# Animation 2025 Project Schedule

This is the development schedule for Blender's Animation 2025 project. This is a living document, not a static schedule set in stone.

## 2025 Targets

The animation project has a lot of lofty goals, and we do want to get to all of them eventually. But we won't get anywhere if we don't define and take concrete achievable steps. Therefore, we are limiting our 2025 development targets to a subset of our larger eventual goals.

The following are our 2025 targets, for Blender 5.0:

- Implement the new base data model:
- `Animation` datablock
- Multiple outputs (i.e. containing animation for more than one ID at a time)
- Animation Layers
- Mix modes:
- Replace
- Combine
- etc.
- Layer influence
- Hierarchical layers
- For organization
- For "takes"
- Animation Strips
- Keyframe Strips
- Reference Strips (i.e. multiple instances of the same strip)
- Anim Strips (strips that reference other Animation datablocks)
- Modifier / Generator strips
- Grease Pencil integration
- Design and implement the UI/UX for the above data model.
- Implement additional features needed to work effectively and joyfully with the new data model:
- Baking for strips and layers.
- Tools for managing your animation data:
- Combining/splitting animation layers.
- Combining/splitting `Animation` datablocks.
- Production tools for library-linking animation data and hooking it up.
- And more TBD as we gain experience during the prototype phases.

More broadly: **in addition to the new capabilities** intrinsically enabled by the new animation data model, the target is to **achieve parity with the old animation system**, but using the new data model. Parity here does not mean strict one-to-one feature parity, but rather means that all use cases that the old system handles should also be handled well by the new system. In other words, all users should be able to switch to the new system for at least new projects. This will set the foundations for future work.

## Out of Scope

We are **explicitly leaving the following features out of scope** for the initial 2025 target:

- Animation Nodes.
- Rig Nodes and associated features like slappable rigs, mid-animation rig swapping, etc.
- Animation-level constraints and drivers.
- Viewport-editable motion paths.
- A new f-curve data model, and associated features like better quaternion rotations.

**We still want to do these.** Moreover, we still need to keep them in mind during this stage and develop some sense of where we're going with them, to help ensure that our decisions at this stage are compatible. But actually finalizing designs and implementing them will come some time after we reach our 2025 targets.

One step at a time.

## Project Phases

To help define our schedule, we have split the project into six phases. The phase progression is designed to get working (even if incomplete) versions of features into users hands as quickly as possible, to **get real hands-on feedback from users** as early as possible, and **course-correct based on that feedback**. This course correction may involve both UI and data-model level changes as we go.

This does mean that the later phases are currently less detailed than earlier phases, because the specific design and implementation work for those later phases depends on the feedback we get in earlier phases. Nevertheless, we have at minimum defined the key goals of each phase.

## Phase 1: Minimal Working Data Model

**Key goals:**

- Implement a minimal working version of the new data model in Blender, and test it for technical soundness and fitness-for-purpose.
- Create designs for an initial UI/UX for the minimal data model, to be implemented in phase 2.

**Details:**

- Minimal `Animation` datablock implementation:
- Outputs
- Animation layers
- Layer mix modes
- Replace
- Combine
- Layer influence
- Keyframe strips
- Infinite
- Finite
- DNA & RNA
- Dependency graph stuff.
- Blend file IO.
- `AnimData` struct changes.
- Verify technical workflow:
- Library linking in production cases.
- UI/UX mockups and design for minimal data model implementation.
- What editors will we have, how will they work, and what will they be responsible for.
- Figure out how animation filtering should work from a UX perspective.

## Phase 2: Minimal Working UI

**Key goals:**

- Implement and iterate on an initial working UI, covering all the features enabled by phase 1's data model implementation.
- Adjust and flesh out the design of the complete 2025 data model.

**Details:**

- Implement the UI/UX design from phase 1.
- Implement the animation filter system.
- New filtering system for the `Animation` data-block.
- Later phases: hook up already-existing animation sources to the new filtering system.
- Later phases: remove the old filtering system, because everything is now available in the new one.
- Design the fleshed-out data model for phase 3.
- **Data-model-level** parity with the old system.
- This doesn't mean it has the same features, it means old-system animation can be converted.
- Animation Layers:
- Hierarchical layers
- For organization
- For takes
- More mix modes.
- Animation Strips:
- Animation Datablock strips
- Modifier strips
- Grease Pencil integration
- Figure out the general direction for and nail down the use cases for the "out of scope" features listed earlier in this document.
- The intention here is to ensure sure we make any small tweaks to the data model that could make these future features easier/cleaner to implement and integrate.
- Figure out how the pose library relates to the old and new data model. (A.k.a. what to do with Actions)

## Phase 3: Fleshed Out Data Model

**Key goals:**

- Implement the fleshed-out data model planned in phase 2.
- Grease Pencil integration.
- Hierarchical animation layers.
- Etc.
- Create designs for an initial UI/UX for the additional features in the fleshed-out data model, to be implemented in phase 4.

## Phase 4: Fleshed Out UI

**Key goals:**

- Implement and iterate on the UI/UX designed in phase 3, covering all core features of the 2025 data model.

## Phase 5: Stabilize

**Key goals:**

- Final polish on existing features and UI/UX.
- Prepare to go mainstream.
- End of phase: remove experimental flag.

## Phase 6: Port Remaining Tools

**Key goals:**

- Start of phase: in use by Blender Studio.
- Get to parity with the old animation system.

## Phase 7: deprecate old system.

- Blender 5.0, deprecate old animation system.

## Backwards Compatibility Story

During the initial development phases, we don't want to bog down our rate of progress and iteration by having to maintain backwards compatibility (either with the old system or within the new system itself). The name of the game in the earlier phases is iterate, iterate, iterate.

However, we of course need (and want) to work towards backwards compatibility for the final release. And even before final release, for the sake of longer-form testing on more realistic test projects, we'll need to maintain a certain level of backwards compatibility for testers.

With that in mind, we've organized the phases into blocks with different backwards compatibility goals:

- Phase 1:
- Zero effort put into new-system backwards compatibility. Lots of breakage.
- No real conversion support for old-system animation data.
- Working in a branch.
- Phase 2:
- Backwards compatibility provided when easy to do so. Still lots of breakage.
- No real conversion support for old-system animation data.
- Working in main, behind an experimental feature flag.
- Phases 3 & 4:
- Backwards compatibility provided when doing so isn't a major time sink. Breakage may still happen regularly, but not arbitrarily.
- Partial conversion support for old-system animation data, where it's straightforward to implement.
- Working in main, behind an experimental feature flag.
- Phase 5:
- Backwards compatibility provided whenever possible. Breakage still allowed if strictly needed.
- Full conversion support for old-system animation data.
- Working in main, still behind an experimental feature flag.
- Phase 6:
- Strictly maintain backwards compatibility.
- Full conversion support for old-system animation data.
- Released in main, non-experimental, exists concurrently with the old animation system.

## Schedule

- https://docs.google.com/spreadsheets/d/1L4uTHV1YggD0YvpXIwWrjYTwSH5f4UcDrRBvIpNhl20/
- https://docs.google.com/spreadsheets/d/1ON5Plh0F0sgWLnjFC7QOtqLzsIY4uWdjpo8aE-hPBI4/