[Bf-taskforce25] Unique sequence strip names

[Ton Roosendaal]

Hi,

We went over this with some people in irc, a cool and interesting 
discussion! :)
Tentative conclusions are (hope I word it correctly):

- names for strips can (optionally) be enforced unique
- having unique names should not be a requirement for the operators to 
work though.
- we need to devise a good internal api (Blender side) that allows 
selection groups to exist, defined either by context (user input), 
stored, or created on the fly. The operator api can use this via 
itterators, python can define them.
- generic 'selection groups' will also do great for future expansion to 
nodes, think of bones groups, vertexgroups, objects, whatever. Same 
goes for future construction history attempts. :)

Promoting sequencer data to ID level ("Library") is a great idea, but 
it shouldn't be done on individual strip level, instead it's better to 
convert the current "Editing" struct into a generic "bSequencer" (or 
"bMovie" ?) ID library data block. That way we have the freedom to 
either re-use a single strip or a complete movie edit.

That new ID entitity then can be used all over in Blender, like for 
textures, but also again to insert in the sequencer as meta strip.

The current "Image" and the new "bMovie" can get a similarly functional 
api, giving image data on request by a user of it. Expanding on the 
idea, we should also allow:

- to insert an Image strip in Sequencer (Blender Image)
- to output the Sequencer result to either Image or Render Result
- Compositor to output to Image.
- Compositor to include a 'bMovie" node.

Here we get a nice generic way to have both sequencer and compositor 
cooperate, with complex dependencies, but that's what we have depsgraph 
for! :)

-Ton-

------------------------------------------------------------------------
Ton Roosendaal  Blender Foundation   ton at blender.org    www.blender.org
Blender Institute BV  Entrepotdok 57A  1018AD Amsterdam The Netherlands

On 26 Jan, 2009, at 1:41, Campbell Barton wrote:

> Some of the sequencer operators would benefit from sequencer strips
> having unique names, (Sequence * types)
>
> adding an effect for instance, would be nice if you could give the
> operator args like this...
> bpyop.sequencer.add_strip_effect(type='CROSS', seq1="anim1", 
> seq2="black")
> ...rather then setting up the selection and running it.
>
> For most functions unique names per metastrip would be enough but
> since you can make and separate metastrips while editing, making the
> names per scene seems less likely to cause collisions while editing.
>
> Even though the Sequence struct is already pretending to be library
> data, I cant see any reason to make strips true  library data (other
> then for consistency).
> It would seem enough that each strip has a unique name per scene.
> since Sequence strips have no inter-scene relationships.
>
> The main advantage with strips being library data is for use in 
> multiple scenes.
> But this (like multi user objects), would mean the position in the
> animation would also be shared between scenes, so it seems a lot less
> useful then a camera in multiple scenes for eg. A strip in multiple
> scenes would be problematic if scenes frame rates dont match.
>
> having real Sequence Strip library data would allow appending strips
> from another blendfile and access strips globally by name, but I'm
> still not convinced its worthwhile.
>
> Sequence strip names can work like bone names, do_versions for older
> files can make the names unique, nothing currently relies on matching
> the names so this wont break anything.
>
> Id like to go ahead with this soon, let me know if there is any
> problems with this proposal.
>
> -- 
> - Campbell
> _______________________________________________
> Bf-taskforce25 mailing list
> Bf-taskforce25 at blender.org
> http://lists.blender.org/mailman/listinfo/bf-taskforce25
>