Modifiers were being mistakenly recalculated at every frame as long as the object had animation, slowing things down due to incorrect depsgraph recalc tags.
Renamed OB_RECALC -> OB_RECALC_ALL to reduce future confusion. During this process, I noticed a few dubious usages of OB_RECALC, so it's best to use this commit as a guide of places to check on. Apart from the place responsible for this bug, I haven't changed any OB_RECALC -> OB_RECALC_OB/DATA in case that introduces more unforseen bugs now, making it more difficult to track the problems later (rename + value change can be confusing to identify the genuine typos).
This was caused by the multi-user data appearing multiple times in the channel list. Now most editing functions filter out duplicates before doing anything to prevent these problems.
Hopefully the additional cost of filtering the entire list an extra time won't be too much of a speed/mem hit...
* Copy/Paste operators for F-Modifiers
Available in Graph and NLA Editors. Use the Copy/Paste buttons beside the 'Add Modifier' buttons.
Copy copies all the modifiers of the ACTIVE F-Curve or Strip depending on the editor.
Paste pastes modifiers from the buffer to all the selected F-Curves or Strips, adding the new modifiers to the ends of each list.
* 'Stepped Interpolation' F-Modifier
This modifier holds each interpolated value from the F-Curve for several frames without changing the timing.
This allows to preview motions 'on-twos' for example without altering the timing, or having to go through setting heaps of keyframes. In this case, Andy wanted to use this for CG <-> StopMo.
Texture animation is now shown in the animation editors. Texture stacks are shown for each Material/Lamp/World block that uses them.
There is currently still a bit of a bug with this which means that unless the owner of the texture stack is animated too, the animation data for the textures won't show up. This will get rectified soon though.
- 'Toggle' operators for channel settings now now act more like the select-all type of "toggle" operator. The old behaviour has now been moved to "invert".
- Channel settings are now flushed (like for visibility and when clicking) for muting and locking when using the operators
* Fixed problem where selecting an individual F-Curve would not set the selection correctly.
Group channels still needed a separate selection check before they get included in the filtered list. I had removed this in an earlier fixing commit today, but overlooked that expanded groups wouldn't get this check. Therefore, group channels would also be flushed on, turning all channels of group on.
* Removed the 'curvesonly' test from deciding whether the selection status + collapsed group fix, from the earlier commit, since this was making a few cases get overlooked (namely for setting visibility toggles, where selected F-Curves in closed and deselected groups still managed to get through)
* Added a debugging print API call for helping with debugging this sort of error in future. It just prints the types of channels being operated on, to easily see what's going on...
This commit introduces a few cleanups and tweaks to the way that timecodes (i.e. the timing indications used instead of frame numbers) get displayed.
1. Custom Spacing of TimeCodes/Gridlines
Made the minimum number of pixels between gridlines/timecode indications a user-preference, instead of being a hardcoded constant. This allows to set the spacing tighter/looser than the defaults, and is also used for the other changes.
2. Default timecode display style, (now named 'minimal') uses '+' as the delimeter for the sub-second frames. This hopefully makes it a bit clearer what those values represent, as opposed to the '!', which can sometimes look too much like a colon.
3. Added various timecode display styles as user-preference. - These include always displaying full SMPTE, to showing milliseconds instead of frams for sub-second times, and also an option to just show the times as seconds only.
- When changing the timecode style, the spacing setting is automatically modified so that the timecodes are spaced far apart enough so that they won't clash (under most circumstances). This automatic modification is only done if the spacing is too tight for the style being set.
4. Unified the code for generating timecode strings between the View2D scrollbar drawing and the current frame indicator drawing.
Added flags for the mapping function offering more control over what values were affected and when. In the WIP code, the values for unselected verts were getting the restoration conversion being repeatedly applied, resulting in the values tending to zero.
F-Curves now internally store radians again instead of degrees.
- This solves problems with inconsistencies when working with drivers.
- No need to version patch old files, potentially screwing them up. As such, removed the version patching for F-Curves.
- Is better suited to optionally showing radians throughout the UI instead or degrees.
As a result, values are now converted on the fly in the Graph Editor for display and operators that operate on values. I've made the conversion system for this rather general, so that other unit type conversions can also be hooked up with the type conversion backend.
Also, made some tweaks to F-Curve RNA wrapping to make it represent the data better.
TODO:
- Transform code currently still needs to be corrected to work with these changes. Currently moving keyframes for rotation curves will make them change too rapidly vertically when using degrees.
Action Groups can now be expanded/collapsed in DopeSheet editors without the same thing happening in the Graph Editor, and visa versa.
This should help improve the workflow, since the channel lists are generally kept more compact in the DopeSheet, while they are more expanded in the Graph Editor, so less time is spent expanding/collapsing stuff. Also this should hopefully alleviate some of the errors from accidentally deleting and then having to restore channels that were not intended to be deleted.
Also, switched the order of the expand/collapse hotkeys (in the channels list region) for channels so that Ctrl +/- now expands/collapses selected channels only, while +/- expands/collapses all channels. This should make it more convenient to quickly open up all groups to select F-Curves for the Graph Editor.
The reverted code was just blindly restoring the old state of the keyframes; changes to selection state, value changes, handle type, etc. were overridden, and the cases where keyframes were deliberately retimed or otherwise were also ignored.
I'm not sure what problems these changes were meant to be solving, but will reassess the situation when I get more info on this.
and other operations the nla mapping would be applied to the curves, but not
restored correctly. The unmap function was not the inverse of the map function,
and it's not clear to me it's even possible to make it so due to repeat, so
now the old coordinates are backed up in a list and then restored afterwards.
Protect and Mute toggles now flush their values when changed, like for visibility, making the workflow a bit smoother. Currently, this only takes effect when clicking on the toggles (i.e. the hotkey+select based toggle setting operators don't take this into account yet).
A solid color backdrop is now drawn behind the mute/protect toggles and sliders, reducing the visual clutter with long names still appearing behind the UI widgets.
* Mesh data/settings can now be animated. It is not recommended that geometry be animated directly, but other settings such as autosmooth, etc. can be...
* Code cleanups for depsgraph, making sure that drivers get included for all object data types.
This commit brings back the functionality whereby the selection of F-Curves and/or relevant Group(s) is synced with the selection status of the corresponding bones when the bones are selected/deselected.
* Added ANIM_list_elem_update and ANIM_id_update functions to call when
changing animation curves, which will then call the RNA property update
functions for those curves.
This is to replace just calling DAG_id_flush_update, that may not always
be the right thing to do, and doesn't send proper notifiers for redraw.
Still only used/usable when transforming in the graph editor, not sure
how this do this with NLA for example, .. needs to be improved.
* Added function wm_data_handle_update function to contain the object
update function, and also added scene animation update there. Actually
it should be doing all datablocks, this makes it work for sequencer.
Joshua, do you agree this is the right direction to go in? I can revert or
change the code if you think it should be done differently. Mainly wanted
to get this working well for sequencer now.
This commit fixes#19908 and #20239. Deleting keyframes will now delete the F-Curves they came from too, if the F-Curves don't have any more keyframes and/or F-Modifiers providing any further motion info.
Improved the hotkeys for toggling the visibility of channels in the keys area for the Graph Editor.
* VKEY - this is now used for making only the selected channels visible, hiding everything else
* Shift-VKEY - the old toggling behaviour
In addition to this, I've made these toggling operators flush the visibility flags up/down the hierarchy, just like clicking on the channels manually do. There are still a few minor oddities to iron out, but it should be better than before.
Also, fixed a bug with these toggling operators introduced during my earlier commit to make filtering work ok. It's always tricky getting these layers of checks just right, so hopefully nothing breaks now again...
* Removed ED_anim_dag_flush_update and ED_anim_object_flush_update. These
were wrapping DAG_* calls and were intended be used instead of them when
doing a DAG update from editors. That goes against the design in my opinion,
no matter who calls the DAG, that should update the editors correctly, so
any special checks in such functions for editors should be avoided.
* Driver RNA properties now do updates again, including DAG scene sorting,
text buttons no longer update as you type anymore, so this should be
safe I think.
* Remove scene.update() RNA function, all properties/functions should do this
automatically, if changing some property or calling a function/operator
does not do the correct update, that should be fixed.
Now when nodes are keyed, they will show up in the dopesheet/graph editor/etc in a new 'Nodetree' category.
Still a major problem left, nodes need unique names in order for the rna paths to hold animation data properly...
* When there is only a single keyframe for a F-Curve, the handles aren't shown anymore. This looks nicer than the fat orange blobs that appeared
* Tweaked the management of GL_BLEND when drawing animation channels in the Graph Editor in an attempt to fix some of the missing text drawn issues.
* Converted the properties panel to use layout engine + added color selectors
Special priority request from Durian team to get this sub-editor of the DopeSheet Editor restored. Originally I was kindof planning to drop it, but obviously it still has a role!
It now supports all the modern features that the DopeSheet supports, complete with selection, muting, locking, DopeSheet summary, and all the other tools that you know and love from the other views.
Also, this no longer uses the old hacky sliders that 2.4x used (instead it uses RNA-based ones), so should function just the same as other DopeSheet views).
Added a summary channel that appears as the first channel in the DopeSheet. For now, this is disabled by default, but can be enabled using the 'Summary' toggle in the header between the mode selector and the standard filtering options. This has been done, since there is a possibility that it will make the DopeSheet run a bit slower.
In this channel you can do everything that you can normally do with DopeSheet channels (i.e. select, transform, edit, etc). It might be worth noting though that care probably needs to be taken when trying to use Copy/Paste, since that is still a bit fidgety...
In the process, I've fixed a few bugs, mostly with selection:
- Selecting keyframes in scene summaries wouldn't work
- Border select only worked in F-Curve and Group channels
Keymaps are now saveable and configurable from the user preferences, note
that editing one item in a keymap means the whole keymap is now defined by
the user and will not be updated by Blender, an option for syncing might be
added later. The outliner interface is still there, but I will probably
remove it.
There's actually 3 levels now:
* Default builtin key configuration.
* Key configuration loaded from .py file, for configs like Blender 2.4x
or other 3D applications.
* Keymaps edited by the user and saved in .B.blend. These can be saved
to .py files as well to make creating distributable configurations
easier.
Also, user preferences sections were reorganized a bit, now there is:
Interface, Editing, Input, Files and System.
Implementation notes:
* wmKeyConfig was added which represents a key configuration containing
keymaps.
* wmKeymapItem was renamed to wmKeyMapItem for consistency with wmKeyMap.
* Modal maps are not wrapped yet.
* User preferences DNA file reading did not support newdataadr() yet,
added this now for reading keymaps.
* Key configuration related settings are now RNA wrapped.
* is_property_set and is_property_hidden python methods were added.
- rename "Nurb" to "Spline" in RNA, eg. bpy.data.curves[0].splines[2].type == 'NURBS'
from a user perspective spline is a more generic term while Nurb is misleading when used for beziers and poly lines.
- added curve.active_spline property so the python UI can display the last selected curve.
- set the active spline when entering editmode (uses first selected spline)
- added back Hide Handles as a curve property (removed the global flag), access from the view panel in editmode.
- added hide normal option for curve, normal size access for curve and mesh display.
- changing orderU/V, endpoints, cyclic, bezierU/V now work in editmode and calls update functions.
- entering editmode was crashing with text objects
- curve.switch_direction() crashed (own fault from last commit)
- Tkey for tilt was overridden by Toolbar, made Tilt Ctrl+T.
- OBJECT_OT_mode_set check for compatible modes before running - so curves dont try go into paint mode with V key for eg.
It is now possible to tag certain keyframes as being 'breakdowns' in the DopeSheet. Breakdown keyframes are drawn as slightly smaller blue diamonds.
Simply select the relevant keyframes and use the RKEY hotkey (or from the menus, Key->Keyframe Type) to choose between tagging the keyframe as a 'proper' keyframe and a 'breakdown' keyframe.
Notes:
* Please note that this feature does not currently imply anything about breakdowns moving around keyframes or behaving any differently from any other type of keyframe
* In future, if there is any such need, more keyframe types could be added, though this is not really likely at all
Now the mute/protect/expand/etc. toggles are drawn using UI widgets. This means that special event handling code to determine when they were clicked on is no longer needed, and also means that there can now be tooltips for these items too.
Also, added visibility toggles for ID-block expanders, which will cause all the F-Curves in the linked datablock to not get drawn. The backend filtering code to make this work will come later...
* Fixed padding for Graph Editor visibility toggles
* Reverted many of the tweaks to Timeline UI for now (for the reasons outlined in earlier mail)
* NLA Editor now (mostly) uses the new channel-drawing API
Finally started refactoring the drawing code for animation editors so that all the drawing (of the channels/names list) is done using standardised code for all the editors.
A little specialised system for this has been made, and is now used for DopeSheet and Graph Editor drawing. This should make it easier to add new channel types with less effort.
Protect/mute toggles are now always clamped to the right-hand edge of the channels list. So you can now make the list wider to view longer names fully. However, since the mouse-handling hasn't been fixed yet, this won't always work.
TODOs:
* make mouse-click handling code use this system too, simplifying things to 1-2 lines of code there. Maybe this could be made to use UI widgets instead at a later date?
* NLA still needs to be ported to this system
* Cleaned up some parts of the code that were unused/could be done a bit nicer
* Added a new option for only showing the keyframes of the selected F-Curves in the Graph Editor, as another way of reducing the clutter.
* Insert Keyframes menu now only displays the 'active keyingset' entry when there are some KeyingSets.
* Moved the validation code for auto-blending/extend modes to NLA editor code, and included calls for this in many of the editing tools for NLA strips.
* Removed obsolete 'ID_IPO' entries from RNA-ID wrapping (these were commented out anyway).
I've gone through all the Animation Editor operators, making sure they send appropriate notifiers, and that these notifiers are handled.
* Added a separate category for animation-related notifiers, since the old-style ones attached to specific datatypes only was turning out to be not too feasible.
* For now, the focus has been on making sure that all Animation Editors update when there have been any potentially suitable changes at all. Later on, we can filter these more carefully to only take the ones we really need (for optimisation purposes)
Recoded the Transform code for NLA-Strips so that they can now be moved between strips and will not get truncated when they get moved into other non-moving strips.
Todos:
* The current code for moving strips between tracks is buggy (only goes up, and has a tendency to move up without being told to)
* Auto-snapping doesn't work yet...
* Made NLA Editing functions more aware of transitions.
- A transition can only be added between a pair of action-clips. Previous, two transitions could be added next to each other, which has undefined behaviour
- Deleting a strip with transition(s) on either side will remove the transitions too. Feedback welcome on this
- The 'type' setting for NLA-Strips is no longer editable. This was dangerous as it could result in transitions with undefined behaviour (though nothing would happen).
* Menus for adding F-Modifiers now only show relevant modifiers (i.e. 'Invalid' is not included in the list, and 'Cycles' doesn't need to be shown for NLA since we've got repeat)
* Function Generator and Noise F-Modifiers now have complete GUI's. A few settings were missed during the porting process.
* F-Modifier buttons now have their source-ID's included in the RNA Pointers used. This didn't get them animateable directly, but is a step closer.
* Most of the F-Modifiers have been ported to use the layout engine + RNA for drawing their buttons now. This plays much nicer with various button-layouts.
--> As a nice demo, try adding a Noise Modifier to a NLA-strip, and change the 'size' setting to about 2 to see some effects.
* Generator and Envelope modifiers haven't been ported yet since they're quite complex (requiring more time + energy), and as such, have been made to have some temporary error prints instead. Will check on this tomorrow.
* Finished/cleaned up the RNA-wrapping of most FModifiers.
TODO's (help requested... Brecht?):
Generator modifier's UI cannot be wrapped yet using the layout engine (though I might try using the old system only), as I'm having some trouble wrapping the coefficients array for this (see rna_fcurve.c - rna_def_fmodifier_generator())
* Using the Ctrl-Shift-M hotkey, F-Modifiers can be added to all the selected strips.
* F-Modifiers can also be added/tweaked from the NLA N-Key properties. The UI now uses the same code as for the graph editor ones.
The UI drawing here is currently messed up from the NLA side, since it seems combining normal layout stuff and old-style uiBlocks doesn't seem to work too well (BUT! the buttons are at least functional).
Next up, I'll need to recode the buttons panel for the Graph Editor so that all of the drawing can be migrated over to use the new layout engine.
