Project 1: Compositing Light and Shadow
March 23, 2019
Over spring break, I started modeling my object for this project, which is an Elmer's glue stick. Overall, the model resembles the reference although the bottom part needs more work and I will probably go back to it later on to make it more accurate.
April 3, 2019
These are the reference photos for the background and the background plate for the project. The was overcast when I took the photos and that caused the light to change a little in from photo to photo. To be able to match the final result to the sphere reference photo I will have to color correct it.
April 4, 2019
The next day in class I merged the different exposures of the chrome sphere photos into the HDR used for fill light. I also matched the camera and key light using the photo with the cube for perspective reference. The results are in the photos below.
April 5, 2019
After Wednesday's class and over the weekend I went back to my model and redid the bottom part and adjusted the shader to match better the material. I also worked on the shadow GOBO, which was difficult as the dark rocks on the ground make it more difficult to separate the light and shadow areas into black and white.
April 8, 2019
Over the weekend, I started to try to integrate the object into the background. However, to do so effectively I had to first match the background clean plate to the background in the sphere photo for accuracy. Color correcting the clean plate was difficult for the same reason as making the GOBO. The dark rocks in the background make it difficult to correct based on shadows, midtones and highlights. The solution that proved successful was using the same GOBO I made for the shadow projection as a mask for my color corrections. This idea was suggested by Sarah Van Alstyne.
April 9, 2019
Object final version composite
Plate color correction node graph
Sphere final version composite
Between last update and this one, I focused on the compositing side of the project more. I made color correction changes to different AOV layers to match the sphere reference. I also spent some time making sure the projected shadow was actually affecting the specular component in the object, which proved to need more color corrections than other AOVS to actually create an effect on the specularity. Although I'm happy with the final results I do believe there is still some things that can be improved and would like to go back to it later, like spending more time on the subsurface element of the shader.
Compositing node graph