Blender Git Loki

Fix: GP array modifier's "count" property should have a lower "sane" limit

Previously, the upper bound for dragging in the UI was effectively unlimited.
This has the unfortunate consequence of causing hard resets (out of bounds/memory)
as I found out last night while testing this.

GP Modifiers - Merge Array and Duplication modifiers

Array and Duplication modifiers basically implemented the same functionality,
except that:
1) Array was slightly more powerful (i.e. it could make a 3D grid instead of
just putting everything in a single line), and
2) Array created new objects when "Applying", while Duplication added to the
current datablock

This commit basically merges the functionality of both into the Array modifier.
I ended up recoding this modifier to hopefully work a lot faster.
* The resulting modifier more closely resembles the behaviour of the normal
Blender Array modifier.
* Performance should be better now with over 25 objects (previous array modifier
implementation lagged badly with more than 25-30 instances). The previous approach
of manually creating objects didn't work too well.

TODO:
* The functionality to disable the instance creation when the simplify options
are enabled is currently lost. Still need to figure out how to restore it
* The stubs/shell for the old Duplication modifier are still around. To be removed!

Port Subdiv GP Modifier

Restored manual gpencil_batch_cache_dirty() call when adding modifiers

There are too many problems leaving it out, even if in theory the depsgraph
calls should be taking care of it.

Temporary (?) fix so that Hero project files get version patched correctly

Porting more GP Modifiers

* Opacity modifier
* Color modifier
* Thickness modifier
* Tint modifier

Cleanup: Remove unused function

Fix: GP modifiers not updating when properties changed

Added back an explicit call to BKE_gpencil_batch_cache_dirty()
in the rna update callback for GPencil modifier properties.

However, instead of polluting the general rna_Modifier_update(),
this time we do this in a special wrapper/update callback for
dedicated GP modifier usage.

I'm unsure why the depsgraph isn't properly updating this case
- there's probably a missing relation between modifier properties
and GP ubereval (which calls the tagging), but where?!

GP Modifier: Set up stubs for geometry modifiers too

GP Modifiers: Fully ported Noise modifier over to new system

Fix: "deformStrokes(..., bGPDframe *)" callback should really have been "deformStroke(..., bGPDstroke *)" instead

Cleanup - Various cleanups for GP modifier files

* Remove redundant checks
* Use M_PI_2 instead of hardcoding a magic number
* Fix license text year info
* Properly handle new palette creation in Tint modifier

Cleanup: De-duplicate mode checking macros for GPencil

Cleanup - Move helper functions for GP modifiers into the modifiers module

Cleanup - Use existing vertex groups function instead of defining out own

Fix compiler warning about const vs non-const

Fix "Apply Modifier" operator to use the new callbacks

GP Modifiers - Change all applyModifier callbacks to bakeModifierGP callbacks

ModifierTypeInfo - Add 3 dedicated callbacks to support GP modifier evaluation and operations

* deformStrokes - Handles the "stroke" modifiers currently evaluated in
BKE_gpencil_stroke_modifiers() - e.g. Thickness, Noise

* generateStrokes - Handles the "geometry" modifiers currently evaluated
in BKE_gpencil_geometry_modifiers() - e.g. Array

* bakeModifierGP - Dedicated callback for hosting the code currently
wrongly-placed in applyModifier. This gets called
when pressing the "Apply" button on a GP modifier.

Compile fix: Missed a reference in previous commit

TODO: Calls to BKE_gpencil_layer_delframe() need to trigger a refresh