I’ve started work on a new keyframe graph with Blender-like handling for the scrollbars and zoom, and as I’m going in and reworking the core architecture to support all of the fundamental stuff like save/load and parenting and precomping and all that (as well as look at some popular AE plugins) I realized there’s a lot of things I’ve taken for granted about the way AE and most video editors work that don’t actually have to be there. I’m not saying I’m gonna implement all of these ideas or even if they’re any good in the first place, but I’d like to share them anyway, sorted from least to most extreme, to demonstrate that they are in fact possible.
Precomps can still be editable from outside comps…
A way of exposing folder-like dropdowns for precomped layers is probably the biggest thing I’ve seen people talk about, and I even saw an Adobe employee in the chat of a recent livestream hint that this feature is actively being worked on for AE. It’d be pretty trivial for me to implement too, but AE is probably gonna beat me to it. I’m gonna go out on a limb though and say they still won’t fix another major movement-related frustration I’ve seen myself and others face, which is not being able to scroll past a composition’s beginning/end times to access keyframes/layer handles that are beyond the boundaries. The way I’ve wired up my new grah it already allows for it by default.
The sidebar doesn’t have to be there…
A decent bit of screen real-estate is lost from the sidebar and all of its switches, and I could make more of the timeline visible by tacking the information onto the layer itself. Some of the buttons like the audio toggle and video frame interpolation toggle aren’t even applicable for many layers. This change would likely be less intuitive by itself, but I introduce it mainly to combine it with the next ideas:
The scale of layer bars doesn’t have to be locked horizontally…
It’d be nice if the timeline was more like a node graph that you could quickly zoom out to see the project as a whole and then back in to focus on smaller pieces of. Or if you could scale up one particular layer/precomp to make it stand out from the rest. As you can imagine the sidebar is what’s holding a program back from implementing this.
Layers don’t have to obey any rules on vertical alignment…
To turn these even further into nodes there isn’t any technical reason for AE’s decision to lock one layer to one horizontal “slab” on the screen or to pack the layers right next to each other. There doesn’t even have to be a concept of video “tracks” like in Premiere that you always have to right click -> add track to get more of; layers can be positioned at arbitrary and odd increments. All that matters is whether one layer is on top of another. The only thing this affects that I can think of would be the visible index number of layers for scripting. And now that we’re this far:
We might as well slap nodes right onto the timeline…
When I say “nodes” I don’t mean a fully-fledged node-based compositing setup, of course, but for what you’d typically use expressions for. Now this is just a half-assed assumption with no real backing to it, but I imagine it’d be more intuitive for many to plug in a wiggle or parent node directly into the layers and be able to see the connections always rather than having to use expressions. I’m not sure the exact system the nodes would follow, but we’d likely always be trading usability for the more versatile options we’d get from a scripting language.
Text could be a bit wilder…
This one is unrelated to the previous and something I’m not sure how to explain, but it might be nice if text by default was broken up into layers that you could manipulate right out of the box with the same tools as everything else rather than having to rely on a separate series of text animation functions that only work for text specifically. I dunno; I don’t do a lot of typography animation, so maybe people don’t have any qualms with how it works now. Also kinda related to the shape layer repeater things in AE. I haven’t thought much about procedural/C4D mography stuff, but I’d want it to be flexible out of the box and work with vector shapes/text and raster layers.