So uh, my bad again. Clearly I'm not very good at this whole regular updating thing. But now there's a lot to put up here so I suppose I should get to it.
At the end of my last post I had just created a rig test, which meant it was time to animate.
First I played around with matching cameras between After effects and Maya. I found methods to export a Maya camera to AE and vice versa, and I found that it was much easier to make the camera move in After Effects and then bring it over to Maya for animating than the other way around. This way I could see the whole scene moving with the camera rather than just the character.
I used an AE script called 'AE3D Export' created by Ryan Gilmore which allows you to select your AE camera and some 3D objects and export them to a maya scene. Then I imported my character and also created an image reference plane showing the moving background so that I could see the character in the scene through the maya camera.
I did this for the 3 shots containing 3D that had complex camera moves. There were two other shots which included the 3D character but had no camera move so I didn't bother with a camera export for those shots.
Once I had all of the Maya scenes ready I went about animating.
This wasn't a particularly difficult piece of animation, though it did take me a little while to wrap my head around the idea of subtle realistic movement. Since Oscar is a real person, not some over the top animated character I wanted to make his animation realistic. I had to figure out what kind of subtle movements he would make even when he wasn't really 'doing anything.' I also had to consider the forces acting on his body, he isn't standing on anything so obviously something is making him float/fly. I decided to treat it as if he was under water, this meant that he was being held up by his chest/torso, and his limbs could just sort of float around on their own. IK was great for this. I was able to move around the model's core control without the feet or hands changing position which made for some fairly convincing movements. I also wanted to consider individual poses, much like you would in 2D. When thinking about movement it is easy to forget about silhouette and pose. I liked the poses in our animatic so I wanted to stay somewhat true to them. So I started with key poses, letting the program in-between them however it wanted.
Then I started tweaking movements. I wanted movements that flowed through his body in an elegant, almost dance-like way. To do this I staggered key frames, for instance I would have a key for the pelvis, and 5 frames later a key for the lower back, 5 frames later one for the rib cage, then the neck, then the head. This way you could see a movement flow up the spine. I used this technique for the arms and legs as well. I also attempted to make use of the 12 animation principles. Mostly I focused on anticipation, follow through, and slow in/slow out.
Of the 5 shots the first two were the most complex. They were close up and demanded more precise animation. The latter 3 were long shots and had fairly simple movements so these were much easier and faster to animate. Then I rendered out the final animation for each shot with the textures.
After animation I moved on to simulation. I had done tests right at the beginning of the year so I generally knew what I was doing. I decided that I should do a two pass effect. One pass of voxel based smoke, and one of simulated particles. I made a 3D container for the smoke and set my character mesh as the emitter. I had a lot of settings to play around with and it took me about 3 days to work out the right look. I used a turbulence field and a wind field to get the smoke moving the way I wanted it to. I turned my settings on the smoke and the fields into presets and then copied them to each other shot.
Next I baked each of the simulations so that they would render faster, and set them up to render. To speed up the process I rendered on about 7 computers at school.
When the renders were done I went to start compositing them I discovered that each shot had something wrong that would require me to re-simulate and re-render all of them. In most shots the smoke was moving too fast, and in others it was moving in an undesirable direction or shape. So I tweaked all of them and re-rendered. In hindsight I should have done more test renders/playblasts before going to the final render in the first place.
Once those renders were done I was ready to do the particle simulations. This process was almost identical to the smoke process. I set up a particle node and set my mesh as an emitter, tweaked settings and fields for a day or two, then baked and simulated. This time I did playblasts along the way. Since I was only simulating about 500 particles at a time is was a lot easier on my machine than the smoke had been, so test renders weren't too difficult. Then I render-farmed again at school and the effect simulation was complete.
Next up was compositing and effects. This ended up being one of the longest processes for me. Compositing can be relatively quick if you're just putting and animation on top of a video or something like that, but I was essentially creating the entire world in compositing using the elements that had been built by myself and the team. I spent a very long time adjusting colours, blend modes, movements, timing, camera moves, effects, and lots and lots of tweaking for every shot. For the character effects alone I used 20 After Effects layers per shot to get the look that I wanted.
Here's a quick video of how that looks in action
I'd say that compositing alone probably took me at least 80 hours of work time, though I should mention I'm pretty terrible at estimating time. In the end I had rendered out between 4 and 8 versions of each shot, that was 54 videos all together not including all the test shots and versions of the whole short film. During this process I even had to do some flash animation for the creeping light effect (though it was pretty janky since I had to rush it to get everything done) because my camera tracking wasn't working for the originally intended 3D effect we had planned on using, and I figured flash would be faster than a rotoscope effect. I should mention that every effect in this short film was done by me. I did the shot smoothing on the live action footage, I did the eye opening effect (Oscar did the 2D animation of it opening, and then I did all the compositing) and the space fly through, all of the effects on the jellyfish, glow effects, particle effects, I made all the space backgrounds using Hubble photos, colour corrections, glitch effects, shaky effects on the creepy shots. All of it. The only effect-like bits I didn't touch were the title and credits, which were done by Oscar. I was meant to do new glitch effects on the credits but I completely ran out of time. So it was a lot of work to complete.
OH! and the phone effects. I did all those too.
I found a free 3D model of the Galaxy S5 phone, edited it a whole bunch to make it not look so janky, textured it and rendered it to match the shots, then I made a phone screen in Photoshop that matched the original screen on the footage, did some animations on it in after effects, composited that all together in the shots and added glow effects to the screen. I'd forgotten all about that hahaha.
Next I took all of those videos over to Premiere Pro to edit together. This didn't take too long. I just had to drag and drop videos over top of our animatic and add the new sound that Oscar had made and I was done.
So here's the final film. :D
In the end I had to hand the project in almost a week late, which I wasn't at all happy about, but It would have been practically impossible to get it in on time with the amount of work I ended up having to do for compositing. In hindsight I could have dealt more of that work out to other members of the team and also managed my own time a little better, but they were all working hard on their portions of this project and on other school work so I was reluctant to give them more. Also I just wasn't used to this amount of work myself. So while I possibly could have worked a little harder to get it in on time, I already felt fairly overworked and pushing myself harder didn't entirely feel like an option. In any case it got done and I think on the whole it looks really cool, though there are still things I'd like to fix up at some point.
So that's it. A semester of work surmised in a single one and a half minute video. Cool beans.
No comments:
Post a Comment