Project 4: Neighborhood pt. II

March 12, 2013

Final Project - Beginning to End

The second part of the "Neighborhood Project" wrapped up pretty quickly, due to the time constraints of the quarter. I had about three weeks total to complete parts 1 and 2 of this final project, which included modeling 11-12 stylized houses, hand-painting and creating each texture, lighting, layout, animation and finally, as a last-minute decision, compositing. Below are still-render images of each house. 

This project ended up being much more difficult than I anticipated because I had so little time to complete it in, and I was overly ambitious in what I wanted to learn and apply. I began to feel particularly overwhelmed this past weekend because I was using the RenderFarm for the first time, during what happens to be its busiest time. Even though I had intended to render my animation on the RenderFarm about 4 days before the due date, I actually ended up finishing the rendering and compositing the day before, probably due to the fact that I wanted to composite the animation as opposed to rendering the final animation from Maya. I felt as though I had too much on my plate because I had to design and model 12 objects, create each texture from scratch with minimal texturing experience, improve upon basic lighting skills, match lighting to a background plate, and composite the animation, all within a short period of time. The learning curve was simply too steep for what I wanted to achieve with this project. 

That being said, I think my first real attempt to create textures and surfaces, match-light an image, and composite went well. However, if I had more time to complete the assignment or less houses to account for, I would have developed the concept and research more thoroughly before beginning the project and pushed the stylization of the houses more to achieve the wonky, moody look I initially had in mind. The latter also has to do with lighting and the backplate image, which are new areas for me as well.

Despite my struggle with this project and the stress of experiencing the RenderFarm for the first time (during finals week, no less!), I learned more than I ever thought possible in the span of a few weeks. I learned a lot about texturing through trial and error and rookie mistakes that had to be made in order for me to understand the process of creating and applying texture maps. I learned about how extensive the research and concept should be before beginning; while I spent three days drawing my 6 concept sketches and pulling together more than 60 reference images, it became clear that I hadn't planned through to the end of the project, where the background image and environment come into play. I also had a difficult time with the color palette, as that was created on a texture-to-texture basis instead of at the beginning. 

I also realized that I have much to catch up on in terms of lighting, having struggled with lighting and color throughout my time as a Fine Arts major as well. I will definitely pursue more resources on lighting and practice lighting for different scenarios and moods. I decided to composite my animation in Nuke as I had done for the Jeep project so that I could experience more compositing and begin to gain a better understanding of the process and the program. That part seemed to be the most troublesome and stressful, as I had to deal with multiple render layers, rendering on the RenderFarm and accounting for dropped frames, and troubleshooting compositing errors in Nuke that I had very little knowledge about. However, the compositing seemed to be the most informative in the end and I feel confident that I had learned enough to give me a good foundation to continue with this next quarter. 

Finally, there were other minor issues that involve geometry, beveling edges, and a displacement problem, which is noticeable in the images where the edges of the houses meet. Throughout this project I have created a very long list of problems and techniques to explore more in depth!

This shows the displacement issue mentioned above, where
the edges of the geometry meet and the displacement of the
roof texture creates black gaps

Wood texture photograph, applied as a bump map to give the
texture a more realistic, believable feel. The only parts of the textures
that were not hand-painted were the photographs that were used as
bump maps and specularity maps. 

A closeup, photographic texture of a rocky cliff which was used
as a bump map for the cobblestones

This was a frame from the TIFF sequence I rendered out of Nuke - not
sure what happened with it! But luckily, it wasn't missed.