Wednesday, October 29, 2008

using reference wisely

I notice the preponderance of rotoscoped animation at the Academy and felt obligated to put up some thoughts on how I approach using reference. This isn't by any means a formula - formulas invariably fail in animation.

So let's start with an exercise, as a way of breaking down process. In an earlier post I mentioned the merit of learning for learning sake by doing more of an exercise rather than gunning for a demo reel piece at every turn. Why? Animation is about evolving one's process, and constantly learning how to see deeper. But pushing ourselves deeper into simple exercises is a good way to solve technical problems in animation as well as push your eye into seeing more detail.

In the spirit of doing a concrete exercise, I chose a throw. But what kind of throw is interesting? A baseball pitch is always good, but I've done enough of those, so I found 2 clips to work from from BBCmotiongallery - a shotput animation (clip 1 clip 2)

So the first thing I did was to watch the clips repeatedly. For me, a great way into seeing is by sketching, so I did a rough sketch sheet of what I took to be the key poses and important breakdowns in the scene.

Admittedly, these sketches are pretty rough. But right now they're a way of searching out information and seeing how poses sequentially work together. In analyzing the motion so far, I broke it down into a few ideas.

  • anticipation..athlete getting into the internal headspace to throw the ball.
  • raises arm and bounces a few times..more of an exaggerated anticipation
  • bouncing up and down in preparation for the throw. This seemed interesting to me. In one of the videos, the thrower just bent down, stored up some energy, and exploded outwards. The bouncing seemed like an interested move in that it's almost practicing what's to come, and getting in tune with his body.
  • weight shift onto back leg as pelvis raises
  • front leg kicks up and raises as he hops to build momentum
  • plants on front leg, explodes out into the throw, torquing his body in the air, landing on the other foot
  • land on other foot, balance on it, settle back to watch the result of the throw. More of a contented leisurely pace
So now I've done my research, chosen elements from 2 clips. Now it's time to put our poses into maya. This next part is the most time consuming part of the process..laying in the key poses. It should be the most time consuming part because if the poses are weak, the animation will be lackluster. For this exercise, I'm keeping it in stepped mode as long as possible. I usually don't work this way for motion pieces, but lately I've been pushing myself to work this way because it's a tried and true method of really focusing purely on posing and timing. So here's the first round...just poses and rough timing. I used the reference footage as a guide and approximated how long things took, but as I'm using a different model and don't want to be too slavish to the reference footage, I pushed it aside after I put in the rough posing and did a rough timing pass based on what felt good.

In looking at this version...there are plenty of problems, but there's enough information to get a general idea across. But some of the poses are weak and they could flow better. So my next pass is to refine the existing poses and add in new ones as needed. I'm still being as economical as possible so as not to overload myself with too much information.

Ok. This is better. The animation is flowing better and the arcs are make a little more sense. Right now areas to watch are the flailing up and down with the arm in the beginning as well as the recovery. From experience, I know these are areas that once splined will be problematic and either require toning aspects down or adding more time. In order to better see what's going on with those areas, I'm going to add a LOT more breakdown frames. Probably about double.

This exposes some problems, but parts are working. There is some fluidity to the throwing part, but the scrunching down feels pretty wobbly. The spine is rotating all over the place and fighting itself on the way down. This will be jittery if I spline it right now, so some of this information I've put down is wrong. Time to stop looking at the reference footage for a while as it's served it's purpose so far. On the next pass, I'm going to refine the jitter, reclaim the key poses so there's time to experience them without noisy motion. I'll spend most of my time eliminating wrong poses and strengthening the ones that are in the right place.

Because the beginning part was getting worse, i deleted the wrong keys and reestablished some holds, simplifying the motion a bit. Also, I'm noticing that some of the breakdown frames are too close together, so on the next pass, it's going to be crucial to focus on timing and spacing, as they've been suffering a bit through looking at the motion. I'm also making a conscious choice to minimize the bounce at the beginning, both in the extremity of the motin as well as the amount of felt like it was overshadowing the next moment and to get better texture to this scene, I'm toning it down, again, taking the IDEA of bouncing in place from the reference and implementing it differently. I get up out of my seat here and act out this motion with the goal of figuring out the lean of my chest, and where my cente of gravity is relative to my feet. I've gone back and added in many more breakdowns on the faster action, as it's time to be decisive.

Next, I spline out my curves (plateau in this case, as it avoids overshoot, but you have to watch for flat tangents where you don't want them). Ok. A little disappointing, but time to carve in. The timing is mush in places, snappy in others, so let's pad keys where needed, add more fluidity and overlap to various poses, take out obvious hitches in the spine rotations and pelvis, cleanup that problematic torso hitching in the beginning, and start to fix the rear settle.

And here we are, at first pass stage. Here, I want the timing and spacing and poses as complete as possible. The only thing left after this (which I didn't get to for this demo) is refining. Refining means many things, but let's focus on splining out to first pass. I slowed down and reworked the beginning - it's not there yet and I'll next want to choose a camera angle and refine to that (as it's an exercise, I haven't focussed too much on it. were it a real shot, we would have locked down the camera from the start). I've heavily revised the weight, especially as he does the weight transfer as he throws. There's still plenty of problem areas, but it's closer now and now we can focus on weight mechanics and getting more fluidity into this piece, which will come by pose tweaking, adding new poses, and refining what we have to make it flow better.

This is NOT finished by any means, but I hope it demonstrated a process for you. Really, this is about the appropriate amount of steps to get to the polish state. Go to the splinedoctors site, as they have some great posts about taking your work past blocking. Andrew Gordon had a great quote to the effect of really spending time on the posing and timing takes a lot more work on the front end, but makes finishing the shot much faster.

So where are we at this point. At 50%. That's how much time it's going to take to finish this shot, but that's a lecture on polish, which is what separates the good from the great.

Not that this is good or great by any means, but I had a little more time today to start the polish pass, setting a specific camera and cleaning with that window open in the corner the whole time to see how my changes affect that angle. Let's call this nearly second pass. At this point, the changes get slower - things I might think about at this point are things like really pushing the moving holds..for the throw and hop, when he lands on his front foot, how is his balance? How long does he hold his foot up before placing it is the spacing on it as he plants it. Looking at the animation, it's still pretty rough, but moving along in a positive direction.

A trick I sometimes use in figuring out stepped to spline is to grab keys in the timeline and turn them back and forth from stepped to spline mode. Being lazy, I made some shelf buttons to do this faster. The mel for these is below. Basically, whatever you grab and is red in the timeslider will change tangency.

These 2 mels are like a toggle between curve types for selected curves in your timeslider

evalEcho "timeSliderSetTangent step";
timeSliderSetTangent step;

evalEcho "timeSliderSetTangent plateau";
timeSliderSetTangent plateau;

These next to set your global keying preferences, so the next key you set will be in a particular tangent type (you can also add in linear, etc, but for this demo, i just go back and forth between these two)

keyTangent -global -itt flat;
keyTangent -global -ott step;

keyTangent -global -itt plateau;
keyTangent -global -ott plateau;


clockwerkz said...

Hey thanks for taking the time to put together this post on your workflow. Very helpful and much appreciated! There's an economy of movement that every athlete pretty much does after thousands of times of doing the same moves over and over.. so I find that using reference of a professional moving around for poses and timing, but maybe either acting it out or filming myself doing the movement will help me highlight where my weight is all the time, mainly because I won't come even close to being as smooth as an athlete. So I'll be clunky doing the moves, but my weight will stand out.
I can never get enough of watching professionals' workflow methods, so thanks again! Love the blog!

jeff said...

glad it's of help. I'll try to keep adding to this post, as it's a useful topic - both for me as a teacher and student...thanks for looking

Steve Masuda said...

Thanks for posting this Jeff, it was one thing before your class that was never taught, and I'm glad someone was able to go over video reference more thoroughly then just "Take video reference!"

Jason Fittipaldi said...

Great post title, or Greatest post title ever? =D

Awesome guide Jeff - very much appreciate it!

jeff said...

thanks Jason.
I hope it's of help..the search for a perfect workflow never really ends.

@b said...

fantastic post jeff...a very nice way to breakdown your you said the search for the perfect workflow never ends, but at least by posts like these we get to hear how different people are approaching their scenes, try different workflows and find out what works best for us... thanks!

jeff said...

glad to see you guys are reading this and it's of use.

I tend to always be looking for a better workflow and find that different animations tend towards a differing workflow. For game animation, I tend to do complete posing, but keep things in spline from the beginning, so I can address the weight as soon as possible.

Pradeep Singh Bal said...

Really really Great POST Jeff ... This is what i never came across : using references for animation, amazing POST.Thanks.

Anonymous said...

Who is the owner of this Blog? Finally i came to good advice for this. But i need to ask more questions.

Please send me email or something, so i can write you with couple of questions about using reference for animation