This is sometimes annoying now, because sometimes one wants to bake stuff
without a geometry. Most notably, volume grids. It's still very easy to bake
geometries though. Link-drag-search from a geometry socket still adds the
geometry bake socket automatically.
Pull Request: https://projects.blender.org/blender/blender/pulls/147519
Add a new constraint called "Geometry Attribute", which directly
samples vector, quaternion, or 4x4 matrix attributes from geometry and
applies these to an object's or bone's transform.
This can be used to transfer data generated by geometry nodes to the
object or bone level. By default the constraint will sample a vector
on the vertex domain. The default attribute is `position` for
simplicity, but the attribute value does not have to have anything to
do with neither the transform of the geometry object nor the geometry
itself.
Pull Request: https://projects.blender.org/blender/blender/pulls/136477
Merges the existing `bpy.context` logging with the new `context` logging
added in [0]. In most cases `bpy.context` & `context` log the same thing.
Where the new `context` logging produces more accurate results.
Ref !146407
[0]: 439fe8a1a0
Support propagation of attributes from the original mesh edges when
generating mesh edges. This implies both deduplication of original edges
and creation of new edges needed for faces. While original edges are not
propagated, attributes are still propagated if an edge was part of any
face. To make original edges distinct we simply choose the first one to
be a value source.
One important part of this patch is definition of invariant of the
mesh on input:
1. All edges must be valid (their vertices must be correct). The
algorithm only deal with duplicate and implicit edges.
2. There is no way to provide info about corner edges attribute (whether
it'svalid or not). If all input edges are valid and there are no new
edges then we could skip re-generation of already valid corner edge
indices. But for now we always do.
The new implementation depends on input mesh invariant so this drops an
old patch to fix hard crash related with duplicate edge vertices. New
code will crash independently to such patch. Hopefully there must be
no problems after #138633.
The changes don't affect performance much. If anything, in some cases
the edge calculation can become a few percent faster. See the PR for
more performance testing details.
Pull Request: https://projects.blender.org/blender/blender/pulls/132492
Currently when a strip has a transform that does not fill the whole
render area, first the image of the strip is transformed, and then
any modifiers are applied on that. This is mostly in the new
Compositor modifier, where procedural textures, gradients, image
coordinates "stick to the screen" instead of following the transformed
strip.
This changes the behavior so that first the modifiers are applied
to the strip image, and then the strip is transformed. This is
potentially a visually breaking change:
- This can alter visual look of existing strip, especially if they are
scaled. Previous behavior was first scale filtering, then modifier;
now it is first modifier, then scale filtering.
- Most obvious change is Compositor modifier (which is new in 5.0).
- Compositor modifier can actually expand the input image (e.g. Blur
node with "expand bounds" option set), and that works.
- Note that Masks continue to be applied in global/screen space. There
can be small look differences with rotated/scaled strips that use
masks, due to Mask application now needing to do filtered mask image
lookups.
- If anyone needs previous behavior (modifier is applied on the
"whole screen"), they can put transformed strip into a meta strip,
and apply the modifier on the meta strip itself.
Compositor modifier examples with images in the PR.
Pull Request: https://projects.blender.org/blender/blender/pulls/146181
Adds a new operator that jumps time by a given number of frames or seconds, forward or backward.
Surprisingly, it was lacking in Blender, and prompted many users (including me) to create extensions.
This PR adds two properties: `time_jump_unit` for choosing whether to jump by frames or seconds
and `time_jump_delta` property that defines by how many frames or seconds the operator should jump,
as well as an actual operator that changes the current frame (`screen.time_jump`).
`time_jump_delta` is a float that gives users the ability to jump by half a second, for example, or by subframes.
Default is set to 1 second, which translates to as many frames jump as frame rate / frame base.
The operator is intentionally not bound by frame range, and can go in negative frames as well.
This is very important because it's extremely common to set frame range to the current workload,
but wish to see animation beyond that.
Operators are added in the new footer for animation editors alongside with pop-up menu where
properties can be changed.
Shortcuts are also added: Ctrl+Left/Right Arrow, which was surprisingly free in Blender.
Now timeline controls are:
- **Right Arrow**: Next Frame
- **Ctrl + Right Arrow**: Jump Forward (by default also Next Second)
- **Shift + Right Arrow**: Jump to End
Pull Request: https://projects.blender.org/blender/blender/pulls/140677
When trying to align bones of 2 different armatures, with different mirror settings
it would crash.
The reason it crashed is that a pointer was not set to anything.
The whole function is not set up to do multi object editing and assumes that all bones gotten
from `selected_editable_bones` are part of the same armature. This is explained by the
comment above that explains that any unselected bones returned from this function are
assumed to be the mirrored bones.
Pull Request: https://projects.blender.org/blender/blender/pulls/147521
Recent changes made the 3D Cursor hard to see in some situations,
according to user feedback.
Roll-back most changes to bring back the old style and improve
readability and contrast.
* Keep the same size as the old cursor.
* Keep the crosshair inside the circle.
* Use red-white for the circle as before.
See PR for details and screenshots.
Pull Request: https://projects.blender.org/blender/blender/pulls/147404
This behavior seems to have been there since initial implementation of
this function in aed1320b24, 8 years ago.
However, it is clearly wrong, and would return wrong positives in many
cases (the Main RNA pointer itself is generated with a nullptr ID by
`RNA_main_pointer_create`, and a _lot_ of code generates RNA pointers
without any ID owner, as shown by a search e.g. on
`RNA_pointer_create_discrete(nullptr`).
WARNING: While logically correct, there is no real way to tell if this
change is not going to break some weird corner case based on wrong
assumption. in addition to unittests, this has also been checked against
several production files.
Indirectly caused by ada45519a7, the actual issue has been there for
years in the PointerRNA diffing code, which was way too basic and unsafe
in non-liboverride case.
Also fixed a copy/paste assert mistake in related code (same thing was
checked twice, instead of also checking another condition).
Finally, make the more general `RNA_struct_equals` call handle properly
cases where one or both of the given PointerRNA operands are Null ones.
The issue here was that I wrongly assumed that the locale would always be available.
Instead just use our built-in function to add thousands-separators (which I didn't
find last time).
Pull Request: https://projects.blender.org/blender/blender/pulls/147515
The scale node X input has no label, while Y does. This is due to the
`.optional_label()` statement which seems to have accidentally been
added in 149a47ba57. To fix this, we just remove the statement.
Pull Request: https://projects.blender.org/blender/blender/pulls/147178
Remove the "Render Viewport Keyframes" mode of the `RENDER_OT_opengl`
operator. This was implemented 6 years ago to avoid camera strobing when
animating on 2s, and was (AFAIK) little used after that initial need
(mostly because of usability issues and limited functionality).
The Fix to Camera operator deals with this at an animation level; that's
a better solution, as it also works for regular (non-viewport) renders.
See #146749
Pull Request: https://projects.blender.org/blender/blender/pulls/147463
This fixes two annoyances when working on node groups: * One has to constantly
switch between the Group and Node tab in the sidebar. * Finding a specific group
input/output in the tree view can be annoying when there are lots of inputs.
This patch implements two features which simplify the process significantly: *
When selecting a group input/output node, activate the corresponding item in the
tree view.
* When the node has multiple sockets, use the one that's closest to the
cursor.
* Show the group interface tree view in the node properties when selecting a
Group Input/Output node.
This constains one UI change that may affect other areas: When a new tree view
is added on a redraw, the active element is now scrolled into view, instead of
the scroll bar just being at the very top. It seems generally good to me, but
could be made optional.
Pull Request: https://projects.blender.org/blender/blender/pulls/147365
If a VSE compositor modifier is used to do keying or introduce an alpha
somehow to the image, the alpha is not used when blending. This is
because the VSE evaluator assumes all modifiers except Mask could not
introduce an alpha channel. So we need to extend the check to also make
an exception for the compositor modifier.
Pull Request: https://projects.blender.org/blender/blender/pulls/147456
Adding new element or changing panel looses the focus on active
element. Now handled in this PR inside the function `scroll_active_into_view`.
Iterate all tree view elements then use the index of active to set the
`scroll_value_`.
Resolves#143930
See PR description for the video
Pull Request: https://projects.blender.org/blender/blender/pulls/143977
Disallow converting objects to mesh while in Edit Mode. When an object
is in edit mode during conversion, its edit data may not be properly
flushed, resulting in invalid geometry and potential crashes.
Ref !147482
Support sync selection in the UV editor, with face-corner selection,
so it's possible to select individual UV vertices/edges in the UV editor
without UV's attached to the same underlying edge also becoming selected.
There is limited support for maintaining the UV selection when selecting
from the 3D viewport, common operations such as picking &
box/circle/lasso select support this, however other selection operations
such as "Select Random" or "Select Similar" will clear this data,
causing all UV's connected to selected mesh elements to become selected.
We may add support for additional operators as needed.
Details:
- UV Sync Selection is now enabled by default.
- In edit-mode the UV selection is stored in BMLoop/BMFace which are
written to custom-data layers when converted to a Mesh.
- To avoid unnecessary overhead - this data is created on demand.
Operators may clear this data - selecting all or none do so,
as there is no reason to store this data for a uniform selection.
- The Python API includes functions to synchronize the selection to/from
UV's as well as flushing based on the mode.
- Python scripts that manipulate the selection will either need to clear
this synchronized state or maintain it.
See:
- Design task: #78393.
- Implementation task: #131642.
Ref !138197
A test of using multi-color SVG icons in place of the current "vector"
(custom drawn) style we are using now. These use separate colors for
the box body versus outline, shadow, handhold. Colored by the theme
colors.
Pull Request: https://projects.blender.org/blender/blender/pulls/126307
Resolve regression in [0] which would mirror unselected faces.
Only copy UVs between mirror faces that are fully selected,
so unselected faces are left untouched. Vertex mapping also now uses
add_overwrite() to keep the last vertex at duplicate positions for a
more reliable mirroring.
Ref !147405
[0]: 919acbbf7f
Fixes incorrect color for dragged links, likely a regression from #145238.
The fix is to pass `TH_WIRE_INNER` to `nodelink_get_draw_config`. This results
in a color ID of 2 from `nodelink_get_color_id`, which satisfies the
`color_ids[0] < 3u` check in the `gpu_shader_2D_nodelink_vert.glsl`.
Pull Request: https://projects.blender.org/blender/blender/pulls/147462
This PR changes the current behavior that closes menus that are open
when your mouse is moved out of its "safe" area (a bit wider than its
visible bounds). With this PR applied menus stay open unless you click
elsewhere, hover a neighboring menu, press escape, or select an item.
Although this PR turns off menu mouse leave, it also adds a preference
that can be used to turn it back on.
Pull Request: https://projects.blender.org/blender/blender/pulls/145523
Selecting "Wide Lines" has always made the playhead vertical line
thicker. But with #145081 the line is an odd width, so when at 1X scale
it is three pixels instead of two with default lines, five pixels
instead four with wide line lines. But that one pixel isn't a big
deal, it is that the SHADOW around the line is now calculated based on
the line width when it used to be based on ui scale. This means at
1X scale and wide lines the shadow is two pixels when it used to
be just one. This PR goes back to the old shadow width calculation
which makes the overall obscuring width 2 pixels narrower at 1x wide.
Pull Request: https://projects.blender.org/blender/blender/pulls/147467
Use the custom color even when nodes are muted, to keep them more
recognizable.
The muted state is still indicated by mixing in the background color
and making the node transparent.
Additionally this makes small tweaks so the header colors of muted
nodes are the same between the collapsed and uncollapsed state.
Pull Request: https://projects.blender.org/blender/blender/pulls/147349
This PR gives access to OpenVDB's level set filtering operations
through individual **SDF Grid** nodes.
- **Mean Curvature**: Smoothens high curvature areas more than flatter
ones.
- **Laplacian**: Approximates mean curvature flow for true SDFs at lower
computational cost.
- **Median**: Reduces noise while preserving sharp features.
- **Mean**: Fast separable averaging filter for general-purpose
smoothing with linear computational complexity.
- **Offset**: Uniform dilation/erosion operation that shifts the SDF
surface by a world-space distance.
- **Fillet**: Rounds off concave internal corners by operating only on
regions with negative principal curvature.
Pull Request: https://projects.blender.org/blender/blender/pulls/147224
Add a function `bpy.data.file_path_foreach()` with a callback function
`visit_path_fn`:
```py
def visit_path_fn(
owner_id: bpy.types.ID,
file_path: str,
_meta: typing.Any,
) -> str | None:
return file_path.replace('xxx', 'yyy') or None
bpy.data.file_path_foreach(
visit_path_fn,
*,
subset=None,
visit_types=None,
flags={},
)
```
When the function returns `None`, nothing happens. When it returns a
string, the reported file path is replaced with the returned string.
`subset` and `visit_types` have the same meaning as `subset` resp.
`key_types` in `bpy.data.file_path_map()`. Since the latter parameter
doesn't affect keys in a map, but rather the visited ID types, I named
it `visit_types` instead.
`flags` wraps most of `eBPathForeachFlag` as `set[str]`, with their
prefixes removed. `BKE_BPATH_FOREACH_PATH_RESOLVE_TOKEN` becomes
`RESOLVE_TOKEN`, etc. `BKE_BPATH_FOREACH_PATH_ABSOLUTE` is excluded,
as it only affects a field in the C++ callback structure that is not
passed to Python at all. If it turns out to be useful at some point,
we can always add it.
`_meta` is for future use, and is intended to give metadata of the
path (like whether it's an input or an output path, a single file or
the first file in a sequence, supports variables, etc.). By adding
this to the API now, we avoid a breaking change when this feature is
actually implemented.
Design task: #145734
Pull Request: https://projects.blender.org/blender/blender/pulls/146261
Fixed the two most obvious entry points (actually making a linked ID
local, and making a local copy of linked data), there may be more
unusual ways to do that still hidden in the code though.
Also added a couple of 'sanity asserts' to the writefile code regarding
cases where IDs are expected to have a null deep_hash data.
This showed an error while it actually worked as expected. The issue was that it
attempted to pack the entires named `.` and `..`, which have special meaning.
Filtering those out quiets the error message.
Pull Request: https://projects.blender.org/blender/blender/pulls/147455
Using a strip mask in a compositor strip modifier causes other strips to
change in color. This is because the code converted the mask image
buffer in place, which affected its other uses. To fix this, convert the
mask out of the place.
Pull Request: https://projects.blender.org/blender/blender/pulls/147446
It's a common mistake that people want to use the viewer node while the
modifier is disabled and wondering why it doesn't work. This patch adds a
warning that helps users diagnose the issue immediately.
Pull Request: https://projects.blender.org/blender/blender/pulls/147443
The function name `for_each_callback` is ambigous (what are we iterating?)
and also doesn't align with the naming of other "for each" functions in Blender
which usually are named `foreach_<thing>` like `IndexMask::foreach_index`
or `animrig::foreach_fcurve_in_action` etc.
Pull Request: https://projects.blender.org/blender/blender/pulls/147440
Any keyframes for properties inside the compositor node
trees that were used by strips did not show up in the dopesheet.
This was because the compositor tree is its own ID and needs to
be manually added by the dopesheet code.
This fix adds the node trees to the filtering code in
`animdata_filter_dopesheet_scene`.
Note that the dopesheet still uses the active scene as its context.
So this means that if the sequencer scene differes from the active
scene, the animation data from the sequencer scene will not
be shown. This is currently expected behavior.
Pull Request: https://projects.blender.org/blender/blender/pulls/147301
Qualcomm driver team found an issue how Blender uses dynamic rendering.
The issue is that between suspend/resume of rendering only a limited
commands are possible. However we add commands that shouldn't be added
there. It is a violation of the Vulkan spec
VUID-VkSubmitInfo-pCommandBuffers-06015
It went unnoticed as it is mere a single line in the 5000+pages spec
and wasn't covered by the validation layers. This has been reported
upstream.
To fix this issue we remove the usage of suspending/resuming render
passes and fix any issue that used this together with clearops. Most
likely the fix will require a second dynamic rendering pass when
clearops are used.
Pull Request: https://projects.blender.org/blender/blender/pulls/147425
This happened if proportional editing was enabled in editmode, not
having an active object plus being in object mode.
The transform system `convert_type_get` assumes
`TransConvertType_Object` if we dont have an active view layer object.
Corresponding code for proportional editing detection [in
`initTransInfo` & `saveTransform`] wasnt though (it used `toolsettings`
`proportional_edit` -- not `proportional_objects` -- in that case).
This is now corrected.
Probably caused by e91d581167
Pull Request: https://projects.blender.org/blender/blender/pulls/147193
This adds a new `PropertyRNA::name_func` property which is similar to the many
existing functions like `editable`, `get_default`, etc. It allows dynamically
getting a UI name for the property.
This is especially useful for node sockets, because those all have the same
hardcoded name "Default Value" which is not helpful. Since we already have
custom tooltips for sockets, this is mostly not visible to the user anymore. The
exception being menu sockets which draw the property name as title. Instead of
"Default Value", this patch makes it show the correct title.
Pull Request: https://projects.blender.org/blender/blender/pulls/147137
Spawning the tooltip or changing values would cause this.
Error in 28827b62f7
Above commit only considered PointCaches associated with **objects**,
not taking into account scene rigid body cache.
This is now added with this PR.
Previously it would cast into modifiers when we actually had a scene as
owner... resulting in a crash.
This was only exposed by 6bf8685628 (it crashes since, was just not
working properly before...). 6bf8685628 changed the data layout of
`struct Object` and only by pure luck the modifiers loop was skipped, now
with the new layout it is entered (but with garbage data resulting in the crash.
Pull Request: https://projects.blender.org/blender/blender/pulls/146577
