2022-09-29 Animation & Rigging module meeting

The meeting will be open for everybody interested to join on Google Meet (link below).

Present: Amélie Fondevilla, Beorn Leonard, Brad Clark, Christoph Lendenfeld, Daniel Salazar, Demeter Dzadik, Fani-Eleni Pandi, Jacques Lucke, James Ledger, Jason Schleifer, Jeremy Bot, Luciano Muñoz, Miquel Campos, MohammadHossein Jamshidi, Nate Rupsis, Sybren Stüvel

People present are referred to by first name for brevity. Others are referred to by full name.


Since the Last Meeting / Announcements

  • New face: Miquel, working on Open Source things (“mGear” http://www.mgear-framework.com/) in Maya, has an interest in Blender.
  • Sybren had talk with Ton about his original ideas for “Animation 2020”. Basically Ton’s ideas overlap a lot with the ideas discussed in the module meetings already. There’s also the vision of making it possible to animate in virtual reality.
    • James has done “mocap” in VR
    • Daniel assumes deformation will move from custom modifier to geometry nodes, but how about shapekeys? Delta deformations should be creatable within the geonodes system dynamically, without being so tightly bound to the specific mesh topology. Sybren: yes, shapekeys should be redesigned.
  • Brad Clark, and Sybren had a video call with Raf Anzovin, creator of Ephemeral Rigging. The most important things were:
    • The bias towards cartoony animation is not an inherent property of the ephemeral system. At first he thought it would be, but nowadays he’s convinced that it’ll also be good for realisitc, on-1s animation. It’s mostly still used in demos though, because that’s what Raf is animating himself.
    • As to the question why his system seems to be more accepted / talked about by other animators, whereas other similar systems failed to achieve that, he said it’s mostly due to him and Rick Lico being animators themselves. It’s more a “hey look at what we can do super easily now” rather than software nerds telling animators “you have to work differently now”.
    • The ephemeral system is aiming at character animation. There’s still use cases for “traditional” FCurve/interpolation-based animation, like more mechanical movements. The overlap between those two worlds (f.e. a character from Sea Beast hopping onto a boat that’s rocking & getting their feet adjusted for that automatically; or a character holding onto a keyframed rotating mechanism) is difficult, and it’s something they’re still working on as well. His view on this, if it could be implemented performant enough, would be to see changes to the keyframed/simulated parameters as “modification” in the same way that using ephemeral tools would be seen as such. In other words, after a change to the boat rocking parameters, those frames where the feet interact with the boat would all be recomputed, resulting again in a dense set of “baked” poses. To me (Sybren) that last point also shows that there’s still a need for traditional constraints (or other ways in which Blender is control over transforms & animation data rather than the animator).
    • Raf also mentioned that with good onion skinning & motion trails you don’t need the graph editor any more. This means that you can work much more in the viewport. This also is directly in line with Ton’s vision, where he sees animators working as much as possible in 3D space and not with spreadsheets, buttons, and graphs; a new character animation system should work in VR as well as (or maybe even better than) with keyboard/mouse/monitor.
    • Brad: the thing that makes traditional systems hard is that constraints are “always on”, relations are always there, and then managing that is hard, especially when the low-level graph editor is the only means to control this.
    • Muhammad: mentions Rumba (video) animation & Cascadeur (video) as tools to look into for how to deal with bone rotations & how they’re represented in the 3D view.
    • Chrisoph made onion skinning in Maya, wants to implement it in Blender as well. His approach for Maya was rather hackish, and based on a 2D projection of the mesh – rotating the camera would show distorted shapes. In Blender he wants to do this proper mesh-based.
    • Brad: Raf was working on pushing updates to neighbouring poses. GreasePencil already has some workflows for this we could look into.
  • Brush Assets presentation by Dalai Felinto & Julien Kaspar, last Thursday at Blender HQ. They’re going to use the pose library as proof of concept for asset region in the bottom of the 3D View.
    • Beorn: will Blender also allow other add-ons to add stuff to this region? Jason sais we should push for this. Sybren has no idea, mostly because this “asset region” is still in the paper design phase.


High-prio Bugs

Apart from these, there are 17 normal-prio bugs and 4 low-prio bugs.

Short-term Goals

Character Animation Workshop Preparation

  • Luciano: interaction-wise, how would global “animation mode” be, rather than per-object Object/Pose modes? Can solve issues with armatures not in pose mode, selecting objects without armature, etc. Could also unify visibility of various things, instead of having various options scattered around.
    • Jason: 100% to Luciano’s idea. We hardly ever just animate 1 character. We’re usually animating multiple things.
    • Christoph: +1
    • Jason: can you make your soapbox bigger? I’d like to stand on it with you.
    • Brad: yeah… too many limits on what it can work on and having to push everything in to an armature to just be able to use the tools and also hack the properties to make it work also :slight_smile: yes please
  • List of Design Principles is getting better.
  • Why do you not animate in Blender? document is getting big. We have to be careful to not dive into solutions. It is meant to show the shortcomings, the blind spots in the current design. At this point in time, a clear problem description is much more important than a possible solution.
    • Monday 3 October, we’ll have a meeting to go over each item in the doc, try and pull back and understand what the goal of the item is, what is the person trying to do, and see if that desire fits in with the design principles. If it doesn’t fit, see if we need to alter our principles or handle it in a different way.
    • Joining are: Brad, Christoph, Daniel, Demeter, Jason, Luciano, MohammadHossein, Nate, Sybren
    • The meeting will be next Monday, 17-19h CEST (Your local time: 2022-10-03T15:00:00Z)
  • Currently the people joining the workshop are: Brad Clark, Christoph Lendenfeld, Jason Schleifer, Jeremy Bot, Julian Eisel, Nate Rupsis, Nathan Vegdahl, Sybren Stüvel

Patch Review & Decision Time

  • D15855: Dope Sheet: distinguish Constant and Linear from other interpolation modes
    • Is this a desired feature? Or does it add too much noise to the UI?
    • Luciano: not against it, but it wouldn’t help him either.
    • Demeter: astigmatism checking in, I had no idea it was 2 lines until you said so.
    • MohammadHossein: it takes two clicks to check the values; is it really needed?
    • Luciano: the design doesn’t convey the meaning, will always have to check anyway. It’s the same with key colors, we should remove the meaning (“keyframe”, “breakdown”, etc.) and just use colors. Leave the meaning to the user.
    • Sybren: will suggest to create a design task for this, because nobody seems to be excited for this in particular. If the lines in the dopesheet are going to be drawn differently than before, this needs to be better designed.
    • Amelie: also, it does not make sense for grease pencil layer keyframes
  • D9199: Remove unnecessary fields from FPoint
    • Selection state of “frozen” curve points is indeed currently not used by Blender. Are there any add-ons that use this? Or is the “feature” really as unused as it seems?
    • Nobody present knows of any add-ons that use this.
    • Brad: baking down mocap data could benefit from this. What’s the 75% improvement mentioned in the patch referring to?
    • Sybren will write followup question to patch author.

Next Meeting

The next meeting will be on Thursday 13 October, 18:00-19 CEST (Your local time: 2022-10-13T16:00:00Z). Again it will be open for everybody who’s interested. The provisionary meeting agenda will be linked in the #animation-module channel before the meeting.


A very exciting idea to be able to capture motion directly in Blender! If you need a tester I have a Vive pro with a set of 5 trackers and Index finger tracking. happy to volunteer.

1 Like