Animation 2025
A New
Hope
Animation System
Sybren Stüvel
—
Nathan Vegdahl
—
Christoph Lendenfeld
<!-- Sybren --> <!-- .slide: data-state="title-page" --> <h1 class="r-fit-text"><strong>Speed</strong> Up Animators</h1> ## & <h1 class="r-fit-text">Make Animating <strong>Joyful</strong></h1> --- <!-- .slide: data-background="media/workshop-2022.webp" data-state="title-page" --> ## October 2022 # Project Starts Notes: - Workshop to look at BIG picture, allow ourselves to dream a nice future. - Defined core principles for the project. ---  Notes: - direction, good idea of where we want to go Paid: - since Nov: Christoph 2, 3 d/w devfund grant, Madrid - since April: Nathan 4 d/w, Amsterdam. Regularly concretely useful: - Brad Clark & Jason Schleifer consulting - Nate Rupsis as developer on NLA & more --- # The **Shinies**! ### Things already in Blender --- <video data-autoplay src="media/ge_speedup_comparison.mp4"></video> **FIFIDS #1: Fast** Notes: - Graph Editor like 12x as fast ---  ----- <img alt="Bone Layer Manager" src="media/blender-bone-layer-manager.png" style="width: 70%"> <small>Bone Manager add-on: https://fin.gumroad.com/l/STdb</small> ----- ## Bone Collections | Armature | Bone | | - | - | | <img alt="Bone Collections - Armature" src="media/blender-bone-collections-armature.png"> | <img alt="Bone Collections - Bone" src="media/blender-bone-collections-bone.png"> | Notes: - 'Armature Collections' was more consistent with 'Scene Collections', but every body kept using 'bone collections'. - Layers & Groups are automatically converted. ----- <!-- .slide: data-background="media/blender-bone-colors.webp" data-background-size="contain" --> Notes: - Colors in edit mode as well. - Channel list has a small swatch instead of making text unreadable. ----- | 3.6 | 4.0 | | -- | -- | | <img src="media/Blender-4.0-channel-group-colors-before.png"> | <img src="media/Blender-4.0-channel-group-colors-after.png"> | -----  <div style="display: flex; width: 100%; font-size: 60%;"> <p style="flex: auto; margin: 0;"><code>Shift+G</code> / Select » Grouped</p> <p style="flex: auto; margin: 0;"><code>Shift+M</code> / Pose » Bone Collections</p> </div> Notes: - Library overrides! New colls, can assign. - Next: Christoph --- <!-- Christoph --> ## AnimAide <video muted data-autoplay loop src="media/animaide_tools.mp4"></video> Notes: - lots of new operators for the Graph Editor - originally implemented by the creator of AnimAide: Ares Deveaux - just finished the code due to Ares' lack of time --- ## Pose Library Improvements <video data-autoplay src="media/pose_library_improvements.mp4"></video> Notes: - thanks to Julian Eisel -> Asset Shelf - tabs for asset catalogues - Flipping while blending - Subtracting poses --- ## F-curve Modifier Improvements <video data-autoplay src="media/fcurve_modifier_improvements.mp4"></video> Notes: - Multi-editing - Copy to selected - modifiers have a name --- ## Better F-curve Smoothing Operators <video data-autoplay src="media/butterworth_smooth.mp4"></video> Notes: - Gaussian (more predictable but volume loss) - Butterworth (keeps curve volume, but has ripple in impulse response) - btw -> blending edges of the selection --- ## Bone Relationship Lines |  |  | |----|----| |  |  | --- ## NLA Strip Shuffling <video data-autoplay src="media/nla_shuffling.mp4"></video> Notes: No more longer need to navigate the maze --- ## Animation Snapping Options <div style="display: flex; width: 100%;"> <p style="flex: auto; margin: 0;">Before</p> <p style="flex: auto; margin: 0;">After</p> </div> <img src="media/animation_snapping.webp" style="margin-top: 0;" /> Notes: - Nicer and more compact menu. (nice icon) - No longer per window: unified across editors. --- ## Parent Space Transforms <img src="media/parent_space_transforms.webp" style="width: 70%;" /> <aside class="notes"> You can now manipulate object and bone transforms relative to their parent's space. </aside> --- ## Many Many More Improvements Notes: - Improved animation baking - Paint modes selection - Copy Global Transform: allow mirroring of pasted transforms - Operator for quickly selecting different parts of key frames, e.g. handles - Improved bendy bone deformations - Lots and lots of bug fixes --- # Near Future Notes: - next: Sybren --- <!-- Sybren --> # Layered Animation Workshop last June <!-- .element: style="margin-bottom: 1.5em" --> **The Goal:** <!-- .element: class="fragment" data-fragment-index="1" --> ## <!-- .element: class="fragment" data-fragment-index="1" --> Speed up animators by **reducing** manual **data management** ----- ## Desires - Gradually build up animations - Try different alternatives - Adjust someone else’s anim - Procedural animation - Stream in anim from other sources - Animate on top of that Notes: - This leads to layered animation. ----- ## More Desires - Keep animation of related things together - Make all animation linkable Notes: - This leads to multi-ID animation - And NLA-like functionality. - Every object has its own NLA, making multi-object troublesome ----- | | | |-|-| |  |  <!-- .element: class="fragment" --> | ----- ### One Character  ----- ### Two Characters  ----- ### Layers <img src="media/workshop-june-2023/10-layers-basic.webp"> <img src="media/workshop-june-2023/11-layers-mix-modes.webp" class="fragment"> <div class="r-stack"> <img src="media/workshop-june-2023/12-layers-children.webp" class="fragment"> <img src="media/workshop-june-2023/13-layers-children-takes.webp" class="fragment"> </div> Notes: - Takes already very relevant to layout ----- ### Multi-Datablock Animation  <aside class="notes"> - Segway to next slides: "Layers have keys" was technically a lie → Strips </aside> ----- ### Stripification  ----- ## Channel Types | | | | -- | -- | | F-Curves | like the current Action | | Time → Index | for Grease Pencil | | ID Chooser | like camera markers, but generic | ----- # The Plan <img src="timeline.svg" class="r-stretch"> ----- ## **Minimal** data model No UI. just keyframes. `anim/animation-id-113594` ----- <!-- .slide: data-background="media/new-anim-viewport.webp" data-background-size="contain" --> ----- <pre class="r-stretch"><code class="hljs language-py" data-trim data-line-numbers="1-3|4-13|15-17|19-21|23-35"> ob_cube = D.objects['Cube'] ob_suzz = D.objects['Suzanne'] # Create Animation data-block. anim = D.animations.new("anim") # Create outputs for the objects. out_cube = anim.outputs.new(ob_cube) out_suzz = anim.outputs.new(ob_suzz) # Assign the animation. ob_cube.animation_data_create().animation = anim ob_suzz.animation_data_create().animation = anim # Create a layer. lay = anim.layers.new('The One Layer') strip = lay.strips[0] # Animate Cube. strip.key_insert(out_cube, "location", 2, 0.0, 1) strip.key_insert(out_cube, "location", 2, 3.0, 60) # Animate Suzanne. def key(out, rna_path, array, frame): for idx, val in enumerate(array): strip.key_insert(out, rna_path, idx, val, frame) key(out_suzz, "location", (3, 0, 0), 1) key(out_suzz, "location", (3.2, -2.2, 2), 60) key(out_suzz, "rotation_euler", (0, 0, 0), 1) key(out_suzz, "rotation_euler", (0, 0, radians(720)), 60) key(out_suzz, "scale", (0.8, 0.8, 0.8), 1) key(out_suzz, "scale", (1.6, 1.6, 1.6), 60) </code></pre> ----- <!-- .slide: data-background-video="media/animation-data-model-single-layer.mp4" data-background-size="contain" --> ----- <!-- .slide: data-background-video="media/animation-data-model-two-layers.mp4" data-background-size="contain" --> Notes: - Next: Christoph --- <!-- Christoph --> # Keying Redefining keying sets and auto-keying behavior. Notes: - overhaul - instead of getting list of keying sets ----- ## Just Press "`I`" <video data-autoplay muted loop src="media/insert_keys.mp4"></video> Notes: - one part: press I -> get keys - keying sets will still be around -> but relative (users in the room?) - more predictable autokeying --- # Also Future Notes: - Looser planning for the future - still not too distant - these will directly speed up animators --- # Ghosting <video data-autoplay src="media/ghosting_system_demo.mp4"></video> Notes: - existing prototype by Falk - common ghosting system between GP and other objects - has a lot of technical hurdles to overcome - shouldn't slow down the viewport - fancy feature, need to make a solid foundation first - Next: Nathan --- <!-- Nathan --> # Selection Sets & Bone Pickers ----- ## Taming Complexity <img src="media/rig_complexity.webp" style="width: 50%;" /> Notes: - Rigs and scenes: complex and difficult to manage. - Rig layers: static, per-rig, not per-shot. - Anything that comes with the rig: not tailored to the animator or shot. - Not just organization: it's also a workflow problem. - Quickly scoping down what you're working with. - Fast and accurate selection. - Keyframing what you intend. - Etc. ----- ## Selection Sets <video data-autoplay onloadstart="this.playbackRate = 1.5;" src="media/selection_sets.mp4"> Notes: - Simple, animator-centric organization of rig controls. - Quickly select sets from a pop up. - Focus. - Key. - Create sets as needed. - Take it with you: across shots and across characters. - Already exists as an addon. ----- ## Bone Picker <img src="media/bone_picker.webp" style="width: 70%;" /> <div style="font-size: 0.7em;">(mockup)</div> Notes: - Not currently in Blender. - Separate pane, for easy selection regardless of 3d view. - De-clutters the 3d viewport. - Good for "controls" that don't really belong in 3d space. ----- ### Arranged for Selection <img src="media/bone_picker_hand.webp" style="width: 70%;" /> Notes: - Arrange for fast selection, not accurate proportions. - Example: quickly box selecting specific digits on all fingers. - Not just selection: buttons could trigger scripts/operators. - Goal: easily built and customized by animators themselves. ----- How to bring this all together into a coherent whole? Still working on that. <!-- .element: class="fragment" --> Notes: - We've shown two solutions: each tackle different complexity issues. - But they're also disjoint. - We want complementary solutions, not just one-off features. - How to bring this together? Still working on it. - Goal: go further and find powerful ways for animators to tame complexity. --- ## Even Further Future <div class="r-stack"> <img class="fragment current-visible" src="media/rignodes-prototype.webp"></img> <ul class="fragment current-visible spaced"> <li>Rig Nodes (prototype)</li> <li>Custom Bone Axes</li> <li>Animation snippets in pose library</li> <li>More rest poses</li> </ul> </div> Notes: - A whole bunch of other things we want to do down the road. - Rig nodes Python prototype. - We've been exploring! - ... - First we need to build the foundations, so not for 2025. - But 2025 is just the start. We'll build on those foundations. --- # Thank You ## Questions!