Crunch Time

I am in full production and no sleep mode. On wendsday I went and shot some footage for shot two and I learned a little something about interlacing footage. It is really hard to track with PFtrack. The reason being is that there are no solid points to track. I attempted to De interlace the footage and that helped a bit however I could not get a good track because of the motion blur it added during a critical camera move. To try to fix it I manually added user points however I was unable to track the footage. I will need to re-shot it this week.

To make up for lost time I pushed on assets. I was able to do a quick render and comp for shot 3. It still needs some work but it is getting there.

I also started work on the spaceship that will be hovering over the city. That and re-shotting the footage will be main focus this week. I hope to have all the CG assets done by Saturday and the footage done by monday night so that I can start rendering and comping my shots.

Shot 3 Test from James Spadafora on Vimeo.

I Have made some progress on Character animation as well. I used the OPSIPA script to create a control rig for my orc. It is robust and a time savor. Normally to create to joy stick controls seen on most rigs now a days you need to set driven keys on blend shapes at different locations on a control curve, which can take a while. The script preps your blend shape to create an intermediary to preserve your original network then provides a GUI for mapping your blend shapes to your control curve. It was also easy to create my own connection inside the hyper shade for the mouth open, which is controlled by joints instead of blends. I used the GUI I made to do a quick lipsync test.

Learning greenscreen hard way......

So this week I tackled the green screen shoot. I learned the hard way that it is super important to get the best plate you can to get the best comp. Here is the original footage:

GreenScreenTest from James Spadafora on Vimeo.

I started with primatte first since that is what we used in class. I really like the step by step process of the toggle however when I moved thru the steps I realized that I was having a real issue with here head. There was all kinds of dark noise close to here hair values. Because of this I was battles to keep the dark wisps of hair while trying to eliminate the noise. This comp did not come out at all. The only way I would be able to easily remove the noise was to re-shoot the shot. I gave the rotopaint a try to draw out a garbage matte however I would need to go in with a tablet every frame and draw it out . I decided jest to re shoot first thing next week or friday becase there was another issue with the footage I will get to later. here was the primatte test:




Next I gave the IBK keyer a try which I did not get to far in because I found that it is a strong keyer if you have a clean BG plate. Pulling a key from IBK would be jest as much work at Primatte with a less intuitive interface. Also I had all the same issues I had with Primatte with the noise around the head.


The last Keyer I used was Keylight. I liked this keyer because I was able to pull a usable key quicker then with primatte. the sliders where a lot more useful and intuitive then the sampling method that IBK and Primatte use. This is by far my favored keyer. I still had all the same issues with the noise and hair however I was able to find a good balance using the black point slider.




There where a few good things that happened in the shoot. My lighing turned out exactly how I imagined it in my mind. I may defuse the red rim a bit but over all I was happy. That intermediate lighting class really paid off. I did not get any spill because I worked far from the screen however this caused the issue with the noise because I was getting the dark areas of the screen which made the noise. To fix this I will drop the intensity of the Keno flows pointing at the screen and work closer. Another issue I had was tracking. The points I put on the screen where to few. as soon as the ladder was out of the shot PFtrack did not has the data it needed to solve the camera. I think I will need to put more points ether on the wall or in the scene to get a good track witch will be important for bringing the scene in and do the set replace.