SYDNEY MARIE SMITH

View Original

Dr. Strange VFX Recreation: Crunch Time

May is rapidly approaching and we’re in the home stretch, something that is both a comfort and a looming threat to me. Nevertheless, I tried my best to keep the momentum I regained last week going, working on iterating and refining my existing elements as well as adding the final element I was missing to my shot.

Lucky for me, the cobblestones fixes for this week were pretty minimal— mostly just fixing the issue with them floating after they are dropped by the RBD sim, which I figured out was an issue with the padding of the collision and the way it was being read as a convex hull. I also think I fixed the material issue with the cobblestones that happened last week, since for some reason they came out strangely low-quality-looking compared to previous weeks. Some experimentation showed me that apparently packed vs. unpacked geometry takes on materials differently, so I went ahead and unpacked all the cobblestones before sending them to render to get it looking correct again.

Magic-wise, I mostly focused on the overall look of both the streams and the sparks. The sparks I also adjusted the influence of the advection volume to add back some more varied organic movement, but the main time I spent on them was adjusting the size and color variation of the particles. One easy major fix in that realm was adding alpha/brightness variation to the longer-lasting particles, since I realized I was only applying the alpha variation to the shorter lifespan particles that shoot straight from the stream. For the magic stream, I ended up adding a noise to the color distribution to add some variation through more bright and dim sections of the stream, which Nelson and I noted was actually what the stream in my reference looked like as well.

The existing element I probably spent the most time on this week was the falling dirt. Even though it wasn’t looking terrible in previous weeks, it still wasn’t quite looking right, so I spent some time adjusting the vellum grains behavior and the way I was dividing it into chunks. I made the clusters of grains smaller so they looked a little less obviously geometric and unnatural and also made the grains themselves a little bit bigger. After the sim, I added a noise to the Cd value to incorporate later into the material applied to the dirt to give it some shade variation so all the grains wouldn’t be the same brown. I specifically turned the dimensions of the noise down to 1 so that I would get a black and white grain pattern that would only affect the brightness or darkness of the material, but not change the overall color.

Finally, it was time for me to add the final element: the cobblestone debris! Although it is a quite small and almost unnoticeable element, I think it’s still one of those little things that makes a larger impact in the end of how real the simulations feel and how the energy is translated in the shot. My setup was pretty simple for creating the debris, essentially just scattering points on the cobblestones and copying a voronoi-fractured piece of debris onto them, and finally adding some variation to the pscale and rotation of each piece. The only challenge with this setup was dictating where and when points should be instanced for debris to potentially fall from, since I specifically needed the debris to come from the bottom half of the cobblestones only when they’re moving across the curve.

This proved to be a lot less simple than you’d imagine, with the animating geometry of the cobblestones not allowing me to group the points I wanted the debris to emit from very easily. After testing a couple of methods, I found the most efficient way for my purposes was to create an attribute from a group on the original geometry of the cobblestones before they are copied and animated, then transfer that attribute onto the final animated cobblestones.

Aside from the actual VFX work I did this week, I also took some time to create an animated camera that matches the movement in my reference for my final shot, as well as a new static cam that matches the end position of the camera in my reference to use for the time being. Additionally, I once again took another shot at fixing the lighting, which I think I’ve gotten at least a little bit closer to what the final product should look like this week. All of that brought me to this work in progress shot for the week:

Crunch time at the end of the semester is always scary, but right on time, I am starting to gain some hope that I’ll be able to complete this project by the end of the semester. The next couple weeks should be just final adjustments and comp work, so here’s to hoping I can make it to the finish line in one piece!