Evaluation

Overview
 This project has really allowed me to explore and expand my knowledge and skills when it comes to camera tracking. I decided to keep the product fairly simple in order to focus on technical skills. Although there are so many improvements that I would have liked to make to the final video, I am fairly happy with the outcome of the project.

Workflow/New Software
As I have mentioned in the other tasks on this project, i feel that my project workflow is starting to form a good structure. On this project, because it was based around match-moving,  I had to adapt the workflow to accommodation this process. All three match-movers /camera solvers I tried work best with image sequences  rather than a movie file. So after the rough edit stage, I converted the shots into .iff files as they work well with Maya.
My original intention was to use Match-mover, however, having been unable to get good results I started exploring other packages. Nukex gave a good track, however, its export system didn't really work. I found that Boujou gave a good track and was able to work with the footage that Match-mover was unable to solve. Unfortunately  by this point I had already re-planned my idea to work in better filming conditions due to the problems faced with match move. I feel that despite needing a lot more practice to get the art of match-moving, the results I have achieved  were good for a few weeks of tracking practice.

Product
Based on the time restrictions and the steep learning curve that match-moving is; I am very happy with the tracks I have achieved. I had originally planned to use Mayas physics to make the domino effect, however, for some reason the computer I was working on kept crashing when I tried to do physics simulations, so I had to animate the domino chain using key-frame animation . While this was time consuming, it was not as time consuming as trying to find a new computer and re-setting up the project.
I changed my plan from a large scale exterior shot, to an interior shot to try and get a better track in match-mover. On moving to Boujou, I had decided that I would continue wiith the new sequences as I had made the most progress with that sequence. While the lighting of the 3D work is not idea and some more tweaking could be done in the composition stage to make the dominos fit better, I feel that the porduct has been successful in teaching me the basics of match-moving to expand upon in future projects.
Conclusion
Overall, I feel that this project has been successful. While the results have not been perfect, I feel the skills that I have learnt will allow me to develop with more practice. I've particularly enjoyed comparing the different match-mov softwares out there.

Final Product

Update

Having got some half decent tracks out of Boujou, I have now started to build my scene up in Maya  I am still keeping with the 'Domino' them. However, due to the change in location/ footage, I have gone for the traditional black and white dominoes. To make the scene more interesting and to really put the track to the test, I am building a 'Bridge' in between two desks and having the domino chain travel over it.
I had originality planned to simulate the dominoes using Maya's built in physics, however, when it came to simulating the computer kept crashing and throwing up various errors. Because I didn't have time to resolve this, I keyframed the dominoes to knock each other over. This doesn't look as realistic, however, the movement is so quick, the inaccuracies are not really visible.

Why Not 123D catch

When I initially planned the project, I had intended to use Autodesks 123D capture a dominos pizza box and texture it. I did several tests with the software in the early stages of the project (with limited results). As I have changed to normal dominos, which have a lot simpler shape, I decided just to model them normally. By doing this I had a lot more flexibility when it came to adding shaders to the object. While I realise that 123D catch is a huge leap forward in the modelling world, I don't think (at this time) it rivals hand modelling.

Exploring Different MatchMovers

After getting limited results from Autodesks MATCHMOVER, I started to explore some of the different software that would produce the desired results.
NUKEX
After using Nuke heavily on the last task, I knew that it did have a very good matchmover/ camera solver built in. I did try to get a track from my footage using this and despite the results being good, the lack of an 'export camera data' node meant that it was tricky and inconsistent to get the camera data out to MAYA. I found that was a similar problem with AfterEffects CS6 on last semesters project.

BOUJOU
I had heard very good things about the results that had been achieved in Boujou, and I have to say, I was not disappointed. The easy to use interface and powerful algorithmic produced results that put the other solvers to shame. As I was completely new to the software, I followed a very good tutorial from an industry 'profession'.http://www.youtube.com/watch?v=OntNstHgkZY  I found that this was so helpful that in the space of a few hours I was already getting decent results. It is also very simple to export camera data from Boujou to Maya (and pretty much any vfx software).

Tracking - MatchMover Not Great

Despite following some good tutorials for Autodesks MatchMover I havent been getting very good results with the exterior shots. I think this is largely down to the confusion of depth and low contrast footage. I have taken some more 'Controlled' interior footage to try and get a better track and after a lot of tweeking I have managed to get an OK track. However, i am going to try and track the shots in NUKE to see if I can get a better result.

Planning

I started this piece by drawing some rough sketches to try and work out what shots to use in the piece. As the focus was on well tracked shots, I don't want more than 3 different shots as it would then became more about cutting and diverting the viewer attention.
I am still undecided about whether or not I want to animate normal dominos or the gimicky idea of Domino's Pizza boxes.

I do however feel that an interior and and exterior shot will best showcase the skills.
To capture the footage I went for a walk around my accomodation complex and took some hand held video shots based on the initial shots planned above. I also went for some different shots that looked good on location. While I was on the location I took a few panaramic photographs in case I have time to do some image based lighting later on in the project.

Initial Reaction

I am actually quiet excited about this brief. I feel that it is the first brief that gives the freedom to concentrate on refining the technical skills in on of the areas that I am probably least experienced in '3D Tracking'. I have done a little bit of 3D tracking before in nuke and boujou. However, it has never really been the main focal point before. Because I want to focus on getting a well tracked piece, I am keeping my ideas very simple. Back before I came to university I had a domino simulation in my portfolio. Which, looking back on, was shockingly bad. So for this project I planned to track some dominos into a scene and use realistic physics to simulate them toppling over. It then came to me that I could make the piece slightly more interesting and give it a but of a gimmicky twist by using actual 'DOMINOS' Pizza boxed as my toppling object. This will give me a chance to get some modelling into the project or even use the object capture software 123Dcatch.