Week 05
- Feb 11
- 5 min read
Critique
try smaller light sources on the ring to catch the bevels
think about using light linking
shot 1 (pedestal/ring roll)
ease out camera
make sure background is tracked well
shot 3 (flower)
ring is too long in the air
make a more graceful fall out of the petal
shot 2 (water)
work more on lookDev of water surface
add caustics + more depth to the water
shot 4 (fabric)
change transition to match cut instead
the pop-up feels weird, 10% longer? more focus on it
shot 5 (break-apart)
pop up ring faster
end motion graphics needs more work
A lot of the critique this week seemed to focus a lot on specific animation and timing notes. We decided that Cae Parkhill should take over shot 1 (which I was previously working on), and it's definitely a better workflow– it gives me a lot more time to focus on comp work! I had previous experience with compositing for CG, however I wasn't super confident and wanted to learn a lot more.
This series that I found was really great, as it not only shows how an AOV rebuild is done, but it also explains to you key concepts of how nuke calculates and operates, as well as the science behind different passes. It's definitely long, but I really appreciated how in depth it was, and I'm sure everyone would learn from it– even if a lot of concepts were refreshers.
I asked my team this week to have two deadlines: The first being a still exr with all the AOVs to be submitted a few days early, then the second to be the full sequence submitted at the normal deadline for comp. This way I can build my node trees and start testing early to make sure I don't get too crunched, and they don't have to scramble to submit early!
Material AOV rebuild of shot 3 (flower):
I wanted to try my hand at an ACES pipeline, as I've used it a few times before on film student productions. After lots of research I created this guide and tested the full pipeline. I felt pretty proud and confidant that I had it properly figured out, which should've been the first alarm bell when it comes to color workflows ;)
Since this guide wasn't complete when the stills were due earlier in the week, most of my testing of those were done in linear, so it was a bit of a surprise when I received the sequences, only to learn that my pipeline did not, in fact, work!
It was definitely a hair tugging moment, however since there wasn't enough time to re-render any shots, the most I could do was to color correct to make it look decent enough and take the loss. I didn't have a great understanding of what the correct color should look like, so I was mostly taking educated guesses.
Example of color with shot 3:
render output
cc
accurate color
It also didn't help that for every shot the whites were extremely over exposed. I made sure to clamp my values, however there was only so much I could do with super blown values. I wonder how this challenge is tackled in an industry environment, and if there are skills and tricks I need to develop next time!
I started taking notes of any issues I noticed in the render, like blown whites, clipping geometry, missing frames, and general oddities (like how the leaves in shot 3 had emission on them), which I'll share with my team next week!
As for the testing stage, we needed to add more depth to the water in shot 2, so I had some testing to do, as my team was curious if I could make light rays in comp.
Below are some videos detailing different methods and use-cases of god/light rays. I particularly liked the one by Compositing Academy, as it showed a CG scene and helped me out a lot!
Some references I collected of god rays underwater.
Working with the still exr, I quickly built up a proof of concept for godrays in nuke.
There's a few issues with the result, being the wrong angle of the light rays, and the general color. I found them to look a bit too strong, and overall the movement and look of them didn't feel like it was underwater. After talking it over with the team it seemed best to pursue light rays as a separate render pass that I comp in.
Shot 2 also had the addition of a green light. Max had to render out the light emitters as mesh lights rather than straight emissive, which brought it's own complications when it came to animating the light to flash. I told him to not fear, as I could very likely do this in comp!
This was a super fun challenge, as I did some funky stuff regarding merge operations and channel shuffling to isolate the green glow, which was present in the volume, transmission, and visible lights AOVs.
I then made linked grade nodes, and in the multiply, I figured out the mathematical expression to make it pulse in a heartbeat pattern! I called in a friend who's a math major in a different school, and we figured it out in Desmos graphing calculator. I then had chatGPT help me translate it to the proper format for a nuke expression.
The result on a still frame is pretty awesome, and it was really fun to mess around with expressions, as I usually don't get the opportunity to!
Unfortunately the result in the moving render isn't very noticeable, which means it's probably a better option to hand key the pulsing, so that it's timed to be at the best moments. I was a bit bummed, but found I honestly didn't mind that much, as it was a really good experiment that I learned a lot from!
All in all, a lot of tinkering in nuke! While most of my tinkering either wasn't really included, or didn't work out, I feel like I made a lot of progress in my learning, so regardless it was worth it! I'm expecting there to be a lot of changes in the coming week, so stay tuned!