Thursday, June 21st, 2018:
Today, I learned about re-targeting, assembly, and polishing from multiple animators. The animation supervisor over one of our projects is Eric, and he wants me to work and see the full flow, and then provide me with a shot on said project to polish up.
So the day began with shadowing Mike. In his first shot, he was showing me retargeting the way it is done here. So he focuses on the contact points - are the hands on hips, is someone laying their palm on another's shoulder, is a leg resting over another? Just areas where we may want to play with FK vs. IK. With MotionBuilder though, with the character IK Blend T and IK Blend R, you have to highlight them in order to turn them on and see the keys that you make for these blends. This is important, because otherwise if you do a switch from FK to IK, back to FK, you won't see your keys unless these are highlighted and selected.
Sometimes when you do this (the blend), you'll need to select those keys (the one at the beginning of the transition and the ones at the end), and interpolate - flat. This keeps them from glitching and spinning on their own axis around, and instead stay put.
Retargeting also though is a step that you don't need to babysit. It's just an initial - find all the bugs and the jitters and knock them out before handing it off to blocked animation, then polish.
When doing a blend between a pose that is moving into a resting position (like the examples explained above), Mike said he typically likes to give an area of 8-12 frames for a transitional blend. When working with Brandon though, he said about 5, so it interesting to see the different points of view on timing (dependent upon the animator). But I also know that they are both working on different projects right now that call for different aesthetics.
He says that, "sometimes you have to sacrifice things to get the essence of the pose - knowing what you can focus on now, and then fix the pose later). Then when is all is done with the re-targeting, plot to control rig.
Then prep for layout. So take each character (in the scene we are working on there are two characters we are focusing on), and plot it to their skeleton. No go into the character tab of the navigator, and delete the mocap skeleton since they are no longer necessary (sometimes called 'divas' from the olden days). Then go into the node editor and space, right click the top node of the mocap skeleton to delete it there as well. This way we are sure that there is no extra info within the scene when we hand it off!
With Chad, who is part of the development team, there was a group meeting with all the animators who are working on the this specific gaming project. What he showed all of us was maya assembly. This is a quick way in using their Shotgun software to run a script that will bring in all of the characters, mocap data, cameras, audio, video reference, body animation, and facial animation into one scene, or 'assemble' it.
Later on in the day, I was able to see this actually carried out and implemented when 6 scenes were needed to have the script run for final gathering.
MIKE & ERIC
After the assembly meeting with Chad and the other animators, Eric called Mike over to do a QA (quality assurance - or in this case, analysis) of shots that he had turned in for re-targeting that now needed approval before being passed off.
In a QA, one of the first things that Eric looks for is hands. He notices that in most of the data, the curls aren't relaxed, and he prefers finger grouping rather than spread out. It is more aesthetically pleasing. So with this in mind for design choices, they made a pose library for hand poses at the beginning of the project that they can pull from to fix data corrections.
When he finds an area that looks somewhat jittery (barley noticeable to the eye), he runs a smooth filter to get rid of that unnecessary motion.
Another suggestion that he makes to Mike in the shots is to always have the translate turned off unless you use it towards the end of the shot, otherwise it is going to fight to be in its own world space throughout the shot.
Avoid things looking floaty. With this, Eric goes in an strategically deletes chunks of data, but does so with care, so as to avoid floatiness, but also have 'moving holds'.
Finally, when shadowing Justin I got to focus on 2 things. Maya assembly and polishing. With the shot that he was working on initially, we dealt with polishing. But he was focusing on a portion of the character out of the camera view, but knew he needed to work there because that would make what was happening inside the camera view feel more authentic.
He also had to work with a prop in this scene. This is the first time that I was able to see prop work done in the scene we were working on, so I tried to really take note of this.
With a prop, you need to set a zero key before beginning so that if you ever need to return to the base pose you can (like most animation). In the shot we were working on, we had a tool being used so we needed to:
- Constrain the TOOL --> MARKER (element) --> WRIST
So for this, we created a marker element and placed it in the palm of the hand where we thought the tool would go. Then we parented said marker to the wrist, and in the base animation layer, we set a (zero) key. Then, we aligned the tool to the marker (both translate and rotate) and moved it into proper positioning.
Next, we duplicated the marker. Now this is what Justin refers to as the 'slide marker'. The original marker is used for the rotates. And the slide marker is used for the translates (in the settings of it though, make it look different from the original marker so that you can differentiate between the two upon selection.)
Now make the slide marker a child of the original marker, and align the original marker to the tool. Now, create a new constraint (a parent/child), and rename it 'Tool_to_RHand' so that you know what the connection of the constraint is between.
- Parent: slide marker
- Child: Tool
Now snap it (in the constraint settings), and clean up your offsets. Now put a new base key for both the original marker and the slide marker.
After doing that scene, we move on to polish with the Maya assembly scene. Justin got all of his assets in one location for the next scene that he was doing, and then polished from there. This was a combination of work from multiple animators that he was working with, so from here he only polished what he thought the camera was going to see. Once he polished the animation (moved some mouth controls, adjusted some jitters, etc.), he baked the key simulation and exported his shot.
I noticed that throughout his entire workflow, he works in order and doesn't jump around. It's clean, assess, then clean your assessments. So doing one character at a time allows for better concentration for him.
With all of this shadowing, and intake, Eric said that they will start me on polishing a scene probably next Monday. At this point I have shadowed Brandon, Mike, Eric, and Justin to get animation techniques down. So as I wait for the scene, I will continue to absorb their style of animation, especially for games and for the aesthetic choices here at House of Moves.
DAILY HOURS: 9:00 a.m. - 6:00 p.m. (9 hours)
CONSECUTIVE HOURS: (82 hours)