Advanced SMR II : Rig architecture


For years we have refrained from giv­ing turn­around rig­ging pre­scrip­tions, because there are many IFs, and we thought that such advice has the poten­tial to be misinterpreted.

We were hop­ing that all the com­pre­hen­sive tool descrip­tions, plus the avail­abil­i­ty of a full turn­around sam­ple file would pro­vide all the nec­es­sary ref­er­ence mate­r­i­al for users to take the next step by them­selves. Unfor­tu­nate­ly, it seems, that this has­n’t happened.

The rea­sons, we think, are most­ly root­ed in the com­mon lack of sol­id under­stand­ing of nesting.
Many times we have seen Ani­mate users (not nec­es­sar­i­ly with EDAP Tools installed) to put all facial fea­tures on the char­ac­ter time­line. Such prac­tice is a very good indi­ca­tor of expe­ri­ence and com­pre­hen­sion levels.

So, it turns out, part of our mis­sion as tools devel­op­ers, who want to improve the lives of ani­ma­tors and the qual­i­ty of ani­ma­tion, is to try to reach those lost souls and some­how con­vince them that ‘Nest­ing is the way to go!’, and Ele­ment Welder is not meant to weld noses onto facial ovals.
Only after that, such users will be able to move fur­ther and start con­tem­plat­ing the engi­neer­ing chal­lenges of turnarounds.

This is Part 2 of our series on advanced Smart Mag­net Rig­ging. If you haven’t read Part 1, it may be good to start the jour­ney there.

SMR architecture for multi-angle rigs

When it comes to mul­ti-angle rig­ging there are gen­er­al­ly two pos­si­ble approach­es to orga­niz­ing assets (or a com­bi­na­tion of the two), and the choice is often dic­tat­ed by per­son­al pref­er­ence or bet­ter suit­abil­i­ty for a par­tic­u­lar ani­ma­tion style.
Each of these approach­es might have pros and cons, while at the same time both pro­vide good flexibility.

The purpose of the rig

It is very impor­tant to not for­get that the rig is only a start­ing point. 

Var­i­ous new states of the body parts have to be added to con­vey the best pos­si­ble pos­es, in accor­dance with the estab­lished prin­ci­ples of ani­ma­tion; the goal should always be to main­tain the styl­is­tic con­sis­ten­cy of the char­ac­ter and the project.

The rig should not deter­mine what the char­ac­ter can do. This lim­i­ta­tion should come from the inter­nal log­ic of the design and the rig should be extend­ed and amend­ed as we ani­mate, to allow full expression.

The rig does not have some spe­cial intrin­sic val­ue. The mean­ing of its exis­tence is to make the process of ani­ma­tion faster and easier.

Organizing assets

Just like in part 1 of this arti­cle we described the tech­ni­cal meth­ods to con­trol and main­tain meta­da­ta, let’s now out­line the var­i­ous approach­es to orga­niz­ing assets.

Con­tain­er of sta­t­ic variants
This is usu­al­ly illus­trat­ed with the exam­ple of a mouth sym­bol or a hand sym­bol, and is the most eas­i­ly under­stood one.
All hand ges­tures can be put inside one Graph­ic Sym­bol, each hand image occu­py­ing one frame. Then on the out­side, the instance is stopped (Sin­gle Frame) and by choos­ing a dif­fer­ent First Frame num­ber we dis­play the desired hand ges­ture at char­ac­ter level.

Con­tain­er of unsynced animations
A con­tain­er can hold not only sta­t­ic images, but also short ani­ma­tions, where one posi­tion tran­si­tions into another.
In this case, por­tions of its time­line will be played out of sync, using the Play Once option and defin­ing the desired range with keyframes and First Frame.

Con­tain­er of synced animation
A synced con­tain­er is a Graph­ic Sym­bol set to Play Once. It plays in sync with both the char­ac­ter and the Main timelines.
The most com­mon exam­ple here is the head, but it can be any body part. If a stroke or a cus­tom brush is used for a limb, it is pos­si­ble that in some shots, these legs and arms sym­bols will be played in full sync, while inter­nal­ly they will be Shape Tweened.

Swap­ping synced containers
A good way to illus­trate this approach is the typ­i­cal head, which has to play in sync from 1 till the end of the shot. In this case we have sev­er­al head angles, each a sep­a­rate Graph­ic Sym­bol, and swap between them. Frame num­bers of inter­nal and exter­nal time­lines always match to main­tain sync with the dia­log and act­ing performance.

All our videos are 1080p. If for some rea­son full HD is not avail­able here, try watch­ing them on youtube.

Swap­ping sta­t­ic containers
For larg­er projects, which have a true pre-pro­duc­tion stage, swap­ping can be effec­tive­ly com­bined with con­tain­ers of sta­t­ic vari­ants as well.
For exam­ple, there can be mul­ti­ple full turn­arounds of a foot, which show it form above and from below. These can be orga­nized each one in its own con­tain­er of sta­t­ic images. If the sym­bols are put in a ‘foot’ Library fold­er and named in a way which alphanu­mer­i­cal­ly orders them the way we want, we can scrub through the con­tent of the cur­rent one with Next / Pre­vi­ous Frame in Sym­bol, and go up and down the grid with Pre­vi­ous or Next Sym­bol In Library to swap.

Video – Mul­ti-sym­bol Arrays

Objectives for functional multi-angle rigging

The goal of func­tion­al mul­ti-angle rig­ging is to orga­nize assets in such a way, so that they can be accessed with the small­est num­ber of clicks or key press­es. They should also pro­vide room to eas­i­ly add more vari­a­tions as need­ed, with­out slow­ing down the ani­ma­tion process, and at the same time keep­ing all these new assets as eas­i­ly acces­si­ble as the orig­i­nal ones.

In the con­text of SMR, best prac­tice is to have only one uni­ver­sal rig per char­ac­ter, where all body parts of one of the views are inter­change­able with the cor­re­spond­ing body parts of the oth­er views.
This means that meta­da­ta has to be syn­chro­nized inter­nal­ly and externally. 

The Sym­bol Instances rep­re­sent­ing the same body part across all views of the turn­around must have the same instance ID, i.e. if the left upper arm in the front view has an ID of 7, the left upper arm in 3/4 view must also have an ID of 7.
This syn­chro­niza­tion is achieved very eas­i­ly by Rolling Over Rig Info (RORI) from one rigged view to another.

Mag­net Tar­gets in these match­ing rig ele­ments must also car­ry the nec­es­sary metadata.
If we look again at the arm exam­ple, its elbow MT will need to be copy/pasted from the front view sym­bol into the 3/4, pro­file and back view sym­bols, if such sym­bols exist.

Necessary decisions

Due to the dif­fer­ent nature of the var­i­ous body parts, they usu­al­ly need to be orga­nized and nest­ed in dif­fer­ent ways. In the Orga­niz­ing assets sec­tion above we out­lined the avail­able options.

Some of the choic­es that have to be made are very obvi­ous. If we will ani­mate to dia­log, the head has to run as a synced Sym­bol Instance, there­fore we will need mul­ti­ple sym­bols for the var­i­ous angles. Putting them in a Library fold­er will allow for quick and pre­dictable swapping.

Hands and feet
In most cas­es the hand or foot sym­bols will be set to Sin­gle Frame and frames will be cho­sen from a range of sta­t­ic vari­ants. Some­times there may be short ani­ma­tions in there as well, which will be Played Once as needed.

If we work on a larg­er pro­duc­tion, an array of hands (and feet) may be orga­nized where there are mul­ti­ple sym­bols with full turn­arounds, such as point­ing, fists and so on. As explained ear­li­er, these can be swapped, and then indi­vid­ual frames cho­sen from the cur­rent container.

The tor­so can be approached in two ways:

1. We can have only one tor­so con­tain­er which has the turn­around views as frames inside, and we can also have some extra views and short ani­ma­tions in there. These short ani­ma­tions can be tran­si­tions between the main angles, or some stretch­ing and squash­ing, bend­ing and so on.
Mag­net Tar­gets will need to be adjust­ed to match the artwork.

2. We can sep­a­rate the views, putting them each in their own con­tain­er and swap them. This allows for clean­er access to short tor­so ani­ma­tions such as bend­ing, twist­ing, shoul­der shrug­ging, etc.

The choice between 1 and 2 is deter­mined by the char­ac­ter design and antic­i­pat­ed range of movements.

Often times for spe­cif­ic parts of the ani­ma­tion, it may be prac­ti­cal to dupli­cate the tor­so sym­bol and work with this dupli­cate. An exam­ple would be an ani­mal run­ning. The tor­so in this case will have a very spe­cif­ic, loop­ing squash and stretch, and it is best to keep that sep­a­rate and synced with the run/walk cycle.

Arms and legs
Arms and legs will most cer­tain­ly need fore­short­ened vari­a­tions. How to nest them will be deter­mined by the design.

Master container

One last thing that we need to men­tion is that we need to have a mas­ter con­tain­er – a sym­bol which con­tains all oth­er symbols.
It does­n’t mat­ter how they are arranged and dis­played there.
The pur­pose of this is to be able to copy/paste or drop all the ele­ments of our rig into the doc­u­ments where we will ani­mate them, so that the ones which are not ini­tial­ly on any time­line are avail­able for us in the Library when we need them for swapping.

This is also the sym­bol (Library Item) which will con­tain the char­ac­ter’s col­or palette meta­da­ta for Sym­bol Palette Con­trol, if we need to have mul­ti­ple palettes.


Now when all objec­tives are set and all meth­ods are well under­stood, we can look at a prac­ti­cal demonstration.

Here the head is nest­ed one lev­el deep­er to allow even faster block­ing out of ani­ma­tion. After the ini­tial pass the wrap­per is bro­ken apart and this expos­es the dif­fer­ent head con­tain­ers, which will run in sync and will have the facial ani­ma­tion inside. This con­cept is known as Break Apart Work­flow.

Video playlist – Advanced SMR techniques

Beyond rigging

A good, well-struc­tured rig can make the work of the ani­ma­tor eas­i­er and more enjoy­able. This can indi­rect­ly lead to bet­ter animation.

Even though in this arti­cle we have tried to describe the process in its entire­ty, and define the var­i­ous chal­lenges asso­ci­at­ed with rig­ging, this is nei­ther a com­plete study, nor it is sup­posed to be fol­lowed step by step. Rig­ging is a cre­ative process and all sug­ges­tions here are only rough guide­lines, show­ing how you can com­bine the exist­ing options and methods.
Real deci­sions will be made on a char­ac­ter-by-char­ac­ter basis and will be deter­mined by the pro­jec­t’s design style.

We have men­tioned this else­where, but it is worth repeating: 

When you start ani­mat­ing, get into the habit of cre­at­ing keyframes across all lay­ers of your char­ac­ter and think of these as pos­es. Nev­er only move a limb in iso­la­tion. This will make your ani­ma­tion look much more organic.

Use SMR to help you, but not lead you. Nev­er for­get that this is not a rigid rig, where the ele­ments are nailed to their par­ents. You can move them away from the MT and every­thing will still work. Only when you need them to snap back, they will.

A good rig is a nec­es­sary pre­req­ui­site to mak­ing high-qual­i­ty cutout ani­ma­tion, but does not guar­an­tee it.
The time spent on rig­ging will only be jus­ti­fied if the ani­ma­tion pro­duced with the rig is worth it.

The biggest chal­lenges in cutout ani­ma­tion are stiff­ness of pos­es, floaty tim­ing and a sense of dis­joint­ed­ness of the fig­ures. All these can be over­come eas­i­ly, when the ani­ma­tor learns to always think of the col­lec­tion of body parts as one organ­ic whole. Expe­ri­ence with hand-drawn frame-by-frame ani­ma­tion can help immense­ly in this regard, but is, unfor­tu­nate­ly, get­ting more rare and more dif­fi­cult to obtain in a pro­fes­sion­al set­ting. Young peo­ple have to invest huge amounts of time on their own to prac­tice tra­di­tion­al ani­ma­tion and life draw­ing, which will then fuel their growth and abil­i­ties to pro­duce bet­ter cutout ani­ma­tion faster.

Nick­o­lay Tilch­eff
April 2022


No comments yet.

Leave a Reply