In this project, I had the pleasure to contribute dust effects. This is something, I’d wanted to do since I watched breakdowns of loads of dust fx for ‘Prince of Persia’. So, I was really happy when Marijn asked me to work on this shot. As a refresher, I’ve included the final graded shot below:
View in full resolution: H264. (more…)
Erik is letting me, Marijn Eken, write a guest post on his blog to introduce a series of posts about a music video we collaborated on.The music video was for the official song of the 2011 Asian Cup, by Jay Sean. You can see the completed video here.
Jay Sean feat. Karl Wolf – Yalla Asia ft. Radhika Vekaria.
For this project I was the visual effects supervisor, overseeing the whole production from concept to delivery and also doing the color grading. I was also on set for the filming, which took place in Doha, Qatar. It was shot on the Red One camera, which gave us the nice option to send around trimmed R3D files to the artists. We used ShotRunner to communicate among the group of 8 people that worked together over the internet.
There are over 100 vfx shots in the video, of which 64 green screen shots and we completed the work in essentially two weeks time. But, for this series we’ll focus on one shot that was a particular challenge.
The client’s idea was to have a pack of dogs run through the desert and turn into a car. They suggested a morph, but I couldn’t see how four dogs morphing into a car would look good. So I opted for having a dust cloud form from the dogs and kind of conceal what actually happened and have the car appear from the cloud. Here is the finished shot.
View in full resolution: H264.
The first challenge was how we would shoot this. Doing this in two passes (one with the dogs, one with the car), would require some kind of motion control to get the same camera move twice, which was not an option. So I quickly realized we would have to do the car in CG. The only real elements would be the desert and the dogs.
I had planned to use golf balls to aid the match-moving. But the reality of the shoot proved I couldn’t use them. The dogs were going to chase after a gazelle. Yes, a real live gazelle. So this made it quite impossible to plan out a route in advance. Luckily the desert floor was hard and cracked, which gave us some features and I just had to hope we could track it later.
We ended up getting very little usable footage. The part we picked had only one dog in it, but we needed four. So we ended up duplicating the dog in the final shot. Linus Hofmann did that and will elaborate on this in a later post.
The footage was particularly shaky, so we needed stabilization. We didn’t want to run it through some stabilizer and then track it, because that would probably make the match-moving harder and less accurate. Marco de Goeij tracked the original shot and provided me with a solid track. What we then did, was use a 3D method for stabilizing, rather than a 2D warp. Since the desert is essentially flat, I created a simple piece of geometry inside Nuke to match that and projected the original plate from the tracked camera onto that surface. I then duplicated the camera and filtered it’s path to create a much smoother camera path and rotation. Looking through this smooth moving camera, we got our stabilized shot. This was the camera that was passed to the others to create all the CG elements from. See the difference between the original and stabilized shots here.
View in full resolution: H264.
It was a bit of a challenge to work out how we were going to have all the elements for the shot come together, as multiple people at different locations and with different software were involved. Richard Levene would render out the CG car in Maya and hand over the animated car model to Erik, so he could use it as a hold out for the dust simulation in Houdini. Linus Hofmann had the job of bringing it all together in the compositing in Nuke, which required a fair bit of tweaking to get just right.
It was a challenging but ultimately fun project to collaborate on. The quality of the end result amazed the client, which is alway nice to hear.
It’s time for another update on the explosion. It’s been silent for a while now, as I’ve been working on some other stuff while running simulations and renders in the background.
As a side note: I’m really pleased with the way HQueue is working on my little farm. It makes simulations and renders much easier to schedule and manage, so its was well worth a little investment in time to set it up. I’m thinking of writing a little post on my experiences with HQueue so far some day soon.
Anyway, on with the fragments and fluids. I’ll start with a render as usual:
View in high resolution: H264.
This is a render of the latest simulation that I have run on both the ‘faces’ and ‘energy fluid’, it also includes the ‘sparks’ particle sim. As with the previous renders, I ran the render passes through my very basic multi pass setup to tune it a little and add some glows. (more…)
Now the exhaust wall stuff is done and made into a tool, attention has shifted back to the explosion animation and simulation. By now all the major systems for animating and hand-over to simulation are in place.
Let me start by showing you a test animation for that I’ve rendered out. I did a very quick multi-pass comp on the render to attenuate reflections a bit and add some glows. The animation is rendered at 8 times slow motion.
View in high resolution: H264.
Here’s another update on the ‘exhaust wall’ part of the ‘derezzing’ sequence. As I was getting the exhaust wall network up to a level I was satisfied with, I figured, why not make it a tool and apply it to the red bike easily. That seemed like a jolly good idea, so I quickly got to work.
First, however, let me show the results coming out of a test render of my new tool (8 times, slow motion). Note that this render focusses completely on the exhaust walls. The background is still a placeholder, so is the grid (too low res texture), and the red cycle animation definitely needs quite some animation love, especially when it hits the ground. It’s also the ‘raw’ render output, with only sRGB applied. But…. with all those disclaimers:
View in high resolution: H264.
To place the cycle crash in context, I’ve started animating a short sequence (roughly based on my ‘previz‘). This is basically what leads to the cycle crashing. My son was adamant that the blue cycle be the one that snuffs it, so that’s what happens. I’ve rendered the second camera angle with the frame rate cranked up by a factor of eight. This will allow for moving in and out of slow motion during this part of the animation.
Here’s a quick flat render of the animation:
Here’s a short update on part 2 of ‘Derezzing’ a crashing light cycle. My background render of the falling apart of the full body of the cycle (in 4 fragments) has completed and here is the result:
View in high resolution: H264.
The render crashed around frame 69 for some reason, so I restarted it from that point on, but had forgotten that I had added a ‘RDB Autofreeze’ micro solver to my dynamics setup. This is why there is a little jump in the faces lying on the ground towards the end of the animation. The cycle doesn’t completely disintegrate as the animation I had setup for this wasn’t reaching the rear end of the cycle, which is why it doesn’t completely fall apart there.
Anyway, this was done by basically turning on all fragments, so I guess I shouldn’t complain about these details. I’m really pleased by the result so far. This animation definitely nicely shows what’s happening to the fragments as they fall apart.
In the mean time, I’m animating a first go at the final animation sequence for the crash, which is progressing nicely. The basic animation path for the crashing cycle is done now, next up is creating the proper fragments, based on how the cycle hits the wall and direct each fragment into its initial path before the dynamics sim will take over. This is a bit more work than just throw the cycle into a collision object and let dynamics do all of it, but it allows me to completely control what happens during the crash and where each piece flies off to with what initial velocity.
Ok, time to get back to fracturing and animating…
Cheers,
Erik
Here’s part 2 about my ‘derezzing’ cycle animation. I’ve now more or less finished the setup for ‘disintegrating’ individual fragments that will explode of the cycle when it crashes into the exhaust wall of its opponent.
After some basic cleaning up of my script, I thought, “let me do a multi fragment test and see what happens…”. This reminded me of how awesome the procedural nature of Houdini truly is! I added another fracture plane to my breaker and held my breath while I ran a simulation of the cycle without any animation, just sitting there in 4 fragments and let the Tron universe do its thing 
The following image is the result that I found when rendering frame 23 once I stopped the sim there…:
I was rather blown away by this, I must admit, even though that may sound really narcissistic (lets just call it lasting boyish love of all things shiny and glowy). The only thing I did to the render coming out of Mantra was throw it into the comp I had already prepaired for a rerender of my original test animation which follows below (which is nothing more than a slightly rebalanced throw-together of the render passes). Note that the blue glowing particle stuff coming from between the fragments is not yet the explosion of fluid as seen in the real Tron Legacy shots. This will be a separate fluid sim.
Anyway, as mentioned above, I rerendered the first test animation through the current setup. You can see it below:
View in high resolution: H264.
The next steps involve quite a bit of keyframe animation of the cycle hitting the wall and the initial movement of the fragments before they are handed over to a dynamic sim. While working on that, I’ll run a render of the full cycle disintegrating in the mean time on my ‘farm’ (as in the single frame above), I just want to see it moving 
.
Until the next post…
Cheers,
Erik
Probably the coolest effect in Tron is the ‘derezzing’ of light cycles after they crash into an ‘exhaust wall’. This effect didn’t end up being part of our Tron test animation. So I figured, I’d give it a go and create one anyway
Anyway, I did a first test setup of a single fragment that’s exploded of a crashed cycle. To figure out what happens during the violent ‘derezzing’ I closely looked at the original VFX test. A few days ago I also found links to Digital Domain’s making off video, which has in the mean time been removed again. However, I did have a close enough look at it to get some idea’s on what to do with my version of this effect. Unfortunately, Tron Legacy hasn’t been released to cinema’s in my little nook of the world (have to wait until Jan 19), so I haven’t had the chance to get better look at all the examples in the movie itself.
As I’m wrapping up another fun project, it is time for some long overdue updates on my blog.
I still haven’t found the time/energy to make that ‘build a wave animation’ screen capture. This is still very much in the planning but I’m going to stop promising it on a concrete schedule as more work is coming up shortly and there is only so much time in a day.
In the mean time, I’ve got some really cool tool development just about to get started, extending my efforts in water/object interactions and bubble advection so far, into a toolset that is able to create some really nice looking splashes and foam effects (both above, on and below the water surface). I’ll be adding the same spray and foam functionality to the breaking wave and ocean displacement tools as well, which should be really cool!
So, plenty to cool stuff on its way!
With a bit of luck I’ll find more time to keep updating the blog in the coming weeks than I have the last month and a half.
Cheers,
Erik