VSE Effect Assets with Encapsulated Animation and Exposed Parameters

Blender's recent VSE Compositor integration is a very promising foundation for user-defined video effects.

With Compositor Effects, Compositor Modifiers, exposed node group inputs and reusable node group assets, Blender is now close to having a system where effect authors can build complex effects while video editors only interact with a small set of useful parameters.

I think this abstraction could become an important design goal for the VSE.

Related work

This proposal builds on ideas that have already been discussed and, in part, implemented:

Per-Clip Compositing Node System in the Video Sequence Editor!

blender.community/c/rightclickselect/P5Z2

Nodes with VSE

blender.community/c/rightclickselect/TLfbbc

The goal of this proposal is not to request the already implemented per-strip Compositor functionality again. Instead, it is about taking the next step: making reusable VSE Effect Assets capable of encapsulating animation, timing and 2.5D/3D-style transformations so that the person using the effect does not need to understand the Blender systems used to create it.

The problem

A Blender user who uses Blender primarily as a video editor should not have to understand every Blender subsystem that is required internally to produce an effect.

For example, imagine a transition where the incoming clip:

• starts small at a configurable position,
• grows according to a non-linear curve,
• optionally stops growing for part of the transition,
• continues growing afterwards,
• simultaneously rotates like a flat card in perspective,
• has independently controlled rotation timing,
• and finally fills the frame.

Blender can already provide many of the individual building blocks required for something like this.

However, a video editor should ideally not need to understand:

• keyframes,
• F-Curves,
• the Graph Editor,
• matrices,
• camera projection,
• compositor nodes,
• or Blender's 3D workflow

just to use such a transition.

Those concepts may be necessary for the person who creates the effect, but they should not necessarily be exposed to the person who uses it.

Effect authors and effect users have different needs

VSE Effect Assets could make this distinction explicit.

An effect author might build something internally from many nodes and advanced Blender concepts:

Input Clip
    ↓
Timing Functions
    ↓
Scale / Position Animation
    ↓
3D-style Transform
    ↓
Camera Projection
    ↓
Perspective Warp
    ↓
Compositing
    ↓
Output

But the video editor could simply see:

Start Position
Start Scale
Growth
Hold
Flip Angle
Perspective
Rotation Timing

This is similar to the advantage of a Geometry Nodes asset: the author can create a complex procedural system while the user only sees the parameters that were intentionally exposed.

Normalized strip-local time

Compositor Effect strips already have a very useful concept: an effect factor that progresses from 0 to 1 over the strip duration.

This could become a fundamental building block for reusable VSE Effect Assets.

An effect should be able to derive multiple independent timing curves from this normalized time.

For example:

Effect Time:  0 ----------------------------- 1

Scale:        grow -- hold ------ grow
Rotation:     wait -------- rotate ----------
Position:     move quickly ---- settle -------
Blur:         high ---- low -----------------

This would allow an effect to adapt automatically when the transition strip is made shorter or longer.

Ideally, effect authors should be able to define such behavior internally so that the final user does not have to create or edit Blender keyframes.

It would also be useful if timing curves or simplified curve controls could themselves be exposed as parameters where appropriate.

2.5D / 3D-style transforms

Many common motion-graphics and transition effects are difficult to express using only traditional 2D position, scale and rotation.

Examples include:

• card flips,
• page-like rotations,
• perspective slides,
• rotating video panels,
• photo walls,
• screen replacements,
• carousel transitions,
• pseudo-3D title and picture animations.

For these effects it would be valuable if reusable VSE effects could perform 3D-style transformations and camera projection internally.

This does not necessarily mean creating actual scene objects.

A lightweight 2.5D system based on transformations, rotation, projection and image warping may often be more appropriate for a video effect.

The important point is that the effect author should be able to use Blender's powerful mathematical, compositing, animation and potentially 3D-related capabilities without requiring the effect user to understand them.

Blender as a video editor without requiring animation or 3D knowledge

An important use case is someone who uses Blender only for video editing.

Such a user may never need the 3D Viewport, materials, rigging or Blender's general animation workflow.

That should not prevent them from using an effect that internally makes use of concepts from those systems.

For example, a user should be able to add a "Flip + Grow" transition and see only:

Start Position
Start Size
Growth
Hold
Flip Amount
Perspective
Timing

The user should not need to know that internally the effect might use interpolation curves, matrices, projections or other advanced Blender functionality.

This separation is important:

The author of an effect may need to understand how it works. The user of an effect should only need to understand what it does.

Advanced Blender knowledge should increase what a user can create, but it should not be a prerequisite for using effects created by other people.

Why this could be especially useful for Blender

This could allow Blender's video effects library to scale without requiring Blender developers to implement every transition and motion effect individually.

Blender developers provide the general infrastructure.

Effect authors build reusable effects.

Video editors use those effects without needing to understand their implementation.

That could lead to community-created libraries containing effects such as:

Flip + Grow
Card Reveal
Perspective Slide
Photo Pop-In
Split Screen Animation
Page Turn
Picture-in-Picture Motion
Animated Lower Third
Screen Replacement

while all of them are built from the same general-purpose Blender infrastructure.

Instead of Blender needing hundreds of individually hard-coded effects, the community could build them from a relatively small number of powerful primitives.

Goal

The goal is not to add one specific transition.

The goal is to continue developing VSE Compositor Effects into a reusable effect-asset abstraction where:

• complex internal setups can be hidden,
• meaningful parameters are exposed,
• timing is relative to the strip,
• animation behavior can be encapsulated,
• independent timing curves can control different parts of an effect,
• 2.5D/3D-style image transformations are possible,
• effects can be stored and shared as assets,
• and effects can be used by people who may know nothing about Nodes, Animation or Blender's 3D tools.

The underlying principle is:

Blender's internal complexity should make VSE effects more powerful without making them more complicated for the video editor using them.