Week 11

Finished fixing the body:

I'm fine with the body mocap data for now. So I will move on with cleaning the face this week.

Cleaning the eyes:

The eyes were the most problematic parts after retargeting, that's why they were my first priority to clean up, as I'm not certain how long it might take me to fix all those jittering and random movements.

Before Cleanup

Looking at the graph editor itself, I can see a lot of bumps and spikes in places where it should be smooth and flat. This explains all the jittery movements from the retargeted data. 

Fortunately, it's actually really straightforward to fix. I could start by deleting all those unwanted frames and only leaving the keyframes that defined the pose. 

I would say this got 80% of the way done. Of course, every subtle movement has been removed but this was a decent base to start adding them back using animation layer.

Animbot here to save the day(my time):

But before going all the way with this method, I wanted to try out more tools in Animbot (cause I paid for it:)) ) and I came across the tool that was exactly what I was hoping for. The "Ease In/Out" slider allows me to smooth out sections of spiky curves and adjust the smoothing value accordingly. This method was really quick and flexible as I can smooth out curves while comparing to the reference footage at the same time,  this way I don't need to go back and re-animate the subtle movements in a different pass.

It was looking pretty decent after cleaning up with the Ease In/Out slider. 

Update: The drop around frames 121-123 was actually too much, which I didn't really notice while moving the slider frame by frame, not until I did a playback at slower speed. But this is an easy fix.

Cleaning the brows:

After I finished cleaning up the eyes, I moved on to the brows, which weren't as problematic compared to the eyes. 

Before Cleanup

The most problematic part was simply the brows not holding at maximum around frames 194-223. Other than that, just subtle jittering here and there, that was an easy fix using the Ease In/Out slider. 

Plus, in some places like frame 43,  the outer brow should raise slightly quicker then followed by the inner brows.

After Cleanup

Cleaning the mouth:

The brow cleanup was done really quickly, I'm starting to get familiar with the Ease In/Out slider more and more. 

Before Cleanup

Looking at the mouth retargeted data, there are a couple noticeable problems I can identify straight on. The mouth shape around frame 46 could probably be pushed more forward using funnel and squeezed a bit more using purse, I think this would minimize the amount of lower teeth showing to better resemble the reference. Moreover, the jaw should be opened slightly more.

The jaw around frame 85-91 was biting too much after the closing action. Miri just finished saying the word "went" around here, and the mouth should be held at the shape pronouncing the "Tst" sound. Moreover, the upper left lip could move downwards a little bit to better resemble the shape in the reference footage.

The mouth corner around frame 20 was pushed too much upwards. I could try reducing the corner pull Y value or I could also try countering the movement with the mouth corner controller.

Most types of problems with the mouth retarget were listed above, mainly the jaw position and mouth corner position in general. The mouth retarget was nearly decent even before the cleanup so I didn't smooth as many curves compared to cleaning up the eyes and brows, since a lot of the subtle movement was retargeted nicely. This certainly makes the mouth animation feel the most natural in comparison. Thanks again to past me for not rushing the mouth-tracking process, and thanks Retargeter for actually capturing the tracking data decently, unlike the eyes :).

After Cleanup

Feedback:

I showed Dr. Kennedy the body and face performance all together for the first time since week 7. He was really pleased with what I've got. The mouth being the most troublesome for most students was captured really well. However, he did point out a couple places that could be further refined. 

2 - 3 parts of the facing of the eyes could be moved slightly more towards the center. As Miri should be looking right at Fred in those moments.

Right at frame 24, both feet were seemingly floating as the center of gravity was shifting from left to right.

OH GOD, she was holding a lip balm:

The problem listed above was not even as noticeable as the missing lip balm. I honestly didn't notice Miri was holding and fidgeting with a lip balm until Dr. Kennedy pointed that out when giving me feedback. He said Miri was being questioned by Officer Fred during the performance so she should be feeling anxious at the moment. Most of the time when her hands were placed closely together, she would be fidgeting with the lip balm as a visual indicator that she was under pressure/stress.

I honestly couldn't notice the lip balm throughout the performance when viewing the Samsun Coverage footage.

Until I took a closer look at the Canon Coverage footage, I did spot the existence of the lip balm in a couple FRAMES(not seconds!!!).

Around frame 760 was the moment I finally spotted it and started to analyze and retrace the movement from here.

Reflection:

I can’t believe I actually thought I was 99% done with this assignment! But setting the frustration aside for a moment, I’m really happy and proud of how the cleaned-up performance turned out. It feels like there were a lot of steps listed in this week’s blog, but I managed to finish all the face performance cleanup in about 5-6 hours. AnimBot’s Ease In/Out slider was a total game-changer in cutting down the time.

As for adding the lip balm, I’m planning to test out the controller first, figure out how the constraints work on this rig, and see if I’ll need to set up any extra controllers for the workflow I’m more familiar with. After that, I should take a closer look at the lip balm movement. The tricky part is that I can barely see it most of the time, so I’ll probably have to rely on my best guess and reenact the action with a lip balm in my hand.