Atlast its done!!!
I finished this movie 3 hours after my submisison deadline. My final render in after effects took 7+ hours when I had estimated a 3.5 hours for that. But on the bright side, the film is done. It was a lot of work and I enjoyed doing this.
I know for a fact that I wont be going back and making changes to this. Whats done is done. I cannot work on this film anymore. But it never hurts to make a detailed analysis of my film.
First of all, the lighting is to strictly tell the story only. Its bad in many places. The character looks different in different scenes as the lighitng in each scene is slightly different. In some scenes his skin color looks almost like a house lizard 🙂 You can also see some GI dark spots on many scenes. The compositing also has its share of problems. Some parts of the character have not been matted out properly in a few scenes.
Lets talk in detail about the different stages of the film and what I would have corrected had I had more time.
1) Story
The story has 1 major issue. This was pointed out to me by my professor. He battles with the switch. But we see no resolution to that problem. Otherwise the story seems to be fine. One point that might be noted is that nothing drastic happened that the janitor had to come in with a fork lift to fix the problem. It seems to exaggerated and too soon for such a drastic move. I had more problems scripted in. The difficulties of these problems kept growing through the film steadily and justified his using the fork lift in the end. But in the interest of time I had to cut out some scenes. So, the fork lift incident happens too soon int film.
2) Modeling
the models used here are very basic. Infact I built a cube, sphere, cone and box with rounded edges and resized and scaled them for every other object in the film. That way they all looked like they belonged to the same family. As my primary focus was not modeling I tried my best to keep the models very basic.
3) Texturing
You cannot put simpler textures than these. Most of them are plain lamberts and blinns. I used wood more predominently in my setup as this office room was based on the previous office that I used to work in.The texturing can be improved a lot, but then agian this was not my primary focus.
4) Animaiton
This was my primary focus. I wanted to explore character animation. I wanted to animate such that even without the lights, textures and sounds, the viewer must be able to see what the janitor was thinking and gong through. To do this, I animated my entire film with just the basic lambert texture on all objects except the character. I did not add lights or sound and finished my animation to make sure that the actions read the way I wanted to. There still are many scenes where I would love to get back and improve the animation. On the whole I am satisfied with what I could do in under two months with other classes and projects going along.
I started out my animations by blocking the actions very roughly to get the timing of the scene laid out. So using the pose-to-pose technique I could layout my actions and time them the way I wanted it. I then continued to block the scene with a straight ahead technique. My blocking stage was pretty intense. I spent a very long time blocking. In some cases I had to block alternate frames to get the right timing. I splined these animations in my second pass. This was fairly simple in most cases and did not require much tweaking as the blocked keys took care of the the arcs, holds, timings 90% of the time. I unfortunately did not have the time to get to stage 3- refinement of the animations.
Lighting:
I had no exprience in lighting and I knew this film had to be lit properly to make any sense. I used indirect lighting (GI and FG) to get the room lit the way I wanted. Then, I added the spot lights, point lights, area lights where I needed the lights to switch ON and OFF. As the bulb flickers, the spot light intensity had to be keyed. This inturn would affect the glow of the bulb, the ambient color and the color of the bulb. So all these were driven with the keying of the intensity of the spot light through set driven keys.
I also used a 3 point light rig for the character. I used directional lights to lit up the character. In some cases I needed to add more lights or add negative lights to remove some lighting, depending on the scene. There are some lighting issues even now in the film. In the first scene that he enters there is too much shadow on him. He is dark. Also the lightning on him is not consistent between scenes.
I had fun lighting the scene where the ceiling collapses. As different part of the ceiling falls down, the lights in those areas needed to dim. The cracks on the ceiling was done using boolean. I modeled the cracks and performed a boolean operation with the ceiling. Now as I move this crack down and as the two objects intersect, the crack starts to develop. The breaking of the ceiling could have been done in a more dramatic fashion as it is the major climax of the film. Some dust, and more chunks of ceiling falling down could do the trick.
Rendering and Compositing:
OMG!!!
I never thought rendering would give me these many issues. Hands becoming transparent, A lights internsity going to zero as tan object moves between the light and camera, objects looking different in different lighting situations, referenced files and its referenced render layers not rendering the way I want. The number of issues I faced with rendering is too big to start listing here. But with just 7 days for submission time, and rendering each scene in atleast 6-7 layers and then compositing them was a PITA (Pain In The Ass). I was driven to extreme mood swings during this phase. I made so many errors and was clueless to why somethings were not working. Many hours of debugging, reading through failed log reports, fighting for nodes on the renderfarm, dealing with missing frames after the render were just a few of my problems. But towards the end, I finally understood why somethings were not working. I could start following the errors. I could debug the problems soon and come with a quick fix. The quick fix in some cases looked dirty, but when the clock is ticking, you don’t really care. I realized at this phase in the project that while manual error and a lack of understanding of how systems worked were the reason for 90% of my failed renders, the remaining 10% cannot is not in your hands. Just pray that you have enough good karma to take you through this 10%. I mean, when the licenses for Mental ray and After effects expire for around 150 systms in a single day, there is not much you can do except catch up on some lost sleep.
I was so glad when all my scenes were rendered and I still had some 6 hours left for my deadline. My music composer, Jesse Lozano, form the Eastman School of Music, had done a fantastic job with the music and I knew it already synced to my animation. Each rendered scene was in its own composition in After Effects and all there was left to do was drag all these compositions in the timeline, add some transitions, add the soundtrack and hit the render button. I hit the render button when there was still 3.5 hours left for submission. I relaxed for the first time in 2 months knowing I had finally finished this film. But apparently my accumulation of good karma ran out that instant. The title of my film was mocking at me. UNLUCKY. TO render out a 4:30 film in After Effects, in draft quality, at a quarter resolution took me 7+ hours. I missed the deadline for submission and turned in an unfinished version of the film. My render at full resolution finished a day later, which I have put up on vimeo.
http://www.vimeo.com/mcvijay/unlucky
I would love to hear some feedback from you guys about my film on the whole and especially character animation.
Animation and Creativity 