I couldn’t find a thread to discuss the general VSE development in Blender so I thought I’d start one.
I was looking for a way to use Blender for quick animatics/previs at our studio and was looking at leveraging the Video Sequence Editor and the new Grease Pencil animation tools to do so, but was quickly dismayed at discovering two things:
We can no longer add the current scene to VSE. I thought this might be a bug at first, but looks to be an intentional decision. Does anyone know why this decision was made?
It’s an inconvenient UX decision but I can’t find a discussion on this anywhere.
I had a very simple test scene with two drawn elements and one moving Suzanne, and Blender was lagging horribly. Then noticed just a single, static cube lags horribly as well.
Is this the behavior everyone’s seeing currently?
It seems logical to clarify that a scene is not at same time its self input and output. Previous behavior was weird.
In 2.79, you can set a montage of several scene strips of same scene.
And scene datablock referring to this montage or to 3D scene used by strips is the same.
Rendering the montage or the 3D scene is just relative to enabling/disabling sequencer under post-processing panel.
Forcing people to create a scene for montage disconnected to any 3D scene is not as fast and practical.
But it is clearly less disturbing for new users and probably easier for devs to track VSE/rendering bugs.
The scene polysemy is avoided.
What I mean is that default scene datablock named Scene contains not just 3D scene data.
It also contains video editing data.
For both, you are using the same Scene name.
When you add a scene strip to VSE, its length is relative to duration indicated in Render settings of the scene.
Scene strip is what Blender would render without video editing.
But because you are making a video editing of that result with sequencer, when you press F12, you are obtaining another result.
Blender introduces a confusion between 3D Scene Rendering and Final Output Rendering.
If there is Video Editing data in a file, you open without launching a workspace with VSE, you could think that Rendering is broken because it does not look like what 3D Viewport is showing.
And about the code, it probably causes looping problems to have different levels of data referring to same scene level.
I really don’t understand the technical reason for the change (and thanks, zeauro, for trying to explain), but I guess the end result is that it was causing many issues that this is the quickest solution to the problem.
Sadly, this does make things slightly inconvenient. And in the case of Previz Camera Tools, basically killed its development as it relied on being able to add the current scene to VSE. Unfortunate, but hopefully there could be be better solution in the future.
That said, can anybody else confirm that adding a scene in in VSE in 2.8 is laggy? Simple test, just add an empty scene into VSE.
I’m guessing (hoping) it’s due to it still being in beta, but just want to confirm it’s not just me.
VSE OpenGL preview seems to need some work.
It does not render Wireframe. Its solid mode is shadeless. Its look dev mode does allow to change HDR background.
In other words, this preview have 2.79 form without Shading popover controls that should be expected in 2.8.