Goals of this refactor:
* Simplify creating virtual arrays.
* Simplify passing virtual arrays around.
* Simplify converting between typed and generic virtual arrays.
* Reduce memory allocations.
As a quick reminder, a virtual arrays is a data structure that behaves like an
array (i.e. it can be accessed using an index). However, it may not actually
be stored as array internally. The two most important implementations
of virtual arrays are those that correspond to an actual plain array and those
that have the same value for every index. However, many more
implementations exist for various reasons (interfacing with legacy attributes,
unified iterator over all points in multiple splines, ...).
With this refactor the core types (`VArray`, `GVArray`, `VMutableArray` and
`GVMutableArray`) can be used like "normal values". They typically live
on the stack. Before, they were usually inside a `std::unique_ptr`. This makes
passing them around much easier. Creation of new virtual arrays is also
much simpler now due to some constructors. Memory allocations are
reduced by making use of small object optimization inside the core types.
Previously, `VArray` was a class with virtual methods that had to be overridden
to change the behavior of a the virtual array. Now,`VArray` has a fixed size
and has no virtual methods. Instead it contains a `VArrayImpl` that is
similar to the old `VArray`. `VArrayImpl` should rarely ever be used directly,
unless a new virtual array implementation is added.
To support the small object optimization for many `VArrayImpl` classes,
a new `blender::Any` type is added. It is similar to `std::any` with two
additional features. It has an adjustable inline buffer size and alignment.
The inline buffer size of `std::any` can't be relied on and is usually too
small for our use case here. Furthermore, `blender::Any` can store
additional user-defined type information without increasing the
stack size.
Differential Revision: https://developer.blender.org/D12986
When proxy size lower than 100% is used, clip strips are rendered with
incorrect image size.
This is because if proxies aren't enabled in movieclip, it automatically
falls back on rendering original media. Sequencer doesn't have knowledge
about this and since 9c99292a16 it assumes that image is proxy,
because it explicitly requested this size.
Check movieclip flag to see if proxies are enabled.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D13080
When cut an stroke using the option Flat Caps, the falt was not done if the cut was done in the middle of the stroke.
Now the flat is applied to the segments created and also some cleanup of the code done.
Cause of this issue is that Custom Node Sockets info type was
initialized as SOCK_FLOAT when registering. Areas within the core that
would ignore custom socket types by checking its type would use the
socket as being a float type.
When custom node sockets have a property called default_value blender
tries to store it as an internal default value what failed in debug
builds.
This patch will set the socket type to SOCK_CUSTOM when registering a
custom socket type and allow, but skip storage of custom default values.
In this case the default values should already be stored as custom
properies.
Reviewed By: campbellbarton, JacquesLucke
Maniphest Tasks: T89260
Differential Revision: https://developer.blender.org/D13174
Allow the use of floating-point values for font point sizes, which
allows greater precision and flexibility for text output.
See D8960 for more information, details, and justification.
Differential Revision: https://developer.blender.org/D8960
Reviewed by Campbell Barton
Suppressing the splash was only done when passing in an argument from
the command line.
Remove G.file_loaded, as it is misleading, only set once on startup,
replace with G.relbase_valid which is used everywhere else to check
if the file path should be used.
Previously both `.` and `_` were used as separators when finding
a unique name for a socket. This removes the use of `.`, since `_`
was more common. It also does versioning for all of a file's node
trees to make sure that they all use the `_` convention.
Differential Revision: https://developer.blender.org/D13181
Blender 3.0 will only support single-frame Actions in the pose library.
The goal of this patch is to lay the groundwork for making it possible
for the Asset Browser to reject/hide "animation snippet" Action assets.
Determining whether an Action has one or more frames (i.e. whether it
has a single pose or animation) requires inspecting the Action itself,
and thus loading the data-block itself. This would make it impossible to
quickly determine from the asset browser.
To solve this, the Action is inspected before saving, and a
`"is_single_frame"` boolean (well, 0/1 integer) IDProperty is added.
Reviewed by: Severin
Differential Revision: https://developer.blender.org/D13202
* Rename the 'copy' functions to make it clear they belong to the same
'group' and are to be used together.
* Fix `flag` parameter of `BKE_copybuffer_paste` being a short instead
of an int.
* Improve documentation.
The mirror modifiers merge option caused unnecessary re-ordering
to the vertex array with original vertices merging into their copies.
While this wasn't an error, it meant creating a 1:1 mapping from input
vertices to their final output wasn't reliable (when looping over
vertices first to last) as is done in
BKE_editmesh_vert_coords_when_deformed.
As merging in either direction is supported, keep the source meshes
vertices in-order since it allows the vertex coordinates to be extracted.
When Constructing bezier splines from dna, the positions of the
left/right handles were set directly in the internal vectors, by
requesting a reference to them. The problem is that
BezierSpline::handle_positions_left() calls ensure_auto_handles()
before returning the reference. That function does some calculations on
uninitialized memory if the positions array is not yet filled.
Differential Revision: https://developer.blender.org/D13107
When clearing geometry the runtime mutexes of a mesh were freed. This
resulted in crashes afterwards. The clear geometry is an RNA function so
would only effect when using from scripts.
This patch separates init/freeing of the mutexes from other code so they
can be used when needed.
Reviewed By: mont29
Maniphest Tasks: T91518
Differential Revision: https://developer.blender.org/D13142
Currently we have a fixed instance recursion limit. While we want to lift this
limitation at some point, that is out of scope for a bug fix. For now just print
a warning to make it easier to detect the issue.
Differential Revision: https://developer.blender.org/D13162
Catalog simple names are supposed to fit into the DNA field `char
AssetMetaData::catalog_simple_name[64]`, and thus should be shortened
appropriately. This was already happening, but is now also covered by a
test.
No functional changes.
Interpolation vertex data on loose edges was writing into already
allocated data.
Resolve this by skipping vertex end-points for custom-data interpolation
which has already been copied from the source mesh.
Reviewed By: sergey
Ref D13082
Currently, when creating a new node tree ID, its `typeinfo` is set to
`NodeTreeTypeUndefined`, but its `type` enum value is left to `0`,
aka `NTREE_SHADER`.
This patch adds a new `NTREE_UNDIFINED` value, and use it for
`NodeTreeTypeUndefined` types of node trees.
NOTE: While it is not clear whether that actually fixes issues currently,
quite a bit of code still relies on the value of `type`, so think it
makes sense to sanitize this.
NOTE: Would have been ideal to reserve `0` value to undefined type,
but at this point this is not possible anymore, so chose to use `-2` instead.
Reviewed By: JacquesLucke
Differential Revision: https://developer.blender.org/D13123
Previously, to get the declaration of a socket, one had to go
through `node->declaration`. Now this indirection is not necessary
anymore. This makes it easier to add more per-socket information
into the declaration and accessing it in various places.
Currently, this system is used by socket descriptions and node warnings
for unsupported geometry component types.
Reference struct members by name instead relying on their order.
This also simplifies moving back to named members when all compilers
we use support them.
Internal struct ObTfmBack had out of sync doc-strings
for members duplicated from Object.
Remove the doc-strings as there is this is just temporary storage.
This commit renames mesh.c to mesh.cc and makes
it compile in C++. Can be useful in the future to be able
to use C++ functionality in existing and new functions.
Differential Revision: https://developer.blender.org/D13134
Evaluated meshes from curves are presented to render engines as
separate instance objects now, just like evaluated meshes from other
object types like point clouds and volumes. For that reason, cycles
should not consider curve objects as geometry (previously it did,
meaning it retrieved a second mesh from the curve object as well
as the temporary evaluated mesh geometry).
Further, avoid adding a curve object's evaluated mesh as data_eval,
since that is special behavior for meshes that is arbitrary. Adding an
evaluated mesh there but not an evalauted pointcloud is arbitrary,
for example. Retrieve the evaluated mesh in from the geometry set
in BKE_object_get_evaluated_mesh now, to support that change.
This gets us closer to a place where all of an object's evaluated data
is stored in geometry_set_eval, and we just have helper functions
to access specific geometry components.
Differential Revision: https://developer.blender.org/D13118
Followup to e65230f0c0.
Pablo and I decided it's fine to reset themes again when saved with the
recent 3.1 builds.
This needed to be done a bit careful, since a normal version patch
resetting the theme would've reset the theme for anybody opening
preferences of a 3.0 build (even the final release build) in a 3.1
build. So make sure the theme is at least from a 3.1 build (but not
newer then this commit of course).
Evaluated meshes from curves are presented to render engines as
separate instance objects now, just like evaluated meshes from other
object types like point clouds and volumes. For that reason, cycles
should not consider curve objects as geometry (previously it did,
meaning it retrieved a second mesh from the curve object as well
as the temporary evaluated mesh geometry).
Further, avoid adding a curve object's evaluated mesh as data_eval,
since that is special behavior for meshes that is arbitrary. Adding an
evaluated mesh there but not an evalauted pointcloud is arbitrary,
for example. Retrieve the evaluated mesh in from the geometry set
in BKE_object_get_evaluated_mesh now, to support that change.
This gets us closer to a place where all of an object's evaluated data
is stored in geometry_set_eval, and we just have helper functions
to access specific geometry components.
Differential Revision: https://developer.blender.org/D13118
During animation playback, data-blocks are reallocated, so storing
pointers to the resulting data is not okay. Instead, the data should
be retrieved from the context. This works when the applied search
item is the "dummy" item added for non-matches. However, it still
crashes for every other item, because the memory is owned by the
modifier value log, which has been freed by the time the exec function
runs.
The next part of the solution is to allow uiSearchItems
to own memory for the search items.
The point domain attributes (stored on splines) are sorted so they
have a consistent order on all splines after the join. However, spline
domain attributes were included in the new order, which didn't work
because the length of the attribute lists didn't match. The simple fix
is to only include point domain attributes in the new order vector.
In some cases when geometry is created in Geometry Nodes
the viewport stats will show 0 because runtime data is not filled.
This patch sets the runtime data on instances.
Differential Revision: https://developer.blender.org/D12738
For single point splines that weren't at the origin, the results were
incorrect. Now take into account the tilt, radius, etc. just like the
general case.
A group node could reference an undefined group when the group
was linked and its source file was not found on load. The field
inferencing code did not handle that case before.
With this change, the file provided in T92799 loads successfully.
Many modifiers and other places use `CustomData_copy_data` to copy data
between different meshes. This function assumes that assumes that the
source and destination `CustomData` objects are "compatible" in some way.
Usually modifiers use `CustomData_copy` to create a compatible new
`CustomData` on the new mesh. The issue was that the optimization I added
for anonymous attributes broke this compatibility. It avoided copying some
attributes when they are no longer used.
This lead to attributes being copied incorrectly.
D13083 contains ideas for how this could be fixed more generally.
For now I just removed the optimization.
Differential Revision: https://developer.blender.org/D13083
By default, when syncing materials slots between object and its obdata,
the amount of slots in obdata is the reference.
Missing linked obdata is replaced by an empty placeholder that has no
material. In that specific case, if we have a valid object ID, we want
to update the (placeholder) obdata's material count from the object one,
and not the other way around.
The version patch for 0cf9794c7e was checking and setting a data name
using the macros for translation. These would access the current
preferences which can mismatch the ones currently being version patched.
See discussion in 0cf9794c7e for details.
Don't handle translation in this version patch, which is more of a
"nice-to-have" version patch, no functionality depends on it.
This properly checks the order of point domain attributes on each
spline, and avoids the map, which makes the code easier to understand.
The assert is also added to realizing instances and the join node.
Differential Revision: https://developer.blender.org/D13071
Currently the curve to mesh node relies on the order of attributes being
the same on every spline. This is a worthwhile assumption, because it
will allow removing the attribute name storage duplication on every
spline in the future.
However, the join geometry node broke this order, since it just created
new attributes without any regard to the order. To fix this, I added a
"reorder" step after all the merged attributes have been created, the
types have been joined, etc. It should be possible to change this code
so that the attributes are added with the right order in the first
place, but I would like to do that after refactoring spline attribute
storage, and not for 3.0.
Differential Revision: https://developer.blender.org/D13074
Previously, `ImageTile->ok` and `ImageUser->ok` were used to indicate
whether an image failed to load. There were three possible values
which (probably) had the following meanings:
* `0`: There was an error while loading the image. Don't try to load again.
* `1`: Default value. Try to load the image.
* `2`: The image was loaded successfully.
This image-wide flag did not make sense unfortunately, because loading
may work for some frames of an image sequence but not for others.
Remember than an image data block can also contain a movie.
The purpose of the `->ok` flag was to serve as an optimization to avoid
trying to load a file over and over again when there is an error (e.g. the
file does not exist or is invalid). To get the optimization back, the patch
is changing `MovieCache` so that it can also cache failed load attempts.
As a consequence, `ibuf` is allowed to be `NULL` in a few more places.
I added the appropriate null checks.
This also solves issues when image sequences are used with the
Image Texture node in Geometry nodes (also see D12827).
Differential Revision: https://developer.blender.org/D12957
rBbe3e09ecec5372f switched the order for vertices referenced by the
start cap's corners, but it failed to account for the offset necessary
for edge indices, since the order changed.
rB43bc494892c3 switched `BKE_libblock_relink_to_newid` to use new ID
remapping and libquery code.
However, that new code does protect by default against remapping an
objects's data pointer when that object is in Edit mode, since this is
not a behavior that generic BKE code can handle (due to required editing
data for most obdata types when in edit mode).
So specific code that does create new IDs and need remapping in Edit
mode has to pass specific exception flags to remaping code.
This commit adds those remapping flags to `BKE_libblock_relink_to_newid`
and add said exception flag to the remapping call from
`ED_object_add_duplicate` when the object is in edit mode.
In theory we should never allow remapping of Objects' obdata ID pointer
when the object is in Edit mode. But there are some cases were this is
needed, so adding yet another exception option to remapping flags.
Preliminary change to fix T92629.
Allows to avoid a global lock being held while reading files from disk,
solving performance issues when Cycles needs to read a lot of packed
images.
Simple test file F11597666
Differential Revision: https://developer.blender.org/D13032
Accessing the default directory in the file selector
would crash if HOME was undefined.
Add BKE_appdir_folder_default_or_root which never returns NULL.
Initial defaults for userdef->fontdir for Mac and Linux.
See D12802 for more details.
Differential Revision: https://developer.blender.org/D12802
Reviewed by Campbell Barton
Those were used in a very few places to detect whether iteration should
be stopped or not, but one can use `BKE_lib_query_foreachid_iter_stop`
now for that.
Also fix early break handling in embedded IDs processing.
Fix T90922: Fix return policy inconsistency in `scene_foreach_id`.
In case `library_foreach_ID_link` would return early in recursive
process, it would not properly free its utils data.
Also add proper iteration break in case some sub-calls requested it.
Finally, make this function return a boolean to know whether iteration
should be stopped or not (will be used in future commit to fix this
handling in embedded IDs case).
Part of T90922: Fix return policy inconsistency in `scene_foreach_id`.
The new `BKE_LIB_FOREACHID_PROCESS_FUNCTION_CALL` execute the given
statement and then check status of `LibraryForeachIDData` data, and
return in case stop of iteration is requested.
This is very similar to the other `BKE_LIB_FOREACHID_PROCESS_` existing
macros, and allows us to properly break iteration when a sub-function
has requested it.
Part of T90922: Fix return policy inconsistency in `scene_foreach_id`.
Add a function to check if iteration over ID usages should stop (using
internal `IDWALK_STOP` status flag).
Use it in `BKE_LIB_FOREACHID_PROCESS_` macros, and in
`window_manager_foreach_id` to handle properly the active workspace case
(previous code could skip the call to `BKE_workspace_active_set` in case
iteration over ID usages was stopped by callback on that specific ID
usage).
Part of T90922: Fix return policy inconsistency in `scene_foreach_id`.
This is a better and more general fix for T92511 and T92508 than
the ones that I committed before.
Previously, we tagged caches dirty when first accessing attributes.
This led to incorrect caches when under some circumstances. Now
cache invalidation is part of `OutputAttribute.save()`.
A nice side benefit of this change is that it may make things more
efficient in some cases, because we don't invalidate caches when
they don't have to be invalidated.
Differential Revision: https://developer.blender.org/D13009
In this case, the uniform index sampling loop would fail to assign any
data to the samples, so fill the rest with the largest value possible,
corresponding to the end of the spline. Animation Nodes has the same
fix for this case.
Remapping code could call collection resync code while processing
remapping, which is a good way to crash by accessing no-more-valid
pointers.
Similar issue as with liboverrides resync, fixed the same way by
preventing any collection resync until whole remapping has been done.
This was probably not an issue in practice in current code, since this
is only used by append code currently, which should not affect
layers/collections in current scene yet.
{F11548100, size=full}
To celebrate the beginning of a new series, it feels like the right time to
give the theme a fresh look while improving on what already works.
The aim of this refresh is to keep a familiar look but with polishing touches
here and there. Like new paint on the walls of your well known house.
The theme for Blender 2.8 was well received but presented a few flaws.
* Transparency on menus and tooltips reduce readability
* Mismatch on certain colors, especially outlines of connected widgets
* Active/open menus highlight was not prominent enough
* Header background mismatch in some editors
At the same time we can make use of new features in 3.0:
* Make panels look like panels again (like in v2.3!)
* Make use of roundness in more widgets
* Since nodes are no longer hard-coded to be transparent, tweak opacity and saturation
* Tweak colors for the new dot grid
This update does not include:
* Meshes in edit mode to match greenish object-data color. This needs tweaks in the code to improve contrast.
* A copy of the Blender 2.8x legacy theme. This could be added to the community themes (shouldn't cost much maintenance, I hope)
There will be certainly small tweaks to do here and there, I've been working using this theme
for months but there can be areas that are missing update. The overall style is presented here.
This commit bumps the file subversion.
Reviewed By: #user_interface, Severin
Differential Revision: https://developer.blender.org/D13008
This reverts commit 03013d19d1.
This commit broke the windows build pretty badly and I don't
feel confident landing the fix for this without review.
Will post a possible fix in D12969 and we'll take it from there.
These nodes allow accessing and changing the stable/random ID used
for motion blur with instances and stable randomness.
Since rB40c3b8836b7a, the stable ID is a built-in attribute, so to be
consistent and allow changing it in the node tree like other built-in
attributes, it has get and set nodes.
Feedback was that "Default" is a bit of a weird name, so switching it to "User
Library". Added versioning code which won't be entirely bullet proof (e.g. will
also rename libraries named "Default" by the user), but it doesn't have to be.
Addresses T90298.
Previously, every node had to create warnings for unsupported input
geometry manually. Now this is automated. Nodes just have to specify
the geometry types they support in the node declaration.
Differential Revision: https://developer.blender.org/D12899
Instance IDs serve no purpose for rendering when they aren't stable from
one frame to the next, and if the index is used in the end anyway, there
is no point in storing a vector of IDs and copying it around.
This commit exposes the `id` attribute on the instances component,
makes it optional-- only generated by default with the distribute points
on faces node.
Since the string to curves node only added the index as each instance's
ID, I removed it. This means that it would be necessary to add the ID
data manually if the initial index actually helps (when deleting only
certain characters, for example).
Differential Revision: https://developer.blender.org/D12980
This adds generic attribute rendering support for meshes for Eevee and
Workbench. Each attribute is stored inside of the `MeshBufferList` as a
separate VBO, with a maximum of `GPU_MAX_ATTR` VBOs for consistency with
the GPU shader compilation code.
Since `DRW_MeshCDMask` is not general enough, attribute requests are
stored in new `DRW_AttributeRequest` structures inside of a convenient
`DRW_MeshAttributes` structure. The latter is used in a similar manner
as `DRW_MeshCDMask`, with the `MeshBatchCache` keeping track of needed,
used, and used-over-time attributes. Again, `GPU_MAX_ATTR` is used in
`DRW_MeshAttributes` to prevent too many attributes being used.
To ensure thread-safety when updating the used attributes list, a mutex
is added to the Mesh runtime. This mutex will also be used in the future
for other things when other part of the rendre pre-processing are multi-threaded.
`GPU_BATCH_VBO_MAX_LEN` was increased to 16 in order to accommodate for
this design.
Since `CD_PROP_COLOR` are a valid attribute type, sculpt vertex colors
are now handled using this system to avoid to complicate things. In the
future regular vertex colors will also use this. From this change, bit
operations for DRW_MeshCDMask are now using uint32_t (to match the
representation now used by the compiler).
Due to the difference in behavior for implicit type conversion for scalar types
between OpenGL and what users expect (a scalar `s` is converted to
`vec4(s, 0, 0, 1)` by OpenGL, vs. `vec4(s, s, s, 1)` in Blender's various node graphs) ,
all scalar types are using a float3 internally for now, which increases memory usage.
This will be resolved during or after the EEVEE rewrite as properly handling
this involves much deeper changes.
Ref T85075
Reviewed By: fclem
Maniphest Tasks: T85075
Differential Revision: https://developer.blender.org/D12969
In future use cases, a volume can contain many grids that represent the
density information. In this case, it's better if the volume to mesh node
creates a mesh based on all of the grids in the volume.
This is also a benefit to share-ability, since one doesn't have to
specify the grid name in the node. Instead, in the future we can have
a way to split particular grids into separate volumes, if only one
grid should be considered.
The code changes are relatively simple:
- Move the old volume to mesh node to the legacy folder.
- Run the volume to mesh node on all instance geometry, like elsewhere.
- Make the blenkernel's volume to mesh API a bit more specific.
Differential Revision: https://developer.blender.org/D12997
This patch makes the background grid of the node editor a grid of dots
instead of lines. This makes the background look a bit more subtle and
reduces visual complexity. The dots are meant to provide a reference
when panning and zooming. Based on the design of @pablovazquez, and
a patch originally authored by @fabian_schempp.
The "Grid Levels" controls how many levels of dots are drawn. As the
editor zooms in, the higher levels of dots fade in, making them closer
together visually. The zoom factor at which each grid starts and ends
fading in is controllable in the code, and could be tweaked further
in the future. The new default value is 7, out of a range from 0 to 9.
Differential Revision: https://developer.blender.org/D10345
The drag and drop feature of objects in 3D View has been modified to include:
- Snap the object being dragged.
- Visual feedback through a box and the placement tool grid.
Maniphest Tasks: T90198
Differential Revision: https://developer.blender.org/D12912
This adds an option to fill the ends of the generated mesh for
each spline combination with an N-gon. The resulting mesh is
manifold, so it can be used for operations like Boolean.
Differential Revision: https://developer.blender.org/D12982
Add calls to `BKE_callback_global_init()` and `BKE_callback_global_finalize()`
to ensure unit tests mimick Blender (and don't trip the assertions added
in rBbeea601e7253).
No functional changes to Blender.
Object types like empties, cameras or lamps will just end up as empty preview
images. We can think about ways to visualize them still, but meanwhile, don't
create such an empty preview.
Differential Revision: https://developer.blender.org/D10334
Reviewed by: Bastien Montagne, Sybren Stüvel
This adds a new image texture node for geometry nodes. It does not
reuse the same node that is used in shading, because we want to be
able to expose the image and frame as sockets.
There is a known update issue when a movie or image sequence is
used. That will be fixed separately (also see D12957).
Currently, the image socket is just a pointer to an Image ID data block.
This can contain single images but also movies and image sequences.
In the future, the definition of an image socket can be expanded to
include images that are generated from scratch in the node tree.
For more details read the discussion in D12827.
Some of the code is a direct port from cycles and should be cleaned
up a bit in the future. For example `image_cubic_texture_lookup`.
For still images, the frame input is ignored. Otherwise, the frame
has to be in a valid range for the node to work. In the future we
may add e.g. automatic looping functionality.
Differential Revision: https://developer.blender.org/D12827
Asset Catalog Paths should only contain forward slashes as separators, but
now the UI is more resilient to people using blackslashes instead.
Manifest Task: T90553
Put related lines in a block of their own, such that each block doesn't
have access to the variables of the previous blocks.
This makes it easier to correctly copy-paste some tests, as the compiler
forces you to update the code afterwards.
Add static boolean to track whether the callbacks system has been
initialised. This makes it possible to make the `BKE_callback_remove()`
function more noisy in case of programming errors, and avoids accessing
`funcstore->alloc` when `funcstore` was potentially already freed.
Thanks @campbellbarton for pointing this out.
This patch renames the node "String Substring" to "Slice String"
to conform to the "verb first" naming convention.
Default length is also changed to 10 to make it easier for users
to understand what the node does.
Reviewed By: HooglyBoogly
Differential Revision: https://developer.blender.org/D12931
This node takes a geometry set with instances as input and outputs
points located on the origins of the top level of instances in the
geometry set (not nested instances). It also has position and radius
inputs to allow overriding the default, and a selection input to only
generate points for some instances.
The use case for this node is a method to use geometry proximity on
instance origins, but in a more generic way that is flexible and useful
in other situations.
Differential Revision: https://developer.blender.org/D12893
Because `segment_is_vector` didn't handle the combined cyclic and
single control point case, it returned false, that the "segment" should
have the resolution evaluated point count. To avoid checking the size in
every call, add an assert for the size and check it elsewhere.
Show assets that have an unknown catalog ID assigned in the "Unassigned"
catalog.
Another catalog named "Orphans" was considered as well, but that would
clash with the usual handling of Blender (discarding orphan data on
save) and thus that idea was discarded.
Manifest Task: T91949
Keep track of unsaved asset catalog changes, in a more granular way than
just one boolean per asset library. Individual catalogs can now be
marked with a flag `has_unsaved_changes`. This is taken into account
when reloading data from the catalog definition file (CDF):
- New catalog in CDF: gets loaded
- Already-known catalog in CDF:
- local unsaved changes: on-disk catalog is ignored
- otherwise: on-disk catalog replaces in-memory one
- Already-known catalog that does not exist in CDF:
- local unsaved changes: catalog is kept around
- otherwise: catalog is deleted.
Because this saving-is-also-loading behaviour, the "has unsaved changes"
flags are all stored in the undo buffer; undoing after saving will not
change the CDF, but at least it'll undo the loading from disk, and it'll
re-mark any changes as "not saved".
Reviewed By: Severin
Differential Revision: https://developer.blender.org/D12967
These nodes just output a single value of their respective types,
making it possible to control multiple inputs with the same value.
Differential Revision: https://developer.blender.org/D12932
Camera, lattice and speaker object types were missing there own proper
`USER_DUP_` flags, leading to not properly handling duplication of their
object data.
NOTE: We could probably simply opions here, by using categories (like
'GEOMETRY', 'SHADING', etc.) instead of exact object types. But this is
beyond bugfix scope.
This commit makes the bounding box node work on each unique geometry
(including instances) individually instead of making one large bounding
box for everything. This makes the node much faster, and is often the
desired result anyway. For the old behavior, a realize instances node
can be used in front of this node (versioning adds it automatically).
The min and max outputs now only output the values from the realized
geometry.
Differential Revision: https://developer.blender.org/D12951
When renaming an asset catalog, also update its simple name.
Catalogs will most likely be created from within Blender, so via the
catalog tree in the asset browser. Here catalogs are always named
"Catalog" until the user renames them, which was reflected in all simple
names being "Catalog".
When there are multiple catalogs with the same path (so different UUIDs
all mapped to the same catalog path), treat the first-loaded one as the
main catalog for that path, and the rest as aliases.
This ensures that the UUID of a catalog (as chosen in the tree UI and thus
interacted with by users) is stable, regardless of whether by some coincidence
later another catalog with the same UUID is created.
Rename `AssetCatalogPathCmp` to `AssetCatalogLessThan`:
- it compares more than paths (so no more `Path` in the name), and
- performs a less-than operation (so no more `Cmp` in the name).
Also restructure its code to make an extra upcoming comparison easier to
add.
No functional changes.
This is mainly for doversion code, when it needs to create new IDs those
should be considered as part of the same library as the current Main's
one.
No practical changes are expected here, this is more of a general
consistency fix, and a pre-requisite for {T92333}.
Since the introduction in rBfc5bf09fd88c, `BKE_pose_minmax` was not
taking these custom transforms into account (making "View Selected"
ignoring these as well and focusing on the bone instead).
Now consider these transforms in `BKE_pose_minmax`.
Maniphest Tasks: T92169
Differential Revision: https://developer.blender.org/D12942
When appending an asset from the asset browser, its asset data needs to
be cleared.
However, linking an asset (or regular append from the file browser)
should not clear such data. In linking case, it would be there again
after a blend file reload anyway.
So this commit introduces a new `BLO_LIBLINK_APPEND_ASSET_DATA_CLEAR`
option.
NOTE: in case the appended ID needs to be copied from its linked data
(instead of making the later directly local), asset data is lost anyway
since it is never copied with the ID currently.
Ref. {T91749} and D11768.
In order to address feedback that the "Stable ID" was not easy enough
to use, remove the "Stable ID" output from the distribution node and
the input from the instance on points node. Instead, the nodes write
or read a builtin named attribute called `id`. In the future we may
add more attributes like `edge_id` and `face_id`.
The downside is that more behavior is invisible, which is les
expected now that most attributes are passed around with node links.
This behavior will have to be explained in the manual.
The random value node's "ID" input that had an implicit index input
is converted to a special implicit input that uses the `id` attribute
if possible, but otherwise defaults to the index. There is no way to
tell in the UI which it uses, except by knowing that rule and checking
in the spreadsheet for the id attribute.
Because it isn't always possible to create stable randomness, this
attribute does not always exist, and it will be possible to remove it
when we have the attribute remove node back, to improve performance.
Differential Revision: https://developer.blender.org/D12903
This commit adds an updated version of the curve to points that
supports fields. Only the position and radius are transferred
by default now, which should improve performance. The other outputs
like tangent and rotation are outputted with anonymous attributes.
I took the opportunity to change a few other small things:
- Name geometry sockets "Curve" and "Points" like other nodes.
- Remove the radius multiple of 0.1, which was confusing.
Thanks to @Johnny Matthews (guitargeek) for an initial patch.
Differential Revision: https://developer.blender.org/D12887
Yet another try at that hairy issue... See comment in commit for
details, essentially this extend the workaround introduced in Objects'
`lib_override_apply_post` callback to try to also properly 're-use'
`OUTDATED` and `BAKED` flags from old source liboverride into new
destination one.
When the input vector gets close to -Y, y and theta becomes totally
unreliable. It is thus necessary to compute the result in a different
way based on x and z. The code already had a special case, but:
- The threshold for using the special case was way too low.
- The special case was not precise enough to extend the threshold.
- The special case math had a sign error, resulting in a jump.
This adds tests for the computation precision and fixes the issues
by adjusting the threshold, and replacing the special case with one
based on a quadratic Taylor expansion of sqrt instead of linear.
Replacing the special case fixes the bug and results in a compatibility
break, requiring versioning for the roll of affected bones.
Differential Revision: https://developer.blender.org/D9551
Separate the huge test into huge logical parts and add more cases
to check. Also add a utility to check that the matrix is orthogonal,
with arbitrary epsilon values and calculations in double.
A couple of tests deliberately fail, to be fixed in following commits.
Ref D9551
This commit adds a node that can be used to find and replace strings
inside of the input string. One initial use case is to have an easier
way to add line breaks to strings to the string to curves node.
Differential Revision: https://developer.blender.org/D12721
No longer save asset catalogs on blendfile save. Instead:
- extend the confirmation prompt for unsaved changes to show unsaved
catalogs.
- In the confirmation prompt, make catalog saving explicit & optional,
just like we do it for external images. {F10881736}
- In the Asset Browser catalog tree, show an operator icon to save the
catalogs to disk. It's grayed out if there are no changes to save, or
if the .blend wasn't saved yet (required to know where to save the
catalog definitions to). {F10881743}
Much of the work was done by @Severin and reviewed by me, then we
swapped roles.
Reviewed By: Severin
Differential Revision: https://developer.blender.org/D12796
Allow creating a new asset catalog in a yet-to-be-saved blend file. The
problem was caused by `AssetLibrary` not having an `AssetCatalogService`
right after creation; only after loading data from disk was this
instance created. It's now always there.
This patch includes an updated version of the raycast node that uses
fields instead of attributes for inputs instead of outputs. This makes
the node's UI much clearer. It should be faster too, since the
evaluation system for fields provides multi-threading.
The source position replaces the input geometry (since this node is
evaluated in the context of a geometry like the other field nodes).
Thanks to @guitargeek for an initial version of this patch.
Differential Revision: https://developer.blender.org/D12638
Make `AssetLibraryService::get_asset_library_on_disk(path)` insensitive
to trailing slashes; i.e. `get_asset_library_on_disk("/path")` and
`get_asset_library_on_disk("/path/¨)` will now return the same
`AssetLibrary*`.
`BKE_callback_remove` now checks whether the callback actually is known,
before trying to remove it.
`BKE_blender_atexit()` runs after `BKE_callback_global_finalize()`. When
an at-exit callback tried to unregister its BKE callbacks, these would
already be unregistered, causing a crash of Blender when exiting,
This commit removes the implicit conversion from points to a mesh
that used to happen before the next modifier. It also removes the
implicit realizing of instances that happened before another modifier.
Now we have specific nodes for both of these operations, the
implicit conversions make less sense, and implicit instance
realizing has already been removed in other nodes.
This adds another geometry nodes modifier before modifiers that would
have realized instances implicitly before. Currently adding another
data-block during versioning after linking means that an assert needs
to be changed. That should be made unnecessary by T92333.
Differential Revision: https://developer.blender.org/D12722
Some compositing applications do not support spaces and dots in layer names, and change
these to other symbols on import. This causes various compatibility issues, including
with Cryptomatte metadata. While technically those could be considered bugs in the
Cryptomatte implementation of other software, where they are not properly accounting for
that layer renaming, it's not ideal.
The OpenEXR channel naming convention is "layer.pass.channel". We get away with dots in
the layer name since we parse this from right to left, but it's a weak assumption.
Now we don't forbid using spaces or dots, and existing files are unchanged. But at
least by default names will be compatible, and hopefully other software catches up in
time to support more flexible layer names.
Ref T68924
OpenEXR uses "." to separate layers/passes/channels, so using AOV.001 is
a problem. Other applications will not be able to parse it correctly.
Default to AOV_001 instead, and don't allow using dots in AOV names.
Fixes T89991
Ref T73266
Ref D12871