Read from bottom to top to read the process in order
This week we had our final class before submission. I had quite a lot of problems this week, thankfully I was helped by Dr Kennedy. For some reason the footage used to do the face animation was not aligned with the Canon face footage which meant my face and body animation was not aligned. I had to select the face controls and move them back 3 frames in the graph editor. I also had a quick issue with the lights. Because Fred is standing in a different direction to the original T-pose it makes the lighting look different on his face and the inside of his mouth was too bright. Luckily all that needed to be done was rotate the lights so they were facing the other way. I also got lots of help fixing up my face cam. It was not set up properly on the frames before the animation but I didn't notice this and tried to set up the camera during the animated section, it moved alright but Fred's chin dipped out of frame and the camera was making bigger movements. We went back before the animation starts and set up the camera so it was aligned and straightened properly. We also had to zero out some of the rotations and translations of the body. After setting up the camera that way it moved much better and kept Fred's whole face in the shot. I moved it around slightly to get the right angles and then parented it to the head controls. I then made my playblasts and rendered them to an mp4 in Premiere Pro. I'm happy with how my playblasts turned out. I think all the face feedback I was given for my formative made it look better and more accurate. The final problem was that when I sent my MB file to my friend to make sure it looked right on her end, the face cam was not parented even though it was on my end. This also happened when I sent my file to Dr Kennedy for feedback. I think there may have been an issue with the reference file. We just duplicated the face cam so it will be easy to be parented if it is not parented during grading.
I have finally finished everything and it is all ready to submit. I'm happy with how everything turned out. I didn't have any huge major problems during this process just little things and things like the alignment of footage which just had to be worked around. I think I worked well during this whole process. It was very helpful having the suggested weekly checkpoints, I just made sure to follow those checkpoints to stay up to date and not fall behind. I only fell slightly behind during week 10 because I spent a lot of time fixing up my blog and adding more detail so I finished the body clean-up slightly later but I caught up and was able to have everything done on time. I think if I was to do this assignment again it would be interesting to do a scene that has a little bit more facial movements. The movement in my scene is subtle but I don't think that makes it much easier to pose and animate because you have to look at all the tiny details and the small changes that aren't very noticeable. So I think it would be interesting to see how I work with a scene with more movement and bigger expressions.
I did quite a bit this week. I finished the clean-up of the arms and imported the performance footage into the scene. I did the props and hand animation for my scene too. In the scene Fred is holding a notepad and pencil in separate hands and then puts the pencil in the same hand as the notepad and shifts the hand around the notepad and pencil. I honestly enjoyed doing the hand animation, it was a bit of a challenge since the performance footage doesn't have a very clear shot of the hands, but I was satisfied to see the outcome, I think it brought more life to the animation to have the hands posed. For the hand swap of the pencil, I duplicated the pencil and positioned them where the pencil would be in each hand. I hid the pencil in the character's left hand until the pencil is swapped and then hid the pencil in the character's right hand so you can only see one pencil the whole time. I made sure the cut between the swap was pretty seamless and it looks like the hand is moving the pencil. The animation of the hand holding the notepad was a bit challenging because all the fingers shift around the notepad, I had to time each movement correctly so it matches up to the footage. I also had to resize the notepad a bit because it seemed a bit too small at first and I couldn't position the fingers around it to match the footage. There are a few points where the hands cut into each other slightly but I made sure it looked normal from the camera angle I was going to use for my submission. I plan to use an angle that is more side on and is similar to the angle of the Canon footage.
I added the facial animation to the animated body as well. I first implemented the feedback I was given for the face. I noticed that the change in focal length of my formative submission made the right side of the mouth droop down which was written in my feedback but in my file with the original focal length it was fine. I implemented the feedback using the focal length of my submission so I could clearly see the changes being made and make sure it looks right for my summative submission. At first I made a new animation layer to add the changes and then I merged the two animation layers and it somehow got rid of all my changes so remade the file and added the changes onto the original animation layer and lifted up the right side of the mouth using the graph editor. I had to lift it quite high but it seemed to match the footage more accurately. I also had to adjust the mouth on frames 108, 123 and 132. The mouth appeared to be too open in those frames so I edited them to be more closed. I also adjusted the eyes since in my feedback it was noted that the upper eyelids were too closed and made Fred appear sleepy, so I also lifted the eyelids to open the eyes a bit more.
I think at this stage I am almost ready to set up my scene for submission. I have asked for feedback on the changes I have made and the body animation and will prepare my scene next week.
I got given my feedback for my formative submission this week. I'm a bit disappointed with myself for the feedback I got for my blog. I understand the feedback and what I did wrong. I think I should have put more effort in my blog for my submission, I was so focused on working in Maya and not so much in my blog. I spent this week working on my blog and writing in my detail about my process and my thoughts for each week of the process. I am now happy with my blog and the amount of detail I added to it. I'm happy with the feedback I got for my work on the eyes brows and mouth. I did get some feedback on things to adjust but it doesn't seem to be anything too drastic and are things that I could change without too much trouble.
I also did the body cleanup this week. There was quite a lot of jitter but it wasn't too hard to clean up it was just tedious. I didn't make any big changes in the graph editor. Most of the curves weren't very big so it the animation didn't look very different after the cleanup just more smooth.
This week I went into the graph editor for the rig and for the controls and deleted all the points that were not in my assigned section so the character only moves for the same amount of time as the face. I also had to adjust the knees since they were broken. I created an animation layer and moved the knees controls so that they were in a position that made sense for the rest of the body. I was lucky that I only had to make one key frame for the knees and they were fixed fot the rest of the section because Fred doesn't move around at all. With the feet I made sure they were flat to the ground and created a plane to be the floor. I was also lucky with the feet because they are flat on the ground for the whole scene I was able to zero out the controls so they don't move or float around at all.
For this part of the process I really followed Dr Kennedy's tutorials step by step because there was a lot of different file making and certain things that had to be done the exact same as the tutorial. This made the process straightforward and I didn't have any difficulties. The cortex was done for us so we didn't to do any data clean up before working in Motion Builder. For previous Motion Capture work I used Autodesk characters so it was interesting learning about the rig and skeleton of a Meta Human in the tutorial. I first imported the Proxy file that I downloaded from Canvas and saved it as a new file. I then imported the cleaned cortex trc file into the Motion Builder scene. Then I added an actor to the scene and placed it over the motion data. I had to adjust the actor a bit to make the data points match up with the actor's body. I had to widen the hips a bit and adjust the shoulders. I had to rotate both the shoulders to 30 degrees and then rotate the arms back down to the T-pose because otherwise the arms would be rotating around the shoulders rather than the shoulders moving and controlling the arms. After that I created a marker set to match the data to the actor. Normally I would have selected the markers from the list in the marker set section which was tedious so it was good to learn that you can physically do a lasso select of the markers and drag it onto the right section of the marker set. The actor was then able to do all the movements. Once I made sure the actor followed the movements accurately I then connected it to the proxy rig so I could use it on the Meta Human in Maya. Then I baked it at 23.97fps, this turned out to be the wrong frame rate so I had to go back in and re-bake it at 24fps.
In Maya I made a new scene and imported the full rig file. I then imported the final Motion Builder file I made. In the human IK section I connected the Meta Human character to the custom rig. I could tell it worked because the shoulders lifted up. The knee controllers did not follow quite well so I will have to fix those up and make them point the right way. After the rig and character were connected, I had to adjust the IK blends for each control to 1.00. Then I was able to bake them together.
Over the break I did the clean-up in the graph editor. It took quite a while because I used a lot of controls. I was able to zero out some sections where the control is not supposed to be moving too much but has slight jitter like some of the tweakers and the mouth suck. I also went through and adjusted some controls that I initially didn't use very much like the teeth, the lips shift and the cheek raise. Sometimes it was really hard to see if the clean up was actually changing anything because some of the controls had such small movements but quite a lot of movement. In the graph editor some of the curves were quite flat and it was hard to tell if it was jitter or just really small but necessary movements. This was mainly with the mouth as there are so many different controls and quite a few do very similar things.
I also had to really change the position of some points in the graph editor and make it almost opposite to the original position but I think it made the result turn out better. I also added some more movement where the points don't move if I notice there is more movement in the footage. Overall, I think the clean up turned out well but it's a little bit hard to tell when the shoulders and rest of the body in the footage is moving or on an angle and in maya it's very still.
I retargeted the data this week. I had to redo it a few times because some of the poses didn't look quite right after being retargeted. It didn't take too long to retarget. I had to go in and adjust some poses. The edges of the upper lips on the character are thinner than in the footage so I had to go through and try and make the edges thicker. I used the tweaker controls for this like corner sharpness and roll. Even using these controls it was a bit hard to make it match. I also used the mouth sticky controls to make the edges thicker. There is also a bit of jitter throughout the scene so I will have to do quite a bit of clean up.
Near the start there is a moment when he speaks and his mouth is closed and then the next frame it's fully open and then the frame after it's closed again. It kind of makes the mouth look glitchy but it matches up with the footage because in the video the mouth looks the same way but it obviously doesn't look glitchy. I don't to try to adjust it too much because I want it to stay as true to the footage as possible.
This week I posed the mouth. Fred speaks a bit in this scene but he doesn't do any wide mouth movements. There are 13 poses of the different mouth shapes. I originally had more but realised that some of the poses I had were very similar to each other so I was able to cut some out. I moved quite a few controls to position the mouth to try and get the right poses . I think I matched up the poses to the footage pretty well. Throughout this scene Fred's mouth is quite pursed mainly when he is speaking. It was a bit hard to make the characters mouth look pursed enough and had to quite a few controls to make it look right. I used all of the Oh controls, funnel, purse and towards. I think what made it a bit of a challenge is that when the characters mouth is pursed it doesn't create wrinkles on the lips like it does in the footage so I really had to just look at the overall shape of the mouth to make sure it matched.
This week I did the retargeting for the brows and the eyes. With the brows I had to retarget a few times because at one point during the blink the brows didn't raise high enough compared to the initial posing. For some reason the poses did not look quite the same after retargeting so that's why I had to repose a few times. The first few times I retargeted I retargeted with the pose value but when I did it with the scene value it seemed to match the poses I made much better.
The eyes retargeted pretty well. Although after looking at the eyes in this video and really paying attention I had to go back in and move the eyeballs around because there were times when the eyes looked like they were pointing in different directions. Also after receiving feedback (also after this video was made) I could see that during one of the blinks, the eyes squint slightly mid blink as if the eyeball is rolling. I wasn't able to fix this by moving the eyeball controllers so I created a new pose while he's blinking where I make the squint more prominent. There is also some jitter on the upper eyelids which I will cleanup.
This week I finished working in Analyzer and paramaterised all my tracking data so I could start working in maya. I was working from home for this part of the assignment so I had to download all the plugins and Facewear Retargeter. It seemed pretty straight forward following Dr Kennedy's lecture tutorials. I didn't have trouble getting the plugins.
In my assigned section Fred doesn't move his eyebrows very much so I created four key poses but there isn't a lot of movement between each pose, the brows raise very very slightly a couple of times throughout the scene. Mainly when he blinks and at the start after the neutral pose. There is a moment when he stops speaking and his brows don't move at all in that section. I had to really look through slowly at the brows to see all the slight movements. I had to be strategic with how I used the controllers because sometimes the controllers wouldn't really create the same pose just by raising a control or lowering a control. I had to use multiple controls to create one movement. Since this was my first time working with these controls it didn't really make sense to me why the controls weren't creating the same kind of movement as the footage. As naïve as that was I came to understand more about the controls and how you had to be strategic when making poses and how you definitely can't create a good pose without using lots of different controls that you wouldn't expect to use. It really made me look at the footage and the poses in much more detail and look at all the tiny differences.
The eyes move more than the brows so I did use quite a few poses for them. The don't move around to high degrees but slight movement when the eyeballs move a bit like looking down or looking in a different direction. There is also a blink and a flutter blink in this scene. I used pretty much every control for these poses except for nose wrinkle, pupil scale, parallel look and eyelashes. Throughout the scene Fred is squinting slightly, mainly after the blinks. It was a bit of a challenge to make the character look like he was squinting and not just tired. For this I had to be strategic as well and put more thought into the controls I used. The squint control did not do enough for the squint I was trying to replicate so I used controls like blink and lid press sometimes even though his eyes are open for majority of the scene. The tracking of the eyeballs was pretty good so I did not have to make many major changes to those. There were just instances when his eye was not looking in one direction enough or looking too much in one direction. Because the iris on the character is smaller than in the footage I had to really pay attention as to where it was sitting because it was a bit harder to match up. I had to look at the whole position of the eye when moving the eyeball around slightly.
I did the mouth tracking this week. The outer mouth markers tracked well. I had to adjust some where there were inconsistencies with placements. Mainly with points that had a bigger black dot it meant there was more space the put the tracker so I had to scrub through a couple times and make sure they were sitting in the exact same spot. After having to do a similar thing with the inner eyes I was able to do this without trouble and more understanding about how consistent the markers should be. The inner mouth markers took more effort. The screen left side had lots of jitter and flew around a bit. My problem was that I didn't place the markers in between the spaces of the outer markers which meant they got a bit lost and inconsistent throughout the scene. Since the inner markers didn't have any black dots to track I had to be more precise and detailed as to where I put them. The other problem was that the nose markers were hidden but for some reason the mouth trackers were still keying on the hidden keys. This threw me off a bit and I was getting confused and redoing the tracking and placements quite a few times. Eventually I figured out the problem and unhid the nose markers and made sure the mouth markers were place right on those keys. After I did that it tracked smoothly.
I had to redo the tracking for my screen left eye this week because I used the point on the eyelid to track while using an eyelash to track on the screen right eye. This meant the screen left eye would be more open. I moved the marker on the screen left eye to an eyelash so it would match the right side
This week I did the tracking the eyebrows, eyes and nose in Analyzer. It seemed pretty straight forward and I was lucky to not have any problems with points flying off the tracking markers. I was worried about how the eye tracking would go because I had to use an eyelash as a tracking marker. There was slight jitter with the inner corner of the screen left eye because there were moments when the marker placement was not very consistent. I went through and looked closer and made sure it was more consistent and sitting in the same spot each time.
Performance capture: the total recording of a performance without cuts using a motion capture system. Performance capture refers to a recording of both face and body that is captured in a single continuous take, as opposed to cobbled together from a series of shorter takes.
This week we were assigned our sections of the performance that we will be working on this semester. I was assigned Fred's performance from frame 2731 to 2970. We had to watch the recording of the actors in the studio acting out the scene as well as the recording of their facial performance and write an action breakdown for our assigned section. I wrote down all the body movements and the upper, middle and lower face regions. I noticed that Fred does not move his upper or middle face region very much in my section until closer to the end so I wrote more about his mouth and the rest of his body. He also only really moves his arms, head and slightly leans at one point. He is very still in my assigned section.