The Scamp Project

Okay, big news. As you may know, this blog's been dead since I started it. I've been working on the final details of my project. Well, it's not completely mine, but nobody has to know. It's going to be a web series surrounding a stuffed humanoid dog. The dog will be completely digital while the background will be filmed using a Flip camera, and it will be put on YouTube in HD. In theory this seems pretty easy, but I've run into quite a few technical issues trying to accomplish this.


The first thing I came across was the camera itself – it wasn't really designed for this. Well it was designed for YouTube, just not video tracking or any type of image stabilization. I'll talk more about those two later. First thing I want to bring to your attention is it has no manual contrast adjustment. It really has no manual adjustments at all. This creates a problem because when you're moving the camera around a scene and it, say, passes a window, it re-exposes, wreaking havoc on your rendering rig, trying to blend the model with the background. This really brings you to the point where you have to start lighting every frame individually.


Still talking about the camera, it's image sensor is surprisingly sensitive to low light, considering it's price-point, but to get video quality good enough, you have to light rooms well. Really well. I find it interesting that as long as the camera doesn't seem to have to do too much of its magically contrast adjustment the image quality stays pretty good, but once the camera starts feeling that it has to change its contrast, the image quality greatly degrades. So my tip if you intend to use this camera is try to watch for when your camera decides it's smarter than you. I assure you, it isn't. Well, then again, I've never met you, you could be a real moron, so you be the judge of that.


Now getting the video back into the computer isn't too hard, except for the fact that FlipShare conflicts with every program on my computer. I can't have Maya open with FlipShare open, but that's not normally necessary so this is more of a pet peeve.


The first thing you need to do is convert your video into image sequences. You can basically forget about the audio. I'm going to be using a separate mic and recording with that. It's not that the mic on the camera's bad, it's just most of the time the only object talking will be the stuffed animal.

P.s. I use PNG images for my image sequence.


Now for tracking, I'm using a mixture of Maya Live and Voodoo. In a later blog post and tutorial I'll go in depth on the subject. But for now, I'll just leave you with it's not that hard.


Now for the digital stuffed animal – his name is Scamp. I fell back on trusty Maya and began modeling from some drawings that my sister drew me. I went through about four different versions of him until I came to the one I liked.

He's rigged using a really primitive rig – nothing special. Just mostly parenting joints to curves and a couple of constraints. The reason I went with this is because of the wide variety of movements he has to do. I'll just add stuff on an as-needed shot-by-shot basis to conserve memory.

So that's basically what I've learned up to this point, with an exception of a few details. If you want to follow this project from the video side of it, hang tight on this blog. To learn more about the digital side of it, my 3dTutorialZone blog and website will be updating with tutorials on how I got this done.

Thanks for reading.

And for those who just skipped to the bottom, don't worry. The next post will have more pictures.