So it has already been done!

The question had been previously asked whether anyone had used lights to track motion, which is something that we had originally intended for the match moving portion of the project as a last minute alternative to to tape marks. After doing some Youtubing, it turns out that this studio had infact used a similar technique that we had intended to use, building a L.E.D. rig to triangulate the position of the cameras. Somehow they had also incorporated iPhones into the mix, which seems to be a common trend in almost any hip project these days.

Test Footage of MatchMoving and Masking Processes.

Test footage of match moving and masking processes used in the special effects project.

Masking

So this is the technique that I have determined will be best for special effects project. This is actually a combination of two techniques.

1. The first technique is to rotoscope the arm on the places that they are on the black background, then you can simply place a black solid on the from underneath and you have yourself a missing arm.

2. The second technique needs  a bit more effort. When the arm moves in front of his uniform it becomes necessary to paint each individual frame in photoshop. Since the video slows down to 59.95 frames per second it makes it quite  few frames. For this I export the exact portion that needs to be painted, then paint it in and bring in the image sequence.

Getting the Hang of Match Mover

Autodesk MatchMover is most certainly one of the most aggravating programs ever created. This is certainly not the forum to air my grievances however. I will tell you what I have learned about the program and some of the small kinks in it's system that will really get you down, as well as an update on my progress.

Some issues and tips-

1. Never Use Automatic Tracking- I found that the most effective way, even though it may be time consuming, to track the footage is to find individual points of contrast and track them using hard tracks. I found that if you define a beginning point, then go forward and track back from there to fill in the gaps, it can streamline the process. The automatic tracking function is more of an accuracy by volume approach as we would have called it in the Army. It projects sometimes thousands of points that are far from accurate and that leads to a shaky and inaccurate camera.

2. Shift Select in the timeline selects a whole chunk of points, Control Shift deselects them. I got so hung up trying to deselect the dragging selection box......lets just say it took a long time to figure that one out.

3. Camera Solving- You need at least 4 points to at any given time in the timeline to solve for the camera as long as you know the focal length of the lens that the film the movie was shot with. If you don't know then the program needs 7 computations points at any given time. Luckily for this past project I had collected camera data and measurements from the film crew so it made it far easier.

Below is a screen shot of just one of the parts of one of the match moving clips. All of the tracks are hard tracks focused on the pieces of tape placed on the backdrop.

Shading Update

Crossposted from my progress blog:

I will be developing a procedural shader for the shatter effect until Week 7. Here are the ideas I've brainstormed: ruby, ice, glowskin, cement, parchment, brick, tribal tattoo, volcanic... To give you an idea of the aesthetic of the film, here is a screenshot:



I created three tests so far. Tomorrow I present them to Kirk and the group. If they don't like them, I will try more of my ideas. The tricky thing is that I was originally imagining glass or something with a lot of caustics, but the way that the short is lit doesn't really provide an environment for good reflections. Here's an example of the predicament:



The ice shader that I created has this problem of fading into the background due to dark reflections. One solution is to give it an ambient, glowing quality. I'm afraid that that may look really cheesy and unrealistic. Another solution may be to rough up the ice, like frosted glass.



Next is volcanic. I like this one! It borders on the edge of totally cheeseball, but I think it fits the martial arts/nature/silent rage thing and the red contrasts nicely against the blue tones.



Lastly, I'm calling this one tribal tattoo. Before anyone accuses me of reusing the spirally thing! - yes, I did. I think a spirally, organic shape ties in to the theme really well. I plan on developing it more (you can still sort of see that it's a marble shader...), but I like the idea. For this one, i would LOVE to go crazy and make the tattoo an animated snake or dragon or something, but I don't think it's possible with the way the UVs will be. Check out the animated test to see what I have in mind for the transition:





It's also my job to figure out how these textures would transition from normal skin in a natural, interesting way. I thought I had it down pat with an ambient occlusion node that would creep the texture over the edges, but it wasn't working as well as I'd hoped. If that's the style we want, I'll have to find another way, but the direction I am going in now is more like the test above.

Meeting with Programmers

I met with Ben Cumings yesterday evening to talk about how things are going so far.  They've decided to stick with an FBX file format for importing and exporting.

Interface:
We no longer need to give them any animation, just points of impact and amount of force.  They should have it set up to allow multiple points of impact to allow us to create different kinds of effects.  When they've finished we'll sit down with them and they'll guide us through their interface until we get the shattered geometry we're looking for.

Timeline:
By the end of week six they will be done with an implementation that works for everything except the UV map and normals.  At the end of week 8 they'll have that stuff finished.  This will be great for our testing period.  We can try using some shattered geometry they give us and see how it works without the proper UVs and normals.  it may work out being more appropriate to simply use Maya's built in shatter algorithm at this point, we'll have to see how it goes.

Video Edit Complete

This post is a bit late, but the video edit is finished.  I've uploaded all of the footage and Adobe Premiere Pro files to our rendershare.