Shot 9 -11: White Sequence – was basically a disaster! This is one of the only shots which requires full motion of the camera in 3D space. A number of other shots use cameras in 3D environments, but the motion path is more linear in these shots. This one requires the camera to swoop in an S curve, passing the ferry on the way and twisting between a pair of mountain peaks at the end. I understand the theory of the shot, and learnt a lot about POI and XYZ just trying to make it happen, but it’s certainly more difficult than I thought. It’s not the action itself that is the problem, though that alone is quite complex – the main problem is timing. The shot is rather long and managing all those camera key frames against the ferry key frames is headache worthy. Earlier tests pointed to this as a problem – things start to get difficult when you have both a camera and the POI in motion at the same time. I had six different environment views up on screen but I still kept losing my subject. The most aggravating thing is that the camera sits well back from the workspace, so the toggles are often off the screen. You have to zoom WAY out to find the manual handles and often can’t see what you are doing. I can understand why industry sets up multiple screens for the same machine as I am so tempted to drag panels off the screen and out of the way. It’s one of those cases where the complexity of the program, and all its damnable options, cause more harm than good. After hours of fiddling I have achieved a rough path. The movement is a bit jerky and too fast, but it’s not something that is unfixable. The view of Russia has been axed, in favour of a smoother cross fade to red sequence. If I have time, I can adjust the end of the shot to put the Pushkin’s Palace back in, but right now it’s going to destroy the motion path I’ve managed, and I’ve had just about enough of swimming for the elusive Z handle under the mountain of layers (no pun intended – they are mountains!).
Shot 12: China Wall – I actually completed shot 13 first, and realising that I was going to have two heavy linear zooms back to back, I decided to head in the reverse direction for shot 12. The shot starts with Nellie close to the camera and then pulls out slowly. It worked really well. From a story perspective as we haven’t seen her up close for a while and she is turned away in the following shot. I installed a custom particle engine which shoots pink petals across the shot, and a few cross over the camera as it zooms out – by coincidence or no, some of the motion is really nice and this is shaping up to be one of my favourite shots. This won’t require anymore work.
Shot 13: China City Fly Through – I seemed to be on a bit of a roll on red sequence today and like shot 12, this one came together quickly and successfully. This made use of 3D layers zooming towards a static back plate. Artwork designed for a straight line zoom, but I didn’t expect it to be so easy after all the problems I had with white sequence. The dragon kite was originally going to “float” in the top corner but as I was dragging it onto the workspace to the required spot, it looked better moving in that direction! It now works in an arch across the shot, emphasising Nellie and the POI. Unfortunately, I couldn’t find a spot for the two Chinese lateens, which is a bit disappointing, but I guess you have to kill your babies sometimes.
Shot 14: Docks Shot – a simple motion path. I really haven’t much to say about this shot. It worked like a charm and took about ten minutes! I added a shift to the back plate so that it didn’t appear too static.
Transition b/w Shot 12 and Shot 13:
Initially I was constructing the shots in isolation, but I’ve realised more thought is required for transitions. I have them planned on paper but actually achieving it is another thing. I performed an integration test with shots 12 – 13, installing a yellow to black fade down and using the dragon kite and petals as bridging. From here on in I will add padding at the end of each shot and think about transitions more actively. A small amount of foreground animation generally helps to distract from a fade down or cross fade. It’s also better to use a black or white solid over the top of all the layers and fade this in or out, rather than changing the opacity of the layers themselves. The first run through I changed the layers rather than the solid and found that because of their size, they came in at different rates (this meant you could see the skeleton/ghost of cropped layers underneath the foreground layers, exposing the illusion of depth). The transition depended on a custom particle flow. I painted two types of pink petal, one curled and one flat, and rendered two engines on top of one another. I hadn’t used custom particles before and got a bit distracted playing with all the engine functions. This was definitely a worthwhile exercise as I have a much better understanding of what I can achieve outside of the presets. Going back to the very early animation theory revision, I made a point of offsetting the two particle comps (petal 1 and petal 2), staggered them so they appear more natural. I also hand animated a few petals that I wanted in the cameras line of sight.
This is probably one of the most important things I learnt when playing around with custom particles:
Start at Birth – Play Once
Plays the custom layer from start once. The particle may die before the custom layer ends, or the custom layer may end before the particle dies.
Start at Birth – Loop
Plays the custom layer from start and re-starts if the layer ends.
Start at Birth – Stretch
Plays the custom layer from start and time-stretches it so it ends exactly when the particle dies.
Random – Still Frame
Grabs a random frame from the custom layer and uses throughout the partlicles life.
Random – Play Once
Starts playing the custom layer from a random frame, and then plays at normal speed. The particle may die before the custom layer ends, or the custom layer may end before the particle dies.
Random – Loop
Starts playing the custom layer from a random frame, and then plays at normal speed. If the custom layer ends, it re-starts from the beginning.