes, the fun begins. After we color correct, clean up and do any necessary compositing, we will assemble one long video sequence. This sequence will include EVERY final shot of the model in EVERY possible combination of clothes from the catalog. The shots will be assembled back-to-back and the order of the shots, as well as their corresponding time codes, will need to be documented in an external file…. probably XML. To create our interactive experience, we will utilize two duplicate flash video streams of this final sequence.




Streaming makes it possible
The ability to jump to any point in time in a flash video stream with very little noticeable buffer time is key. By streaming our two final videos we can jump from 01:12:14 to 5:22:01 and begin playing rather quickly instead of having to wait for the over 3 minutes of video to download in between. Streaming enables to us to have a VERY long final sequence of shots where we can navigate to any section as needed.

A great example of a streaming FLV with built in navigation was the Sprint Connectile Dysfunction Site…which is no longer available
This site utilized a very large streaming FLV at a very high bitrate. The user was able to navigate from room to room by clicking on the video. The video then seamlessly transitioned to the section that the user clicked with little to no wait time or noticeable hiccup. Very cool if you had a chance to check it out. If I come across a more current example I will update this post with it later.

Mash the button (Interaction Design)
O.K. so now that we have our two streaming sequences all set up we can allow the user to interact with them. What I envision is an interface that would allow the user to select any item from the clothing catalog and either drag it directly onto the video model or choose the item in the nav and click between available options (colors maybe) to see them change on the model.

Below is a ROUGH prototype of this idea. You can click either the red shirt or yellow shirt and see the video seamless transition the clothes on the model. Cool right?

How does this work?
When you really start to think about this idea and you see some of the amazing moco work that is being done you can really go crazy with ways to expand the idea. The guts of the idea however involve only two key concepts:

Repeat pass moco shots that seamlessly loop.

Streaming the video so that we jump to any timecode

Here is a basic diagram of the idea of the seamlessly looping video clip

Within each section of the video stream, the first and last frames of the section exactly match. Both the camera and the model are in the same position. If the user does not make a selection to change what the model is wearing by the time that section reaches the end, the video stream will simply loop back to the beginning and seamlessly loop.

If the user DOES make a selection to change the model’s clothes at ANY time during the video playback a few things occur.

We first find out the current timecode of video stream #1.

We then look at our XML file of sections and timecodes and find out where videosteam #2 should cue too based on the user’s selection

We then tell video stream #2 to go to the corresponding timecode for the new section.

The diagram below shows video stream #2 cueing up to the matching timecode in the yellow shirt section to where the user has clicked in the red shirt section.

Once video stream #2 is synced up, we then transition it in, over top of video stream #1. In my prototype, I use a simple dissolve for this transition, but this could be anything that reveals the new matched up video over top.

What makes this transition so cool is the fact that the only pixels or elements that change on the screen will be the model’s clothes. The background and any other elements will remain the same, all in full motion, since both shots were moco shots. FUN!

This technique would repeat itself as the user made more selections. The incoming video stream would cue up to the corresponding section and transition on. Since the clips are streaming, we could cue up to a section several minutes downstream.

Imagine the possibilities
Hundreds of different outfits all seamlessly transitioning on the model as she walked through varying terrain or interacted with other models would be awesome. I haven’t even mentioned exporting the camera data from the moco rig for use in CGI or motion tracking or the new HD video support if streaming flv….more ideas coming soon. ☺

I have had this idea for quite a long time and get very excited every time I think about. This idea is one of the main reasons I started IDEAfont. I hope that someone out there can make this a reality and I hope that I can be a part of it.