Scene 1 shot 4 is mostly done. This time I had to figure out how to climb a steep staircase.

Allthough I knew I had to plan the animation properly I again got distracted by the mechanics of a stepping cycle. To put the real animation on top of the stepping cycle was some additional work I could have avoided. I will see how I fare in the next shots.

Nonetheless the result isn’t too bad.

With now 3 shots animated (okay scene 1 shot02 has to be redone) I did a rough test on editing and cutting these 3 shots for the opening sequence.

What I had seen during animation of the individual shots, was that each shot was in effect longer than the planned sequence in the storyreel animatic. If the ratio was to hold true for each shot, the final movie would suddenly be at ~7min runtime with all the effects on total render time and such.

Now with the first editing, the time is down again. not exactly as the initial planning but close. Lesson learned here is, that the context gives the timing and pacing. A single shot is a context in itself and seems to demand more runtime.

And again, to avoid unnecessary animation a good planning and rough blocking could be of some help here.

See here now for the first edited 3 shot sequence

Contrary to my idea of reusing the animation from shot 02, I started doing the walking animation from scratch. Its a good thing I did this, as I have learned a lot during this shot.

I studied the advice given in Richard Williams’ book ‘The Animator Survival Kit’ ( which I should have right from the start and used this to get Ara walking up the bridge towards the stairs.

Until now I wasn’t quite happy with the footroll mechanism I had built for Ara, but as it turned out, I just used it in the wrong way. Reading carefully through the walkcycle basics I started to use the footroll control in a different way and to my astonishment it worked quite well.

What I have learned too, is that more planning is required before starting animating a shot. This time the planning is for the actual acting. I have to know know what will happen in this shot to lay it out correctly. This is certainly an area where I will have enough opportunity to practice.

I was not quite sure how the workflow is for layering actions over a basic animation. I tried to use the NLA editor for this but in the end went back to do it all in one action as it gave me the best control and more important overview over the animation.

Another point I discovered was, that the lookat control for the eyes was nice but not really usable in the animation. The control was parented to the head, so that if you set it once, the eyes would always stay fixed in respect to the head. This is not what the eyes are doing in reality. They focus on something out of the reference of the heads coordinate system and stay locked on it. For this to work I had to rework the lookat control somewhat. Ii am still not completely satisfied but for this shot it is sufficient.

And, I couldn’t resist to test my theoretical cloth workflow on this shot. Better test it early than be sorry after all the shots are animated

See here for the current version of scene 1 shot 3:

And, btw, I will certainly redo scene 1 shot 2 from scratch with all the new knowledge I have now regarding walkcycles for Ara.

I am happy to say, that the production phase for Ara’s Tale has started. That means that I am now actually starting to animate the individual shots.

It was quite some work and major headaches to get the template scene files and models ready for action.

First I checked again my setup from my previous research for having a simple rough cloth ( without the simulation) for animation preview and when the animation is finished to actually turn on the cloth simulation. It turned out that I had to modify my setup a bit.

The dress is linked in from the model file but I have local copies for the rough dress and the cloth simulation. The rough dress is driven by Aras rig and the cloth simulation is only parented to the main bone of the rig and the rest is done by the cloth simulation and collision.

Next I ran into a bug when using proxy armatures for rigs with action constraints. After saving the scene file with the linked armature all action entries from the action constraint fields were gone. I learned that if I protect the bone layers in the model file this prevents this problem to occur. Alas, it also prevents to add custom constraints in the scene file.

What I had totally ignored was to setup a camera rig. I soon figured out, that this was something to have and came up with a first solution, where I have a camera with bones for main position, camera position, camera point at and lens value. In combination with a  follow path constraint I think it will pretty much cover all the shots I have to do.

Speaking of shots, there are a total of 42 in it. It seems like there lies an universal truth in this answer …

I have chosen as first shot to animate the second one in the movie, where the camera pulls back from the closeup of the cage and we see Ara carrying the cage while walking a path. I have read somewhere, that you shouldn’t start animating with the opening shot, but I thought I give it a try nonetheless.

What I learned so far, is that animating is hard and tedious work. And I have yet to find a workflow which suits me and gives good results in an acceptable time frame.

What I also plan to do, is try to reuse certain parts of the animation in other shots. Shot 3 from scene 1 e.g. is just Ara walking with the cage along the path seen from front. I hope to reuse the walking animation I have now for shot02 with just added head movements and some tweaking.

You can see here the first production animation done so far for Ara’s Tale. This is version 1 and for sure not final as there are quite some rough edges to it. I may leave it though in a not totally perfect way and go on to other shots and come back later.

I am glad to announce that Mikkel Nielsen has now joined the project and will be doing the sound design on Ara’s Tale.

Mikkel approached me a couple of months ago and I gladly accepted his help, as this is an area where I have no experience.

And now, with the production phase imminent, we had sort of a kickoff mail exchange with all my ideas so far on certain aspects. These will act as basis for Mikkel to actually start designing all the countless sounds which are needed to enhance the experience of this project.

And what started as an one man show suddenly has grown into a team effort.

Welcome on board Mikkel.

Now that I am happily married, I start to gain momentum and motivation in returning to and working on Ara’s Tale.

I totally underestimated the challenge of rigging the face. Its very demanding, both technically and artistically.

Based on my research my first naive approach was to set up some bones for the areas of the mouth, brow and cheeks and play with the weight painting. That way led to frustration. Coincidently the Durian team ( more precisely Nathan Vegdahl ) posted on their blog, that he abandoned a bone based rig and returned to a shape key based one.
Following this, there was also a very informative thread on blenderartists about ways to rig a face.

I did a step back and considered the situation. My rig should just do what the storyboard showed me it should do, no more, no less. So no need for a generic rig with a full expression range. I started anew and created several shape keys based on some basic expressions found at the artnatomia site. Only the jaw is bone deformed.

I ended up with 3 shape keys for the brow area, 3 for the mouth area one for the cheeks and one for the lower eye lids. I hooked all the shape keys to control bones in my rig and started playing and tweaking.

I still have some problems with the mouth area, but I think the system I have now will fit my needs.

Initially I thought it necessary to create higher level actions for basic expressions, but after playing around a little bit I think I will use the bare controls for the moment.

See here for a sequence of images where Ara goes through a series of expressions. This is all done with the shape key system combined with the already in place (bone driven) eye rig system. Its only greyscale since GIFs can only display a 8bit palette based images and the color version has some serious color banding.

I may still tweak the shape keys and even may add some more, but the moment is nearing when production is finally starting, meaning the animation phase will start in earnest and I will start with the first shot to animate.

I have decided to continue with the facial rigging as this suits my currently available time schedule better.

I started with rigging the eyes. Major sources of information and inspiration were the Mancandy rig as well as the facial rig tutorial by Paul Cageggi. I also got a lot of information from Gordon Goodwin’s rigging repository.

The rigging process itself was relatively straight forward and the used techniques quite clear. The really time consuming part was the weight painting and additionally reworking the mesh.

After I had set up the rig the deformation was horrible. Even with decent weight painting I had grotesque deformations on the eye lids. I was nearly at the point of abandoning this rig setup when I remembered ,that I had weight painted the whole head (including the eye area) for the head bone. The weight was set to 1 for this bone. This was the source of the horrible deformations I was witnessing. If the surrounding mesh is all nearly pinned with a weight of 1, you have not much control left.

So I cleared the head vertex group and started weight painting again. As I should have guessed this wasn’t a great idea either, as now the slightest weight – and be it only 0.001 – immediately started moving the vertices without damping. So the solution was to put some weight on the head vertex group which provided some initial pinning which is overcome by the influence the eye lid bones are providing. Once this was done the actual weight painting of the lids was a relatively easy thing to do.

The result was almost pleasing, almost. This ‘almost’ sent me back to reediting the mesh. Having not only a clean topology but also a clean mesh helps the deformation in extreme setups like those of the eye lids. By clean mesh I mean having the edges (without the subsurf modifier) going in a smooth fashion and not zig-zagging around wildly. The subsurf modifier hides these problems and only when starting deforming you realize something’s amiss.

I am quite satisfied with the current setup and think it will also be sufficient for the several extreme closeups of the eyes. At this time there are no corrective shape keys used, its pure bone deformation.

See here for some examples of eye-posing

I have also done a small test animation. See here two versions, one a viewport rendering where you can see rig controls and then the same animation rendered.

I will continue with the face rigging. The next part should be the areas around the eyes and the brow.

Contrary to my last comment about putting the project to rest for some weeks and to my own utter surprise I was able to invest some time on Ara’s Tale.

This time I started the rigging for Ara. Before actually doing a rig I had to do quite some research. I have never done a serious rig before, so I read a lot of tutorials, threads on rigging on blenderartists and cgtalk, watched various video tutorials just to get an idea what the essence of rigging looks like.

I soon realized, that I should not create a general purpose rig, but only implement the things I really need. The shot breakdown done with the storyboards gave me a good idea what to expect.

As always, learning is iterating. A lot of the problems discussed in various threads and the solutions provided in tutorials were just theoretical concepts and hard to categorize for me. So I just set up a simple rig and did some posing and first animation attempts. That gave me quite some insight and the tutorials and discussion threads made much more sense and provided additional insight.

After quite some iterations I am now at a point where I have a basic body rig for Ara which seems to work for all of the animations tasks ahead (face rigging will be done later, after I have done some serious animation work).

The key facts of my rig are as follows:

• Strict separation of deforming, helper and control bones
• The leg setup is IK based, with a foot roll control driving action constraints. I may add a FK chain and an IK/FK switch as I noticed some problems getting the walkcycle correct (shin swinging …)
• arm setup is FK but I will use auto-IK for posing. There are two shots were an IK setup may be helpful, so I also may introduce an IK chain for this.
• The fingers on each hand are driven through separate controls for curling and and splaying. For very fine tuning I have left access to the individual finger control bones.
• The toes are not controlled separately.

I have read a lot about FK/IK switching and its inherent problems for smooth transitions without jumps and as it seems, even more importantly, the workload when you start tweaking the animation. As of now it seems to me that using Auto-IK with FK is a good compromise for the animations tasks ahead of me.

Rig controls

Full rig

The weight painting to tweak the deformations and maybe even adding corrective shape keys is still on my todo list, but that shouldn’t interfere with the actuals animation work.

There were countless pitfalls setting up the rig and I may still come back and built it from scratch, but having a working rig for Ara, I started to create a walkcycle. My very first walkcycle ever done.

I learned two things from it:

• Animation is a hell of a lot of work to get it anything near watchable.
• I really have fun doing it

On a personal and very subjective view I am quite pleased with the cycle produced. On a more objective perspective, I realize that this walkcycle has a lot of problems.

What helps in this project ( at least I think it does) is that Ara is wearing a long dress and thus obscuring various glitches.

Anyway I am really curious to what level I can push my animation skills during the work on this short.

See here for Ara’s first steps.

The next step will be to actually animate one of the opening shots where Ara is walking along a path with the cage in hand. To do this I have to investigate a little bit on a proper animation workflow.

One thing for instance is the use of the NLA editor in combination with cyclic actions. It would be cool to setup a walk using a basic walkcycle and using the NLA modifier system to layout the walk path. To work then on the details it would be nice to ‘bake’ the NLA strip to a new action and work on this action from then on. I will try to find an answer on the forums.

Since my last post I worked mainly on two aspects of Ara’s Tale.

• preparation work for Ara’s rigging
• adapting the roughsets to match the new storyboard and mapping out the shots

The rigging preparation included some rework on Ara’s mesh and (again) some research on how to combine Ara’s animation with the cloth simulation.

I was able to get some very useful information from these threads on blenderartists.org

http://blenderartists.org/forum/showthread.php?t=147902

http://blenderartists.org/forum/showthread.php?t=97413

A good point made there was to see cloth simulation as seperate step during production, which will demand its own portion of time that has to be done carefully. That means that I will concentrate on Ara’s animation and once thats finished (ha!) I will tackle the cloth simulation for her dress. The same I think should be true for her hair.

Rigging will be interesting, to say the least. For mapping out the storyreel in the rough scenes I just set up a extremely simple rig with deforming bones only and using Auto IK for very crude posing. The process of mapping out the shots showed me quite well where emphasis has to be made on the rigging and where I can get away with less effort.

The rigging process will be next, new territory again but just lets see …

I already created an animated storyreel based on the original storyboard. I decided not to do a full animatic this time as it tended to get me distracted and in the end gained nothing from the extra effort put in. Once I have broken down the production to individual shot files I may come back to 3d animatcs to try out camera work.

The 3d storyreel isn’t completely finished yet, but I have put a video of the frst 80 secs on vimeo to see for the interested. The sets are VERY rough, but at least Ara is used with her actual mesh, but very crude posing.

Once animation for the actual shots starts, I will have to enhance the sets, maybe not final but certainly finalizing the terrain Ara moves in. All the parts where Ara and the dragon are interacting with the environment should be finished. I hope this workflow turns out to be feasible.

Production note

Work on Ara’s Tale will be dormant now for the next 6 weeks. After waiting for 25 years I really do not want to jeopardize my real life wedding by spending too many hours intimately close to a girl like Ara

There has been a lot of activity behind the scenes, reworking the storyboard and visual storytelling.

I am working through ‘Directing the Story’ and try to relate the principles shown there to Ara’s Tale. It gives very good hints and invaluable insights.

The main theme is how to draw the audience into the story. This is achieved by a multitude of techniques. One principle which I find quite true is that as soon as you create an emotional bonding you have reached the goal of getting to the audience. For an emotional bond the audience need to understand what its seeing, being guided to ask questions about whats happening next. By delaying the actual answer the audience will try to generate their own answers and thus get drawn into the story.

Upon review of my storyboard so far, I noticed that I presented a detached view. During rework, I tried to show more Ara as a person and the development more from her view. The questions asked there are simpler, easier to understand and thus help the storytelling.

A lot of work has gone into the second half of the story – the part which is not shown here – as it incorporates the climax of the story – the confrontation between Ara and the dragon. Here I tried to display both characters and show the dialogue and personal transformation which happens here.

The book also mentions composition as tool to portray feelings and give visual hints. I tried some variations and I got a glimpse of how powerful careful composition planning can be.

As very simple example consider the following two storyboard drawings. It shows Ara walking through the canyon from above. It should show some suspense and give the impression of the oppressing walls upon Ara.

The left image is the original one, the right shows the same scene but is slightly tilted. This gives a more dramatic and dynamic effect and supports the story. The question currently pending is, where is Ara going, where does this canyon lead.

Some aspects mentioned in the book will be explored during the animatics phase. Especially the pacing/editing is very hard for me now at this stage. With more experience I am sure you can plan this even during storyboarding, but I am not there yet.

Right now with the storyboard converted into a storyreel the estimated length will be 3:40 min (without title, credits etc) and will consist of 48 shots in 2 (maybe 3) sets.

See here for the first 100 sec ( yes its getting more , but this time without sound ) of the actual storyreel. If all goes well, this will be very close to the ‘final’ version. It has to pass internal QA (aka feedback from my family and some chosen critics).

The story is still the same, but I think the mood has changed quite a bit from the first version. And it has changed for the better ( in my opinion).

The actual time tracking sheet clearly reflects the shift of focus to the storytelling aspect:

----------------------------------------------
 8:45    Animatics
42:02    Concept Art
   12:00    Ara Design
   18:35    Color Tests
   11:27    Dragon Design
78:06    Modeling
   35:01    Ara
   43:05    Dragon
21:51    Research
    1:16    2d-art
    1:30    animation
   12:00    effects
    7:05    modeling
4:00    Rigging
   4:00    Dragon
 5:30    Script
43:43    Storyboard
----------------------------------------------
203:57 Total

The next steps will be to rework the rough sets and try to do a simple animatic based on the actual storyboard now with the new models and sets.

Then I have to shift my focus to the rigging and minor reworking of the models.

Thanks to Jack-Jack for bringing this to my attention.

A new version is now uploaded, which should work now correctly. You can download it directly from here.

I apologize for any inconvenience this may have caused, but then, until now nobody has complained .