HFC 2020 Production Blog - Misha

Worked on materials and lighting more. Render farm finally went back to normal time/frame towards the end of the week which allowed me to render out the creature segment. It’s still not perfect as evidenced by the half rendered images so hopefully they figure that out. Considering going back to glass for the creature’s head as the metal/pink part was done to cut down on render time and I feel like it makes their movements look more abrupt/jerky. Not super sure how to fix that otherwise. Not able to put as much time in as I would’ve liked but that’s kind of the nature of life at the moment.

Finally able to work again as of this past Friday! I’ve been working on materials for the whole segment and got as far as the part past the creatures, but the render farm isn’t playing nice with that bit yet. So far I’m pleased with what I’ve been able to do, will probably tweak the shader on the non-glowing part of the heart to see if I can’t get it to be less dark/more vibrant. Otherwise pleased with how things are going!

It’s really hard to actually do anything on this computer. I’ve attached a pretty common error I’m running into where my program does this and I have to close and reopen it (a process which takes about ten minutes on this laptop) to fix it. This file is very heavy, and that wasn’t a problem on the Monty computers, but its a huge problem for my older laptop that was pretty middle of the line five years ago. I’m trying my best to work on materials and lighting, but it’s slow going and in all honesty its pretty hard to find the motivation to fight through these constant technical issues.

This week was again focused on the timing and adjusting the mandelbulb so that it transitions better into the sphere segment, although I haven’t figured it out completely. I’ve been experimenting more with Mandelbulb 3D but still haven’t quite found what I’m looking for yet, as it is a new program and I’ve been very busy with other films and classes. I’ve also been experimenting with materials for the final render.

This week was mostly focused on adjusting the timing of my animation to better reflect the animatic and starting to seriously consider the background of the segment. For this I started looking into a program called Mandelbulb3D which is a fractal visualization software. There are so many options it’s a little crazy, but from some experiments I’ve done I created some possible directions to consider. The colors would be adjusted to better suit the palette in the final version.

Here’s a look at everything together so far. I’ve put in some rough transitions between the different parts of the sphere so that everything is smoother. I’ve also tweaked the creature’s spawning, the length of the tails, and the movement of the camera. I’d like to add more interest to the sphere as it turns into the lotus, maybe with a rotation or some kind of tile effect, just something to make it more fun. Next week I need to lock down what the creatures are going to do when they reach the glass.

This week we got a lot closer to the animatic. From here it’s mostly just fine tuning for the creatures. My efforts are focused on the tails at the moment, and ensuring that the creatures move more a bit more slowly.

This week I mainly focused on getting everything organized and consistent in the scene. I also got to work on creating the curves that the particles will advect along after they exit the lotus-heart, and a method of switching them from the intial velocity necessary for them to properly exit the heart to the curve based velocity required to get around the camera and to the glass.

This week I figured out how to get the creatures out of the heart. Rather than the squishy, explosive method, I thought the heart melting to reveal something similar to a lotus seed pod would look interesting, be more effective, and provide a nice callback to Mackenzie’s segment. The creature’s are also propelled out of their own chambers and into space, after which they will be advected along a curve as previously planned. This is good, as all the pieces of the first segment have been figured out and now I can move on to the neuron segment, or find an alternative.

We really wanted to just try different things and see if they work so we could further push the experimental nature of the film. When our time in the screening room came, the crew all worked super hard to get everything filmed and photographed that we needed, and I’m very excited to see what kind of material we might be able to pull from it.

Another Mandelbulb. This one was the result of switching the equation for the X value from cos(theta) sin(phi) to cos(theta) cos(phi). I really love the star patterns that emerge from this one.

After fixing this up so it worked in VEX (Houdini’s code language), I created the phase variable as a channel and specified that the value could only ever go between 0 and 2𝛑, as specified in the text. This allows me to key animate on that channel or possibly even write a function so that I don’t have to go back into the wrangle node. I’m very excited about this as I’ve been wanting to use fractals for a while and now that it’s actually possible to do, I can look into ways of further streamlining the process!

The final method I’ve decided on, advecting along curves. It’s faster and looks better. I’d like to eliminate the swirling, or at least at some more disruption so it isn’t quite so even.

Swarm intelligence test. Slight modifications from Junichiro Horikawa’s code. If I were to take it further, I would modify things more extensively. As it is currently it doesn’t seem to be worth pursuing.

Velocity field method.

This is just a quick test of what everything might look like together. 

This is much more robust, allows a greater degree of control, and allows many more creatures to be instanced without me having to individually draw and animate every curve, which would frankly be a huge waste of time. I’ve done a few tests with the synapse/neuron formation and I’m still a little unclear as to how I’m going to tackle that. The few things I’ve tried have been pretty unwieldy, and I need a simple solution if this is going to work on the scale I’m attempting.