This week instead of presenting a breakdown of our progress to the mentors, we presented the first full composite of our project so far.
Based on the feedback we received, my main goals for this week were to work on the reflections on the car and to make the particles the main light source of the scene. If there was time, I also wanted to start working on the lights from the car headlights.
The comparison on the right of our render compared to the video reference we shot and the feedback we received made it clear we needed to refer back to the reference more. However, the reference in the comparison was from slightly later than in the clean plate we are using, so I went back through out footage and found the reference we shot from the same time of day (below) to work to match.
We received comments about the angle and positioning of the car compared to the reference as well, so Bua worked to update the car animation for shot 1 and 2 before passing it back to me to continue iterating the lighting.
Before starting on the reflections, another issue brought up when comparing to the reference was the light positioning. The side of the car closest to the camera should be much darker than we had it, so I double checked the HDRI positioning and realized a mistake I made. I had positioned the HDRI as if it were shot from the same place as our main camera, when we actually shot it from the middle of the road where the car passed through. Readjusting the HDRI to match the POV of the camera it was shot with started to darken the side of the car more. I also adjusted the exposure of the HDRI to better match the reference.
Original HDRI positioning Adjusted HDRI positioning HDRI Exposure: -1
Another issue I addressed, which I realized after the last render, was I neglected to adjust the ray depth, so I increased the number of diffuse and specular bounces.
Ray depth: 1 Ray depth: 2 Ray depth: 3
The reflections of the HDRI onto the car weren't quite as clear as they were in the reference, so I went back into the car shader and increased the coat IOR as well as the IOR of the windshield. By this point this is how the comparison to the reference looked.
The HDRI reflections were looking closer to the reference, but they looked distorted and stretched from the way the image was projected on the geodome. To fix this I added in environment cards over the car with an image of the forest canopy from one of Joleen's HDRI shoots that would hopefully provide better reflections. The forest photo was more saturated as it was taken at a different time of day, and it colored the car blue before I adjusted it to match the HDRI. The reflections from the cards were less distorted, but I had a lot of trouble positioning them so they weren't interfering too much with the HDRI lighting, but also covered the car while it was moving so there weren't obvious differences between the HDRI and card reflections.
Environment card: original image Environment card: color corrected image
Prof. Gaynor's also advised to curve the environment card planes to prevent distortion, which did end up helping some of the positioning issues as well. The positioning will need to continue to be adjusted, as there are still visible gaps, but there is much less distortion with the cards, which can be seen more clearly in shot 2.
HDRI reflections Environment card reflections
As previously mentioned, one of the lighting issues we got feedback about was how light the side of the car close to camera was. Although repositioning the HDRI did begin to address this issue, I also added a render layer of the ground shadow reflecting back onto the body of the car for Bua to use as a mask to darken it further. This did darken the side down closer to the reference, but it doesn't reach as far up the side of the car as it should which I will adjust for the next render iteration.
Ground shadow reflect layer Car render without GSR Car render with GSR
For my second goal this week, making the particles the main light source, I focused on testing how the glowing particles would effect the car and the road. We had to reconsider our Houdini to Maya pipeline with .ass files to implement the Houdini shader. This method involved me importing particles as points with said shaders into Maya, but Prof. Fowler advised we use spheres rather than points. This would allow us to experiment with shaders and mesh lights to find the best method for lighting.
Due to the size of the file with the particles as spheres and not needing the shader, Joleen went back to exporting alembic (.abc) files for the next particle iteration. She had also began to separate the particles in front of the car from the ones behind it, so for the first test I only had the front section of particles.
For these tests I applied an aiStandardSurface with a high emission value to the shaders to start to create the glow effect on the environment. For the glow on the road, I initially applied a mirror material to the ground plane to capture the light reflecting, but the reflections were too sharp for a glow effect and the positioning felt off. Realizing from this test a road would be much more diffuse of a surface, for the second test (which included the front and back particles) I applied a lambert with a gray value close to that of the road.
Particle reflection on mirror (front only) Particle reflection on lambert (front and back)
For the second test of the glow on the road I also rendered a version of the beauty layer with the reflection of the particles with emission to test how it would come across and judge the best way to get the glow effect.
Comments