Foxhole

The team continues to work hard.

Joel has finished his model of our creature, Frostridge. Cewon immediately took the model and began texturing it in the most brilliant colors. Joel has also started to rig the creature after discussing movement details with Paul.

Sai has finished some great motion capture tests. He’s had a chance to clean up a lot of the material and has discovered it might be too big a job for the data we collected in Organic Motion. But, the data we collected from the main campus system, Vicon, is doing really well in the pipeline. We will discuss further and maybe recapture some of our older data on Vicon.

Chen has finished his first pass at compositing which has given us a better sense of all that is needed in every shot. Our next task now is to look to collecting assets to replace his placeholders in the scenes.

Dorothy has all but finished Ramesh. His eyes were finished this week, his body is rigged, and as of Friday she was very close to finishing the rig for the eyes.

Julia has been hard at work doing more cloth sims. The tests are turning out great and we will be talking about just how to integrate them next week.

Paul was away at the Boston Science Fiction Film Festival most of the week and made some great contacts: Three curators of film festivals who all gave their personal contact information so that Paul can bypass sending films through withoutabox and send work directly to them. The team will definitely be sending them our animation.

Compositing is gearing up.

Over the last week the artists have been sending Chen different assets to begin rough compositing. The team discovered discrepancies in perspective and general feel of the test shot, but we’ve started to compensate by beginning the discussion of perspective and tone at an earlier stage, keeping Chen (our compositor) in the loop at every stage so he knows what’s going into the shot.

We’re also just about done with the modeling. Dorothy ran into some problems with the 2D art not translating into a 3D model skeletal structure, so the model of our lead character, Ramesh, had to be adjusted to accomodate. Dorothy also worked hard on getting a basic animation rig for the Bird Boss to Sai so he could move to the next step in his process of animating the moment when the Bird Bosses join as one.

Julia and Cewon have been smoothing out their workflow but discussing at length the items Cewon needs modeled in a shot to give her a template to draw from. They are both working very hard to accomplish our two test shots, but as they are the most complicated composites in the short, is makes sense that they are finding bumps along the way.

   Joel has finished a model of the creature, Frost, along with an extra model of the head that will portray the animal with its mouth open and teeth bared. Next week he will start with rigging Frost for animation. Joel, Paul and the rest of the team spent time on Friday going over every shot of the film in which Frost and the other creatures appear and discussed just what movement is needed from the creature.

Sai continues to clean up the motion capture data, transferring it from Maya to Soft Image, but for now, he has retargeted the data to Ramesh’s skeleton. We had originally planned for the face to be hand animated, but after experiencing the sophistication of the Main Campus motion capture system, we have decided to capture facial expressions as well. The problem is that we’ll need a face that is more “rounded” like that of Ramesh and therefore our other motion capture actor, Himanshu, will not do because of his slim build. So the director, Paul, is stepping in to act as the face model, his face being a little more “fleshed out”.

 

We’ve spent a lot of time this week trying to refine our pipeline and experiment with technologies.

The challenge Chen, our compositor and head of pipeline, had was mainly how to setup a proper work flow with render farm. We had some hiccups early in the week with getting others on the team access to the render farm and teaching them how to use it, but there were also questions the head of Computer Support needed to answer for himself before our setup was complete. Chen also spent some time figuring out how to organize 3D animatic assets and find a better way to lay out those assets within each shot in the film..

For our lead artist, Cewon, it’s been tough trying to estimate the workflow of this type of art because she’s never been through an animation pipeline like this before. During the BVW round 5, she made a hand-drawing of 3840×1080 pixels in two days, so she assumed a 1920×1080 rough background would take one day. But she soon realized that the quality of an animation background needs to be better than her work in BVW because she’s defining so many buildings in Ramesh’s city. She’s beginning to get accustomed to the type of workload needed for animation though and she’s well on her way to completing the background for the first test shot.

Sai, Chen and Paul finished the first portion of motion capture with the Organic Motion system here at the ETC. But, they found that Organic Motion does not do well when the subject has their limbs close to their body, and in one scene of the short, the lead character, Ramesh, is sitting and crawling on the ground. So, they visited the motion capture lab on the main campus and set up a time when they could capture the remaining scene with their more accurate system. Sai, Chen and Paul also discovered that the system on main campus can be used for facial capture. They will be making an appointment sometime soon to begin that process.

There are a few things on hold as we smooth out the modeling pipeline. As the team discovered when Sai started to run through the motion capture data, the way Ramesh was modeled does not make him conducive to animation–the way his shoulders moved made his arms collide with his body. This was not the fault of our talented modeler, Dorothy. She actually modeled the character perfectly from the model sheets, but we found that the 2D sheets didn’t translate well into the 3D realm. Dorothy is now working on fixes but unfortunately, Cewon’s texturing of the model is on hold until all that work is done.

Joel has continued to show his value by giving us a great model of the “creature.” But we also came across some hiccups when trying to translate the 2D concepts to the 3D realm. On a flat surface, the creature looks positively plausible in it’s shape and suggested motion, but in 3D is seemed a little off, so we’re now taking the model beyond the original concepts and tinkering with the way it moves and carries its weight.

Julia created a 3d mockup of the scene Cewon is developing to help get the perspective of the shot correct. After three different people delivered three different perspectives on a single shot, the team found a need to find reference material and speak often to make sure everyone was seeing the shot in the same way. They have since laid out the shots Julia will be doing and they are meeting daily to look over the perspective and scale of everything in the shot.

Paul continues to bounce from desk to desk, working to keep the aesthetic of the piece cohesive while also keeping up with all the housekeeping of the project. As the sound designer, he delivered a locked soundtrack earlier in the week and will continue to revise it throughout the semester. But, the elements are now timed and the animators can therefore get started on moving the characters to the sounds on the animatic.

The team ended the week with a little break.

Having started the semester in full working mode, the team has had little time to get to know one another, so Paul ordered pizza and had everyone take an hour out of their schedule on Friday to have a little pie and play a rousing round of Pictionary.

 We’ve met a lot of challenges this week.

One of the biggest pieces of our puzzle that must fit is the motion capture pipeline. Chen, Sai, Julia, and Paul went down to the Organic Motion room at the ETC early in the week and learned how it works. Sai then went down later and did a test. Looking at the output, one can easily say that it gives a good estimation of the movements, but when Sai imported a model from Maya to Motion Builder the resulting recording was not that good. On the other hand, skeleton output was very nice. So he tried to figure out the basic problems behind it. The main problems were …

1. Not having a proper IK/FK controls for the rigged model and characterization.

2. Losing data of rigged models which are modeled in higher versions of Autodesk products when exported to lower versions of it.

3. Skeleton faces in the opposite direction when you create it in Motion Builder for recording.

In general, the procedure for motion capture is to export a model to Motion Builder and then record actions. Having a lower version of Motion Builder in the MoCap room, Sai was facing problems with movements of the model. So he thought about reverse engineering; why not just take the skeletal data and retarget it to a model in Maya? Sai gave it a shot and it worked perfectly with a little bit of data correction.

We also did tests on the rendering and art pipelines. On Monday, Chen will speak to the team about what he needs from the creative team as far as using our render farm is concerned because we found that we were keeping him out of the loop during our tests. We also learned that all artists need to understand the perspective of the shot, so we built a pipeline around shots with distinguished perspectives which involves Julia making a rough model of the shot in Maya then sending the file to Cewon to do an art pass.

We tested different shading techniques on our Ramesh model and found a problem with the eyes. The concept art looked great on the page, but the eyes didn’t translate well to our model. So, we have had several meetings and sought the opinion of both our advisors. We are now working on a fix–Cewon will be doing a detailed drawing of the type of eyes the team decided upon, then Dorothy will remodel the eyes to that specification.

Joel Ogden is a great asset to our team and has delivered a rough of the Creature. We are planning on meeting with him next week to discuss the model.

We also had a final editing pass done on the animatic, fixing some timing issues. It is now ready for a rough soundtrack to be locked for use by the animators.

The whole team is pleased to be working so hard in the first few weeks, but we did discover one big problem: We don’t know each other all that well. While most teams are taking these first weeks and learning about each other, we’ve been at the grindstone. So, part of our mission within the next two weeks is to take our team dinner and have a little mingling time.