After 8 months of development and hard work… it’s finally… completed. I’m very proud of what I’ve achieved this year.
Thanks to Interpol for the awesome song to which I now know every single beat and vocal of. Thanks to Lisa Millar who created the sound for my project, her job was to subtlely enhance the theatrics of the video, she did a great job. Thanks to Chris Barker, who as my teacher helped me out with those technical problems and kept me optimistic about a goo solution. Thanks for all the feedback as well!
Finally to my parents. Thanks Mum for keeping me fed and alive during crazy work hours and Dad for setting up all those computers for me to render on.
Leif Erikson would not have been possible without the support of these people.
I think I’m FINALLY finished. Thought I should give an overview of how the project went.
Problem: Running out of time.
Ok I admit, I should have started working solidly from the beginning (JANUARY), rather than say… August. Solution: START EARLIER.
Time was a huge factor and drawback which I have learned my lesson on.
Problem: Renders that didn’t match.
To save time, I started rendering bits and pieces before I had completely finished. Which meant when the rest was done, sometimes the new renders didn’t coincide with the old stuff, which was I’m-gonna-tear-my-hair-out-annoying when I had to rerender and find out what was different in the scene to the previous one.
Solution: Render all together or keep track of render layers.
Of course if time permits, I would have rendered it all in one go. But I should have at least kept track of render layers and stopped making new ones that made it confusing.
Problem: Too many freaking files. Solution: I’m not too sure there would be one.
Problem: MAYA.
Well…Yeah ’nuff said.
I might leave you with some stats of my animation:
1 song
4 minutes
6000 frames
4 computers worked simultaneously at one point
73,340 renders in 232 folders
100.2 GB of data
1 person in need of 1 serious break
Now if you’ll excuse me, I’m going to find something to do till I get tired at 4am. Hello, retarded body clock.
I have come to realised I really really hate fixing up renders, when there’s just the little things that are different. Spent the pass week fixing up little things. Oh-so-annoying little things.
Things like:
shadows that suddenly change in the scene because I unlinked a light and forgot to link it back when I had to render the rest of the scene,
little changes in the specular or diffuse or shadow, and it would take me hours to figure out what was different from the original to the newly rendered batch
particles that suddently stopped moving in a batch render (I had to rerender the end 20 frames one by one of leg particles because every subsequent frame after the first one would render exactly the same!)
the very last bit where the door opens: the door and the mannequin would not line up, too many hours it took to fix that
backgrounds that would suddenly move quicker than the objects, even when it is set to the RIGHT frame rate and I hadn’t changed the camera at all
so on and so forth…
The only thing that’s worse than a bad render, is rerendering said render more than once. BUT. IT’S. DONE.
For these eyes, I motion tracked the paintings an image of the holes in the wall in AE: instant creepiness! For added creep, I used a procedual eyeball and rendered out a short sequence of the eyes looking as she walks past.
Used the exact same method as the attack, but because this is a close up paint effects looked a little dodgey. So I animated an extruded NURBS surface along the animated curves and applied the goo shader onto that to get it looking good.
Finally after a whole semester procrastinating.. I have finally completed THE GOO.
Chris has been sending me source files throughout the semester as I have been saying early on, how daunting I found creating the goo. A couple of his files involved hair follicles which gave some really nice and easy dynamic movement. However, when I used this method in my scene, especially with the collisions on for each body part and the ground, the scene became very sluggish. I also found I didn’t have enough control on each hair, especially when the goo strands float between being agressive and passive. So I decided against hair, and chose to animate the curve points instead. This might have been more time consuming in the long run, but I had a lot of control.
After much, much trial and error, the following is the process I settled upon:
Created curves leading up to the important parts of her body (neck, torso, left and right arm, left and right leg). I rebuilt these curves with 25 spans and applied a modified intestinal paint effect to it. I rebuilt them back to 2 spans - that way the paint effect would still be smooth.
Created 4 Point on Curve deformers on each curve, so that a smaller joining curve could be constraint to the deformers. The end locator was then Point constraint to the nearest joint on the body.
Keyed each Point on Curve deformer and paint FX, making sure their movements matched the mannequin’s body movements.
Repeated process for the smaller point constraint curves, which I scattered randomly between the big curves. I created a cluster deformer in the middle of these ones.
For the goo that wrapped around the body, I converted the paint FX to NURBS or polygons (depending on the number of faces) and soft binded them to the skeleton. Painting the weights on the torso wrap was very tedious.
In the end, it probably took the better part of three days for this scene. And the process was very simple because I had some base scenes and methods sorted already. I didn’t have to be so scared of it.
Using blendshapes again, leakage was pretty simple. Just two NURB spheres, on inside and one oozing out. NEXT UP: goo attack - no more procrastination!
