Understanding animation planning and steps

In recent years, Blender got me fascinated with the world of animation. I started by watching tutorials about animating in Blender, then got straight to doing whatever popped to mind using that knowledge. My focus has been on getting better at posing / sounds / compositing / etc, thinking it’s the key to improving the quality of my works. I recently found a Youtube channel about CGI, which also describes how professional animations are made. One thing I seen there made me question my approach, and if there’s something as essential which I haven’t considered equally.

Unlike me, professional animators do something else: Planning and preparing. I was surprised to see that most take time to sketch their scenes on paper first, covering every important moment in the animation. Even after that, they don’t immediately snatch the actual models and start animating them. They first make a mockup animation, with characters made of untextured boxes as well as very low-polygon versions of world objects. Only later they replace those with the character mesh and start working on the final render.

This is something I never did. I just think of a basic idea (eg: I want a city where a robot will walk by, turn around, then go through a door) then load the models for the world, the model for my character, and start posing. I get an idea of how things look as I keep working on the final render, whereas others go the other way around: They know how things have to look at each point before there’s even a 3D scene. In my search for cooler effects and focusing on just poses materials and composite nodes, I never even thought about this until now.

I’m looking for someone to explain this better. Why exactly do professional animators sketch what they’re about to do first, then animate box characters before getting to the final ones? What are the real steps someone should take before creating an animated film? And how should I handle this in Blender? I know it’s a less technical domain, more focused on thinking and common sense rather than blunt functionality… but I’m hoping someone found a way to explain it so I can do things the right way.

Only because, if you fail to plan you’re planning to fail, in real productions you can’t re do something over and over again, it takes too much time and too much money, so we sketch quicky ideas to know if we’re on the right direction, consult with everyone (mostly if there is someone above you like a director), then we move to the next stage, same thing, prep quickly because at this early stages if you decided that your character needs to be a dolphin instead of a man, then you’re still in time to go back, without loosing that much work, it’s all about having the ability to go back and forth between steps, once yo get to a certain point, there are many things that you can’t change without having to scratch everything and starting over.

That’s mainly the reasons.

It’s about storytelling. People who make animations professionally do so because they want to tell a story of some kind. So in the beginning, the question is: what story do I want to tell? This is the stage where the writers and producers and authors and agents are all taking meetings with each other, doing story pitches for things that might not even be written yet, or business executive sit down with advertising salesmen who try to convince them that their agency can tell such a good story the business will improve.

Once they have a story in mind, then they move on to: how best to tell this story? This gets the concept artists and the script writers and the character designers into the picture, as the story moves from written words to pictures and storyboards. The overall timing begins to come into focus as the storyboards and dialog give some indication of how long things will take.

But it’s only an indication. That’s when the animators start moving boxy characters around in a low poly world, with somebody reading the dialog into a cheap microphone so the boxy characters can have a sound track, and the timing is worked out, and scenes that don’t get the story across can be easily changed, or cut, or added to so they ‘work’ to tell the story.

Only when everyone is convinced that the scenes are good, the dialog works, the pace and timing of the animation is right, does the production move from all the planning to actual production. Actual voice actors are hired to do the dialog. Props and backgrounds and sets are modeled and textured and lighted. Characters are modeled and rigged and turned over to the animators, who use low poly versions to save render time while they get everything moving properly and then switch to the high poly versions to render the actual images that will go into the finished animation.

But everything throughout the whole process is driven by the story. At every stage, the question is: does this work to tell the story?

It’s fair to say that for everything I made, I had the course of events in mind before starting. I knew what I want the animation to be about, what theme I’m aiming for, what characters I will use, how long it should take, and so on. But after that, I went straight to working on the final animation with the final models. I never prepared anything in practice, established a simpler test scene, or used any drawn art to envision how the environment should look before it’s modeled in place.

I think there’s another point to this as well: If you base the environment and objects and characters on drawn art, you get better results. For example, if’s probably better to have a beautiful drawn picture of a landscape in front of you and model your landscape to look like that, rather than creating a 3D scene and drawing the terrain as it comes to mind.

Problem is that I’m not sure how much time is lost rather than won… especially if you have to do two sets of animations (one for boxy mockup, other for final render). I imagine it helps if the armature and animation can stay, but you temporarily swap the mesh rigged to it with a set of boxes before starting, then delete the boxes and re-enable the final mesh when ready.

The MultiRes modifier probably helps here. I haven’t used it yet, but heard it lets you store multiple versions of a model for different quality levels. One could have a box interpretation at step 1, a very low-poly version (enough so it’s recognizable) at step 2, and the full quality version at step 3. Would this be a correct approach too?

Can’t comment on multires, but there’s three things in previous comments that need repeating: first, those teams are large, so it’s a way to discuss the content of animation early by using simple models and mockups. If you animate alone, that is not an issue. Second is voices and lip-sync, those guys get to work earlier, if they have something to work on; you get better timing. And number three: sometimes it is easier to throw something away, if it’s poorly done. Animators are not keen on cutting stuff they have worked weeks on.

Moved from “General Forums > Blender and CG Discussions” to “Support > Animation and Rigging”

Short version: it’s easier and less time-consuming (and therefore cheaper) to re-do/revise/destroy things in the planning phase.

Let’s take the example you gave earlier: a robot walks down a city street, turns around, then goes through a door. So many possibilities, so we will focus on two: Caves of Steel and Terminator.

In a scene from the first story, the robot is a detective, going to somewhere with his human partner. As they walk down a crowded passageway (future New York’s equivalent of a street), they pass a mob inside a shoe store, where a customer refuses to be served by a robot clerk. The Luddite mob wants to break the robot clerk into tiny pieces. The robot detective (who appears human, not robotic) after questioning his partner about the situation, turns around and enters the shoe store to resolve the situation single-handedly.

In a scene from the second story, the robot is an assassination machine on a mission to kill a certain person. He is on the hunt in a deserted slum neighborhood of Los Angeles, and detects a clue, perhaps a sound, from a doorway he has passed by on the street. He returns to investigate, entering carefully because it may be a trap.

Now, you have your rigged robot model and a city street model. I think you can imagine the way the robot would move in these two situations would be very different. How would you go about animating one of these shots? Assume for the moment you are just working on the robot, and someone else is going to animate any other characters. What sort of thoughts go through the robot’s mind (there is no dialog in either shot), and how do you get that across to the audience with motion and gesture? How does the robot turn, approach the door, open the door and go through it? At what speed does he move in the different parts of the shot? When does he speed up? When does he slow down? What do you want the audience to feel, and how will you go about getting that response?

You may believe that you can do this planning in your head, and when you jump in to pose your robot character directly it will all work out, but most people would need to plan what they are going to do, and would discover that their initial plan had flaws that needed to be corrected and spots where it could be improved. This shot might go through four or five drafts, each an improvement over the previous one, before it is really satisfactory. Before it really tells the story it is intended to tell.

Now: storyboard sketches take a few seconds to a minute or so to draw. In an hour, an artist can draw, compare, toss and redraw dozens of storyboard sketches to get the essence of the story across. In that hour, the artist has done two or three of the necessary drafts. So when the animator begins blocking out the shot, he is already working on the fourth draft of the shot. And will probably find some things that don’t work in 3D rather than the 2D storyboard, or some things that could be improved still further, and will produce the fifth draft of the shot. This high quality plan is what the finish animator takes and works with to produce the renders for the finished animation.

Had the finished animator just jumped in with no planning, either a lot of time would be wasted starting over to incorporate needed changes, or the finished work would not be as good as it might have been.

Thank you, this info is helpful. One thing to note is that I tend to work alone, so I didn’t analyze this from a team’s point of view. If there are multiple authors, it might be easier for members to progress in parallel with the preparation steps. I assume that animations made by a single person work by the same principle… though in this case one person first does a set of sketches, then the test animation, then the final animation.

I guess there would be one more question: What’s the best way to approach this in Blender? First of all, is it common for people to use drawn sketches to build their 3D scene upon directly? I know that Blender lets you assign an image to the 3D viewport background in any of the side views… should that be used to model the world? Second, what’s the best way to prepare an armature with multiple quality levels? Like I said the MultiRes modifier sounds like a good option, but I’d like to know what others use.

Oh… and since it’s relevant, here’s a video which shows this process in a very good way:

If you are willing to shell out a few bucks, Blender Cookie’s Animation Toolkit is well worth the price of admission, and goes into detail about the process and the methods. Both presenters have extensive experience working solo and working as team members.

My personal process usually begins with stick figures in Blender, objects thrown together out of cubes and cylinders and icospheres, which i move around (just grab and translate or rotate: no rigging) and render in various positions as my storyboards. I use front and side views in background when I model characters, but for scenery and buildings I generally just find some photographs similar to what I had in mind and put them up on a second monitor while I model freehand in Blender.