The first and easier part of the project I tackled this week was shooting the actual footage I’m going to composite my VFX on top of. As you may have read in my first post about this project, one of the new things I wanted to try to elevate my composting skills was trying to integrate my VFX on top of real life footage instead of just a static backplate. So, with the help of my very lovely and helpful friend, Vy Lam, I was able to shoot some video footage of myself mimicking Annie’s scene in The Nevers! Be sure to check out her website to see the awesome work she’s doing this semester for this independent study as well. 

Now I am not an actor by any means, my main goal in recreating this footage was just to get the movements and timing as close as I could to my reference shot--though I did also want to extend the timing of the overall clip a bit since it’s a little short for my liking. So, with a camera, an ipad and iphone with an orange image on them to act as lights, and a dream, Vy and I were able to film a short clip for me to use.

Once I got the raw footage, I went ahead and did some basic compositing work on it, doing some color grading and fixing the levels to make sure it wasn’t too dark or too bright--basically doing my best to match the overall aesthetic and feeling of the shot from The Nevers. I also did some rotoscoping through After Effects to remove Vy’s hands holding the precarious light sources, and played with different ways to integrate the footage of myself into the background. And with that, my base shot was ready to go for some pyro magic VFX!

With nothing else left for me to distract myself with, it was finally time for the simultaneously more fun and also more painful part: revisiting my pyro simulation. As you might remember from my last post, my pyro sim was at a questionably functional state as of last week. Lucky for me, Nelson was able to take a look through my file this week to help me clean up my nodes and fix the biggest issues inhibiting me, and boy was his walkthrough of the fixes he made helpful!

Firstly, if you are interested in working with volumes in Houdini and you have just one takeaway from this blog post, please let it be this: always turn on clamp coverage in your volume rasterize attributes node! This one check box basically solved the biggest issue I have had since the beginning of my exploration of pyro sims, which was my volumes and subsequently my fire coming out very blocky. Turns out if you just check this one box, it automatically makes your source volumes look so much nicer and doesn’t try to fill each voxel with volumes. Just take a look at the difference between these two volume sources:

Look at that, smoother and less blocky volumes with the click of a button!

Along with this quick fix, Nelson helped to adjust my particle simulation to be shorter and punchier to better match my reference and also fixed the pscale of the particles I was generating to turn into my voxels—another very small detail that makes a huge difference in the look of your simulation.

With Nelson’s revisions in mind, I went back to work on my simulation to play with the newer and more functional pyro sim that he helped me achieve, focusing on different methods I could use to improve the shaping of the fire burst. One thing that Nelson suggested was looking into utilizing the velocity from the curves I sourced the volumes from to push out the burn and temperature attributes from a more dense center. Now that definitely makes sense on paper, but implementing it was another matter entirely.

I struggled with attribute wrangles and using my limited knowledge of coding for an entire night of work just trying to figure out how to call attributes and let the curves affect the burn and temperature values. I started out trying to specifically multiply the velocity by each attribute, but after many failed attempts, I ended up trying to multiply them by the curveu values generated by my curves instead. This allowed me to remap the burn and temperature to be strongest at the beginning of each curve and weakest at the end, which I don’t think is quite what Nelson was getting at, but an improvement nonetheless.

The final change I made to my fire for this week was playing with the pyro shader some more, partially out of necessity since it didn’t quite work right with my adjusted pyro sim, but also because I wanted to attempt making the colors more accurate to my reference. Seeing the final render, I’m pretty sure I overshot the saturation, but I suppose it’s better to do too much first. It’s still very much a work in progress, but it is slowly but surely becoming more functional as the weeks go by and I get more comfortable with working with volumes in Houdini.